dist/ANNOUNCE100660 765 144 4545 6071542763 11537 0ustar lordusersGNU sed 3.0 is now available. Copies can be obtained by ftp from prep.ai.mit.edu:pub/gnu/sed-3.0.tar.gz and from other sites (see the list of mirror sites at the end of this message). It is more P.C. (Posixly Correct) and has more reliable performance than older versions of GNU sed. Please report bugs to bug-gnu-utils@prep.ai.mit.edu. [ Most GNU software is packed using the GNU `gzip' compression program. Source code is available on most sites distributing GNU software. For information on how to order GNU software on tape, floppy or cd-rom, or printed GNU manuals, check the file etc/ORDERS in the GNU Emacs distribution, ftp the file /pub/gnu/GNUinfo/ORDERS on prep, or e-mail a request to: gnu@prep.ai.mit.edu By ordering your GNU software from the FSF, you help us continue to develop more free software. Media revenues are our primary source of support. Donations to FSF are deductible on US tax returns. The above software will soon be at these ftp sites as well. Please try them before prep.ai.mit.edu as prep is very busy! thanx -gnu@prep.ai.mit.edu ASIA: ftp.cs.titech.ac.jp, utsun.s.u-tokyo.ac.jp:/ftpsync/prep, cair-archive.kaist.ac.kr:/pub/gnu, ftp.nectec.or.th:/pub/mirrors/gnu AUSTRALIA: archie.au:/gnu (archie.oz or archie.oz.au for ACSnet) AFRICA: ftp.sun.ac.za:/pub/gnu MIDDLE-EAST: ftp.technion.ac.il:/pub/unsupported/gnu EUROPE: irisa.irisa.fr:/pub/gnu, ftp.univ-lyon1.fr:pub/gnu, ftp.mcc.ac.uk, unix.hensa.ac.uk:/mirrors/uunet/systems/gnu, src.doc.ic.ac.uk:/gnu, ftp.ieunet.ie:pub/gnu, ftp.eunet.ch, nic.switch.ch:/mirror/gnu, ftp.informatik.rwth-aachen.de:/pub/gnu, ftp.informatik.tu-muenchen.de, ftp.win.tue.nl:/pub/gnu, ftp.nl.net, ftp.etsimo.uniovi.es:/pub/gnu ftp.funet.fi:/pub/gnu, ftp.denet.dk, ftp.stacken.kth.se, isy.liu.se, ftp.luth.se:/pub/unix/gnu, ftp.sunet.se:/pub/gnu, archive.eu.net SOUTH AMERICA: ftp.inf.utfsm.cl:/pub/gnu, ftp.unicamp.br:/pub/gnu WESTERN CANADA: ftp.cs.ubc.ca:/mirror2/gnu USA: wuarchive.wustl.edu:/systems/gnu, labrea.stanford.edu, ftp.digex.net:/pub/gnu, ftp.kpc.com:/pub/mirror/gnu, f.ms.uky.edu:/pub3/gnu, jaguar.utah.edu:/gnustuff, ftp.hawaii.edu:/mirrors/gnu, uiarchive.cso.uiuc.edu:/pub/gnu, ftp.cs.columbia.edu:/archives/gnu/prep, col.hp.com:/mirrors/gnu, gatekeeper.dec.com:/pub/GNU, ftp.uu.net:/systems/gnu ] dist/COPYING100660 765 144 43112 6071377322 11447 0ustar lordusers GNU GENERAL PUBLIC LICENSE Version 2, June 1991 Copyright (C) 1989, 1991, 1992, 1993 Free Software Foundation, Inc. 675 Mass Ave, Cambridge, MA 02139, USA Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed. Preamble The licenses for most software are designed to take away your freedom to share and change it. By contrast, the GNU General Public License is intended to guarantee your freedom to share and change free software--to make sure the software is free for all its users. This General Public License applies to most of the Free Software Foundation's software and to any other program whose authors commit to using it. (Some other Free Software Foundation software is covered by the GNU Library General Public License instead.) You can apply it to your programs, too. When we speak of free software, we are referring to freedom, not price. Our General Public Licenses are designed to make sure that you have the freedom to distribute copies of free software (and charge for this service if you wish), that you receive source code or can get it if you want it, that you can change the software or use pieces of it in new free programs; and that you know you can do these things. To protect your rights, we need to make restrictions that forbid anyone to deny you these rights or to ask you to surrender the rights. These restrictions translate to certain responsibilities for you if you distribute copies of the software, or if you modify it. For example, if you distribute copies of such a program, whether gratis or for a fee, you must give the recipients all the rights that you have. You must make sure that they, too, receive or can get the source code. And you must show them these terms so they know their rights. We protect your rights with two steps: (1) copyright the software, and (2) offer you this license which gives you legal permission to copy, distribute and/or modify the software. Also, for each author's protection and ours, we want to make certain that everyone understands that there is no warranty for this free software. If the software is modified by someone else and passed on, we want its recipients to know that what they have is not the original, so that any problems introduced by others will not reflect on the original authors' reputations. Finally, any free program is threatened constantly by software patents. We wish to avoid the danger that redistributors of a free program will individually obtain patent licenses, in effect making the program proprietary. To prevent this, we have made it clear that any patent must be licensed for everyone's free use or not licensed at all. The precise terms and conditions for copying, distribution and modification follow. GNU GENERAL PUBLIC LICENSE TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION 0. This License applies to any program or other work which contains a notice placed by the copyright holder saying it may be distributed under the terms of this General Public License. The "Program", below, refers to any such program or work, and a "work based on the Program" means either the Program or any derivative work under copyright law: that is to say, a work containing the Program or a portion of it, either verbatim or with modifications and/or translated into another language. (Hereinafter, translation is included without limitation in the term "modification".) Each licensee is addressed as "you". Activities other than copying, distribution and modification are not covered by this License; they are outside its scope. The act of running the Program is not restricted, and the output from the Program is covered only if its contents constitute a work based on the Program (independent of having been made by running the Program). Whether that is true depends on what the Program does. 1. You may copy and distribute verbatim copies of the Program's source code as you receive it, in any medium, provided that you conspicuously and appropriately publish on each copy an appropriate copyright notice and disclaimer of warranty; keep intact all the notices that refer to this License and to the absence of any warranty; and give any other recipients of the Program a copy of this License along with the Program. You may charge a fee for the physical act of transferring a copy, and you may at your option offer warranty protection in exchange for a fee. 2. You may modify your copy or copies of the Program or any portion of it, thus forming a work based on the Program, and copy and distribute such modifications or work under the terms of Section 1 above, provided that you also meet all of these conditions: a) You must cause the modified files to carry prominent notices stating that you changed the files and the date of any change. b) You must cause any work that you distribute or publish, that in whole or in part contains or is derived from the Program or any part thereof, to be licensed as a whole at no charge to all third parties under the terms of this License. c) If the modified program normally reads commands interactively when run, you must cause it, when started running for such interactive use in the most ordinary way, to print or display an announcement including an appropriate copyright notice and a notice that there is no warranty (or else, saying that you provide a warranty) and that users may redistribute the program under these conditions, and telling the user how to view a copy of this License. (Exception: if the Program itself is interactive but does not normally print such an announcement, your work based on the Program is not required to print an announcement.) These requirements apply to the modified work as a whole. If identifiable sections of that work are not derived from the Program, and can be reasonably considered independent and separate works in themselves, then this License, and its terms, do not apply to those sections when you distribute them as separate works. But when you distribute the same sections as part of a whole which is a work based on the Program, the distribution of the whole must be on the terms of this License, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it. Thus, it is not the intent of this section to claim rights or contest your rights to work written entirely by you; rather, the intent is to exercise the right to control the distribution of derivative or collective works based on the Program. In addition, mere aggregation of another work not based on the Program with the Program (or with a work based on the Program) on a volume of a storage or distribution medium does not bring the other work under the scope of this License. 3. You may copy and distribute the Program (or a work based on it, under Section 2) in object code or executable form under the terms of Sections 1 and 2 above provided that you also do one of the following: a) Accompany it with the complete corresponding machine-readable source code, which must be distributed under the terms of Sections 1 and 2 above on a medium customarily used for software interchange; or, b) Accompany it with a written offer, valid for at least three years, to give any third party, for a charge no more than your cost of physically performing source distribution, a complete machine-readable copy of the corresponding source code, to be distributed under the terms of Sections 1 and 2 above on a medium customarily used for software interchange; or, c) Accompany it with the information you received as to the offer to distribute corresponding source code. (This alternative is allowed only for noncommercial distribution and only if you received the program in object code or executable form with such an offer, in accord with Subsection b above.) The source code for a work means the preferred form of the work for making modifications to it. For an executable work, complete source code means all the source code for all modules it contains, plus any associated interface definition files, plus the scripts used to control compilation and installation of the executable. However, as a special exception, the source code distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable. If distribution of executable or object code is made by offering access to copy from a designated place, then offering equivalent access to copy the source code from the same place counts as distribution of the source code, even though third parties are not compelled to copy the source along with the object code. 4. You may not copy, modify, sublicense, or distribute the Program except as expressly provided under this License. Any attempt otherwise to copy, modify, sublicense or distribute the Program is void, and will automatically terminate your rights under this License. However, parties who have received copies, or rights, from you under this License will not have their licenses terminated so long as such parties remain in full compliance. 5. You are not required to accept this License, since you have not signed it. However, nothing else grants you permission to modify or distribute the Program or its derivative works. These actions are prohibited by law if you do not accept this License. Therefore, by modifying or distributing the Program (or any work based on the Program), you indicate your acceptance of this License to do so, and all its terms and conditions for copying, distributing or modifying the Program or works based on it. 6. Each time you redistribute the Program (or any work based on the Program), the recipient automatically receives a license from the original licensor to copy, distribute or modify the Program subject to these terms and conditions. You may not impose any further restrictions on the recipients' exercise of the rights granted herein. You are not responsible for enforcing compliance by third parties to this License. 7. If, as a consequence of a court judgment or allegation of patent infringement or for any other reason (not limited to patent issues), conditions are imposed on you (whether by court order, agreement or otherwise) that contradict the conditions of this License, they do not excuse you from the conditions of this License. If you cannot distribute so as to satisfy simultaneously your obligations under this License and any other pertinent obligations, then as a consequence you may not distribute the Program at all. For example, if a patent license would not permit royalty-free redistribution of the Program by all those who receive copies directly or indirectly through you, then the only way you could satisfy both it and this License would be to refrain entirely from distribution of the Program. If any portion of this section is held invalid or unenforceable under any particular circumstance, the balance of the section is intended to apply and the section as a whole is intended to apply in other circumstances. It is not the purpose of this section to induce you to infringe any patents or other property right claims or to contest validity of any such claims; this section has the sole purpose of protecting the integrity of the free software distribution system, which is implemented by public license practices. Many people have made generous contributions to the wide range of software distributed through that system in reliance on consistent application of that system; it is up to the author/donor to decide if he or she is willing to distribute software through any other system and a licensee cannot impose that choice. This section is intended to make thoroughly clear what is believed to be a consequence of the rest of this License. 8. If the distribution and/or use of the Program is restricted in certain countries either by patents or by copyrighted interfaces, the original copyright holder who places the Program under this License may add an explicit geographical distribution limitation excluding those countries, so that distribution is permitted only in or among countries not thus excluded. In such case, this License incorporates the limitation as if written in the body of this License. 9. The Free Software Foundation may publish revised and/or new versions of the General Public License from time to time. Such new versions will be similar in spirit to the present version, but may differ in detail to address new problems or concerns. Each version is given a distinguishing version number. If the Program specifies a version number of this License which applies to it and "any later version", you have the option of following the terms and conditions either of that version or of any later version published by the Free Software Foundation. If the Program does not specify a version number of this License, you may choose any version ever published by the Free Software Foundation. 10. If you wish to incorporate parts of the Program into other free programs whose distribution conditions are different, write to the author to ask for permission. For software which is copyrighted by the Free Software Foundation, write to the Free Software Foundation; we sometimes make exceptions for this. Our decision will be guided by the two goals of preserving the free status of all derivatives of our free software and of promoting the sharing and reuse of software generally. NO WARRANTY 11. BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION. 12. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. END OF TERMS AND CONDITIONS Appendix: How to Apply These Terms to Your New Programs If you develop a new program, and you want it to be of the greatest possible use to the public, the best way to achieve this is to make it free software which everyone can redistribute and change under these terms. To do so, attach the following notices to the program. It is safest to attach them to the start of each source file to most effectively convey the exclusion of warranty; and each file should have at least the "copyright" line and a pointer to where the full notice is found. Copyright (C) 19yy This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. Also add information on how to contact you by electronic and paper mail. If the program is interactive, make it output a short notice like this when it starts in an interactive mode: Gnomovision version 69, Copyright (C) 19yy name of author Gnomovision comes with ABSOLUTELY NO WARRANTY; for details type `show w'. This is free software, and you are welcome to redistribute it under certain conditions; type `show c' for details. The hypothetical commands `show w' and `show c' should show the appropriate parts of the General Public License. Of course, the commands you use may be called something other than `show w' and `show c'; they could even be mouse-clicks or menu items--whatever suits your program. You should also get your employer (if you work as a programmer) or your school, if any, to sign a "copyright disclaimer" for the program, if necessary. Here is a sample; alter the names: Yoyodyne, Inc., hereby disclaims all copyright interest in the program `Gnomovision' (which makes passes at compilers) written by James Hacker. , 1 April 1989 Ty Coon, President of Vice This General Public License does not permit incorporating your program into proprietary programs. If your program is a subroutine library, you may consider it more useful to permit linking proprietary applications with the library. If this is what you want to do, use the GNU Library General Public License instead of this License. dist/INSTALL100660 765 144 11134 6071377322 11444 0ustar lordusersRecommended short-cut when using a generic make(1) program. % tar -zxvf .tar.gz % cd % mkdir =inst % ./configure --prefix=`pwd`/=inst % make install That will compile everything and install under ./=inst. For example, `pwd`/=inst/bin will contain the executables. Recommended short-cut when using GNU Make. This uses the VPATH feature to keep the intermediate files apart from the source files. % tar -zxvf .tar.gz % cd % mkdir =inst % mkdir =build % cd =build % ../configure --prefix=`pwd`/../=inst % make install That will also install everything in =inst, but the object files will all be contained under =build. The rest of this file is generic INSTALL documentation for utilities distributions. To compile this package: 1. Configure the package for your system. In the directory that this file is in, type `./configure'. If you're using `csh' on an old version of System V, you might need to type `sh configure' instead to prevent `csh' from trying to execute `configure' itself. The `configure' shell script attempts to guess correct values for various system-dependent variables used during compilation, and creates the Makefile(s) (one in each subdirectory of the source directory). In some packages it creates a C header file containing system-dependent definitions. It also creates a file `config.status' that you can run in the future to recreate the current configuration. Running `configure' takes a minute or two. To compile the package in a different directory from the one containing the source code, you must use GNU make. `cd' to the directory where you want the object files and executables to go and run `configure' with the option `--srcdir=DIR', where DIR is the directory that contains the source code. Using this option is actually unnecessary if the source code is in the parent directory of the one in which you are compiling; `configure' automatically checks for the source code in `..' if it does not find it in the current directory. By default, `make install' will install the package's files in /usr/local/bin, /usr/local/lib, /usr/local/man, etc. You can specify an installation prefix other than /usr/local by giving `configure' the option `--prefix=PATH'. Alternately, you can do so by changing the `prefix' variable in the Makefile that `configure' creates (the Makefile in the top-level directory, if the package contains subdirectories). You can specify separate installation prefixes for machine-specific files and machine-independent files. If you give `configure' the option `--exec_prefix=PATH', the package will use PATH as the prefix for installing programs and libraries. Normally, all files are installed using the same prefix. `configure' ignores any other arguments that you give it. If your system requires unusual options for compilation or linking that `configure' doesn't know about, you can give `configure' initial values for some variables by setting them in the environment. In Bourne-compatible shells, you can do that on the command line like this: CC='gcc -traditional' DEFS=-D_POSIX_SOURCE ./configure The `make' variables that you might want to override with environment variables when running `configure' are: (For these variables, any value given in the environment overrides the value that `configure' would choose:) CC C compiler program. Default is `cc', or `gcc' if `gcc' is in your PATH. INSTALL Program to use to install files. Default is `install' if you have it, `cp' otherwise. INCLUDEDIR Directory for `configure' to search for include files. Default is /usr/include. (For these variables, any value given in the environment is added to the value that `configure' chooses:) DEFS Configuration options, in the form '-Dfoo -Dbar ...' LIBS Libraries to link with, in the form '-lfoo -lbar ...' If you need to do unusual things to compile the package, we encourage you to teach `configure' how to do them and mail the diffs to the address given in the README so we can include them in the next release. 2. Type `make' to compile the package. 3. Type `make install' to install programs, data files, and documentation. 4. You can remove the program binaries and object files from the source directory by typing `make clean'. To also remove the Makefile(s), the header file containing system-dependent definitions (if the package uses one), and `config.status' (all the files that `configure' created), type `make distclean'. The file `configure.in' is used as a template to create `configure' by a program called `autoconf'. You will only need it if you want to regenerate `configure' using a newer version of `autoconf'. dist/Makefile.in100660 765 144 23466 6071420202 12455 0ustar lordusers# Copyright (C) 1994,1995 Free Software Foundation, Inc. # # This program is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by # the Free Software Foundation; either version 2, or (at your option) # any later version. # # This program is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with this software; see the file COPYING. If not, write to # the Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. SHELL = /bin/sh subdirs=@existingdirs@ localfiles = ANNOUNCE \ COPYING \ INSTALL \ Makefile.in \ configure \ configure.in \ config.sub \ config.guess \ doc \ install-sh localtreats = ANN.BX # `all' # Compile the entire program. This should be the default target. # This target need not rebuild any documentation files; info files # should normally be included in the distribution, and DVI files # should be made only when explicitly asked for. all: @for dir in ${subdirs}; do \ cd $$dir; \ ${MAKE} all; \ cd .. ;\ done #`install' # Compile the program and copy the executables, libraries, and so on # to the file names where they should reside for actual use. If # there is a simple test to verify that a program is properly # installed then run that test. # # Use `-' before any command for installing a man page, so that # `make' will ignore any errors. This is in case there are systems # that don't have the Unix man page documentation system installed. # # In the future, when we have a standard way of installing info # files, `install' targets will be the proper place to do so. # # Cygnus extention: # # `install' # Should *not* depend on the target `all'. [feh!] If the program is # not already built, `make install' should fail. This allows # you to install programs even when `make' would otherwise # determine them to be out of date. This can happen, for # example, when the result of a `make all' is transported via # tape to another machine for installation. # subdir-inst-target=install-nobuild install: all ${MAKE} subdir-inst-target=install install-nobuild install-nobuild: for dir in ${subdirs}; do \ cd $$dir; \ ${MAKE} ${subdir-inst-target}; \ cd .. ;\ done #`uninstall' # Delete all the installed files that the `install' target would # create (but not the noninstalled files such as `make all' would # create). uninstall: for dir in ${subdirs}; do \ cd $$dir; \ ${MAKE} uninstall; \ cd .. ;\ done #`clean' # Delete all files from the current directory that are normally # created by building the program. Don't delete the files that # record the configuration. Also preserve files that could be made # by building, but normally aren't because the distribution comes # with them. # # Delete `.dvi' files here if they are not part of the distribution. # # Cygnus extention: # # `clean' # Should remove any file that can be regenerated by the # `Makefile', excepting only the `Makefile' itself, and any # links created by `configure'. That is, `make all clean' # should return all directories to their original condition. # If this is not done, then the command sequence # # configure HOST1 ; make all install clean ; # configure HOST2 ; make all install # # will fail because of intermediate files intended for HOST1. clean: for dir in ${subdirs}; do \ cd $$dir; \ ${MAKE} clean; \ cd .. ;\ done #`distclean' # Delete all files from the current directory that are created by # configuring or building the program. If you have unpacked the # source and built the program without creating any other files, # `make distclean' should leave only the files that were in the # distribution. distclean: for dir in ${subdirs}; do \ cd $$dir; \ ${MAKE} distclean; \ cd .. ;\ done #`mostlyclean' # Like `clean', but may refrain from deleting a few files that people # normally don't want to recompile. For example, the `mostlyclean' # target for GCC does not delete `libgcc.a', because recompiling it # is rarely necessary and takes a lot of time. mostlyclean: for dir in ${subdirs}; do \ cd $$dir; \ ${MAKE} mostlyclean; \ cd .. ;\ done #`realclean' # Delete everything from the current directory that can be # reconstructed with this Makefile. This typically includes # everything deleted by distclean, plus more: C source files # produced by Bison, tags tables, info files, and so on. # # One exception, however: `make realclean' should not delete # `configure' even if `configure' can be remade using a rule in the # Makefile. More generally, `make realclean' should not delete # anything that needs to exist in order to run `configure' and then # begin to build the program. realclean: for dir in ${subdirs}; do \ cd $$dir; \ ${MAKE} realclean; \ cd .. ;\ done #`TAGS' # Update a tags table for this program. TAGS: for dir in ${subdirs}; do \ cd $$dir; \ ${MAKE} TAGS; \ cd .. ;\ done #`info' # Generate any info files needed. The best way to write the rules # is as follows: # # info: foo.info # # foo.info: $(srcdir)/foo.texi $(srcdir)/chap1.texi $(srcdir)/chap2.texi # $(MAKEINFO) $(srcdir)/foo.texi # # You must define the variable `MAKEINFO' in the Makefile. It # should run the Makeinfo program, which is part of the Texinfo2 # distribution. info: for dir in ${subdirs}; do \ cd $$dir; \ ${MAKE} info; \ cd .. ;\ done #`dvi' # Generate DVI files for all TeXinfo documentation. For example: # # dvi: foo.dvi # # foo.dvi: $(srcdir)/foo.texi $(srcdir)/chap1.texi $(srcdir)/chap2.texi # $(TEXI2DVI) $(srcdir)/foo.texi # # You must define the variable `TEXI2DVI' in the Makefile. It should # run the program `texi2dvi', which is part of the Texinfo2 # distribution. Alternatively, write just the dependencies, and # allow GNU Make to provide the command. dvi: for dir in ${subdirs}; do \ cd $$dir; \ ${MAKE} dvi; \ cd .. ;\ done #`dist' # Create a distribution tar file for this program. The tar file # should be set up so that the file names in the tar file start with # a subdirectory name which is the name of the package it is a # distribution for. This name can include the version number. # # For example, the distribution tar file of GCC version 1.40 unpacks # into a subdirectory named `gcc-1.40'. # # The easiest way to do this is to create a subdirectory # appropriately named, use `ln' or `cp' to install the proper files # in it, and then `tar' that subdirectory. # # The `dist' target should explicitly depend on all non-source files # that are in the distribution, to make sure they are up to date in # the distribution. *Ref Making Releases: (standards)Releases. distname = brand-x distdir = $(distname) treats = $(localtreats) announcefile = ANN.BX manifest-file: rm -f MANIFEST cp $(announcefile) ANNOUNCE for treat in $(localfiles) $(treats) ; \ do echo $$treat >> MANIFEST ; \ done for subdir in $(subdirs) ; do \ make -s -f $$subdir/Makefile.in SUBDIR=$$subdir manifest >> MANIFEST ; \ done sed -e "s%^%$(distdir)/%" MANIFEST > M2 sed -e "/Entering dir/d" -e "/Leaving dir/d" M2 > MANIFEST rm M2 dist: manifest-file mkdir $(distdir) - cd $(distdir); \ for file in $(localfiles) $(treats) $(subdirs); do ln -s ../$$file; done; \ cd ..; \ tar -zhcvf $(distname).tar.gz --files-from MANIFEST rm -rf $(distdir) #`check' # Perform self-tests (if any). The user must build the program # before running the tests, but need not install the program; you # should write the self-tests so that they work when the program is # built but not installed. check: for dir in ${subdirs}; do \ cd $$dir; \ ${MAKE} check; \ cd .. ;\ done #`installcheck' # Perform installation tests (if any). The user must build and # install the program before running the tests. You should not # assume that `$(bindir)' is in the search path. installcheck: for dir in ${subdirs}; do \ cd $$dir; \ ${MAKE} installcheck; \ cd .. ;\ done #`installdirs' # It's useful to add a target named `installdirs' to create the # directories where files are installed, and their parent # directories. There is a script called `mkinstalldirs' which is # convenient for this; find it in the Texinfo package.You can use a # rule like this: # # # Make sure all installation directories, e.g. $(bindir) actually exist by # # making them if necessary. # installdirs: mkinstalldirs # $(srcdir)/mkinstalldirs $(bindir) $(datadir) $(libdir) \ # $(infodir) $(mandir) installdirs: for dir in ${subdirs}; do \ cd $$dir; \ ${MAKE} installdirs; \ cd .. ;\ done # Cygnus extention: # # `install-info' # Install all info files. install-info: for dir in ${subdirs}; do \ cd $$dir; \ ${MAKE} install-info; \ cd .. ;\ done # Cygnus extention: # # `clean-info' # Remove all info files and any intermediate files that can be # generated from texinfo source. clean-info: for dir in ${subdirs}; do \ cd $$dir; \ ${MAKE} clean-info; \ cd .. ;\ done # Cygnus extention: # # `Makefile' # Calls `./config.status' to rebuild the `Makefile' in this # directory. Makefile: ${SHELL-/bin/sh} config.status dist/configure100770 765 144 56704 6071421127 12326 0ustar lordusers#! /bin/sh # Guess values for system-dependent variables and create Makefiles. # Generated automatically using autoconf version 2.3 # Copyright (C) 1992, 1993, 1994 Free Software Foundation, Inc. # # This configure script is free software; the Free Software Foundation # gives unlimited permission to copy, distribute and modify it. # Defaults: ac_help= ac_default_prefix=/usr/local # Any additions from configure.in: # Initialize some variables set by options. # The variables have the same names as the options, with # dashes changed to underlines. build=NONE cache_file=./config.cache exec_prefix=NONE host=NONE no_create= nonopt=NONE no_recursion= prefix=NONE program_prefix=NONE program_suffix=NONE program_transform_name=s,x,x, silent= site= srcdir= target=NONE verbose= x_includes=NONE x_libraries=NONE # Initialize some other variables. subdirs= ac_prev= for ac_option do # If the previous option needs an argument, assign it. if test -n "$ac_prev"; then eval "$ac_prev=\$ac_option" ac_prev= continue fi case "$ac_option" in -*=*) ac_optarg=`echo "$ac_option" | sed 's/[-_a-zA-Z0-9]*=//'` ;; *) ac_optarg= ;; esac # Accept the important Cygnus configure options, so we can diagnose typos. case "$ac_option" in -build | --build | --buil | --bui | --bu | --b) ac_prev=build ;; -build=* | --build=* | --buil=* | --bui=* | --bu=* | --b=*) build="$ac_optarg" ;; -cache-file | --cache-file | --cache-fil | --cache-fi \ | --cache-f | --cache- | --cache | --cach | --cac | --ca | --c) ac_prev=cache_file ;; -cache-file=* | --cache-file=* | --cache-fil=* | --cache-fi=* \ | --cache-f=* | --cache-=* | --cache=* | --cach=* | --cac=* | --ca=* | --c=*) cache_file="$ac_optarg" ;; -disable-* | --disable-*) ac_feature=`echo $ac_option|sed -e 's/-*disable-//'` # Reject names that are not valid shell variable names. if test -n "`echo $ac_feature| sed 's/[-a-zA-Z0-9_]//g'`"; then { echo "configure: error: $ac_feature: invalid feature name" 1>&2; exit 1; } fi ac_feature=`echo $ac_feature| sed 's/-/_/g'` eval "enable_${ac_feature}=no" ;; -enable-* | --enable-*) ac_feature=`echo $ac_option|sed -e 's/-*enable-//' -e 's/=.*//'` # Reject names that are not valid shell variable names. if test -n "`echo $ac_feature| sed 's/[-_a-zA-Z0-9]//g'`"; then { echo "configure: error: $ac_feature: invalid feature name" 1>&2; exit 1; } fi ac_feature=`echo $ac_feature| sed 's/-/_/g'` case "$ac_option" in *=*) ;; *) ac_optarg=yes ;; esac eval "enable_${ac_feature}='$ac_optarg'" ;; -exec-prefix | --exec_prefix | --exec-prefix | --exec-prefi \ | --exec-pref | --exec-pre | --exec-pr | --exec-p | --exec- \ | --exec | --exe | --ex) ac_prev=exec_prefix ;; -exec-prefix=* | --exec_prefix=* | --exec-prefix=* | --exec-prefi=* \ | --exec-pref=* | --exec-pre=* | --exec-pr=* | --exec-p=* | --exec-=* \ | --exec=* | --exe=* | --ex=*) exec_prefix="$ac_optarg" ;; -gas | --gas | --ga | --g) # Obsolete; use --with-gas. with_gas=yes ;; -help | --help | --hel | --he) # Omit some internal or obsolete options to make the list less imposing. # This message is too long to be a string in the A/UX 3.1 sh. cat << EOF Usage: configure [options] [host] Options: [defaults in brackets after descriptions] Configuration: --cache-file=FILE cache test results in FILE --help print this message --no-create do not create output files --quiet, --silent do not print \`checking...' messages --version print the version of autoconf that created configure Directory and file names: --prefix=PREFIX install architecture-independent files in PREFIX [$ac_default_prefix] --exec-prefix=PREFIX install architecture-dependent files in PREFIX [same as prefix] --srcdir=DIR find the sources in DIR [configure dir or ..] --program-prefix=PREFIX prepend PREFIX to installed program names --program-suffix=SUFFIX append SUFFIX to installed program names --program-transform-name=PROGRAM run sed PROGRAM on installed program names Host type: --build=BUILD configure for building on BUILD [BUILD=HOST] --host=HOST configure for HOST [guessed] --target=TARGET configure for TARGET [TARGET=HOST] Features and packages: --disable-FEATURE do not include FEATURE (same as --enable-FEATURE=no) --enable-FEATURE[=ARG] include FEATURE [ARG=yes] --with-PACKAGE[=ARG] use PACKAGE [ARG=yes] --without-PACKAGE do not use PACKAGE (same as --with-PACKAGE=no) --x-includes=DIR X include files are in DIR --x-libraries=DIR X library files are in DIR --enable and --with options recognized:$ac_help EOF exit 0 ;; -host | --host | --hos | --ho) ac_prev=host ;; -host=* | --host=* | --hos=* | --ho=*) host="$ac_optarg" ;; -nfp | --nfp | --nf) # Obsolete; use --without-fp. with_fp=no ;; -no-create | --no-create | --no-creat | --no-crea | --no-cre \ | --no-cr | --no-c) no_create=yes ;; -no-recursion | --no-recursion | --no-recursio | --no-recursi \ | --no-recurs | --no-recur | --no-recu | --no-rec | --no-re | --no-r) no_recursion=yes ;; -prefix | --prefix | --prefi | --pref | --pre | --pr | --p) ac_prev=prefix ;; -prefix=* | --prefix=* | --prefi=* | --pref=* | --pre=* | --pr=* | --p=*) prefix="$ac_optarg" ;; -program-prefix | --program-prefix | --program-prefi | --program-pref \ | --program-pre | --program-pr | --program-p) ac_prev=program_prefix ;; -program-prefix=* | --program-prefix=* | --program-prefi=* \ | --program-pref=* | --program-pre=* | --program-pr=* | --program-p=*) program_prefix="$ac_optarg" ;; -program-suffix | --program-suffix | --program-suffi | --program-suff \ | --program-suf | --program-su | --program-s) ac_prev=program_suffix ;; -program-suffix=* | --program-suffix=* | --program-suffi=* \ | --program-suff=* | --program-suf=* | --program-su=* | --program-s=*) program_suffix="$ac_optarg" ;; -program-transform-name | --program-transform-name \ | --program-transform-nam | --program-transform-na \ | --program-transform-n | --program-transform- \ | --program-transform | --program-transfor \ | --program-transfo | --program-transf \ | --program-trans | --program-tran \ | --progr-tra | --program-tr | --program-t) ac_prev=program_transform_name ;; -program-transform-name=* | --program-transform-name=* \ | --program-transform-nam=* | --program-transform-na=* \ | --program-transform-n=* | --program-transform-=* \ | --program-transform=* | --program-transfor=* \ | --program-transfo=* | --program-transf=* \ | --program-trans=* | --program-tran=* \ | --progr-tra=* | --program-tr=* | --program-t=*) program_transform_name="$ac_optarg" ;; -q | -quiet | --quiet | --quie | --qui | --qu | --q \ | -silent | --silent | --silen | --sile | --sil) silent=yes ;; -site | --site | --sit) ac_prev=site ;; -site=* | --site=* | --sit=*) site="$ac_optarg" ;; -srcdir | --srcdir | --srcdi | --srcd | --src | --sr) ac_prev=srcdir ;; -srcdir=* | --srcdir=* | --srcdi=* | --srcd=* | --src=* | --sr=*) srcdir="$ac_optarg" ;; -target | --target | --targe | --targ | --tar | --ta | --t) ac_prev=target ;; -target=* | --target=* | --targe=* | --targ=* | --tar=* | --ta=* | --t=*) target="$ac_optarg" ;; -v | -verbose | --verbose | --verbos | --verbo | --verb) verbose=yes ;; -version | --version | --versio | --versi | --vers) echo "configure generated by autoconf version 2.3" exit 0 ;; -with-* | --with-*) ac_package=`echo $ac_option|sed -e 's/-*with-//' -e 's/=.*//'` # Reject names that are not valid shell variable names. if test -n "`echo $ac_package| sed 's/[-_a-zA-Z0-9]//g'`"; then { echo "configure: error: $ac_package: invalid package name" 1>&2; exit 1; } fi ac_package=`echo $ac_package| sed 's/-/_/g'` case "$ac_option" in *=*) ;; *) ac_optarg=yes ;; esac eval "with_${ac_package}='$ac_optarg'" ;; -without-* | --without-*) ac_package=`echo $ac_option|sed -e 's/-*without-//'` # Reject names that are not valid shell variable names. if test -n "`echo $ac_package| sed 's/[-a-zA-Z0-9_]//g'`"; then { echo "configure: error: $ac_package: invalid package name" 1>&2; exit 1; } fi ac_package=`echo $ac_package| sed 's/-/_/g'` eval "with_${ac_package}=no" ;; --x) # Obsolete; use --with-x. with_x=yes ;; -x-includes | --x-includes | --x-include | --x-includ | --x-inclu \ | --x-incl | --x-inc | --x-in | --x-i) ac_prev=x_includes ;; -x-includes=* | --x-includes=* | --x-include=* | --x-includ=* | --x-inclu=* \ | --x-incl=* | --x-inc=* | --x-in=* | --x-i=*) x_includes="$ac_optarg" ;; -x-libraries | --x-libraries | --x-librarie | --x-librari \ | --x-librar | --x-libra | --x-libr | --x-lib | --x-li | --x-l) ac_prev=x_libraries ;; -x-libraries=* | --x-libraries=* | --x-librarie=* | --x-librari=* \ | --x-librar=* | --x-libra=* | --x-libr=* | --x-lib=* | --x-li=* | --x-l=*) x_libraries="$ac_optarg" ;; -*) { echo "configure: error: $ac_option: invalid option; use --help to show usage" 1>&2; exit 1; } ;; *) if test -n "`echo $ac_option| sed 's/[-a-z0-9.]//g'`"; then echo "configure: warning: $ac_option: invalid host type" 1>&2 fi if test "x$nonopt" != xNONE; then { echo "configure: error: can only configure for one host and one target at a time" 1>&2; exit 1; } fi nonopt="$ac_option" ;; esac done if test -n "$ac_prev"; then { echo "configure: error: missing argument to --`echo $ac_prev | sed 's/_/-/g'`" 1>&2; exit 1; } fi trap 'rm -fr conftest* confdefs* core core.* *.core $ac_clean_files; exit 1' 1 2 15 # File descriptor usage: # 0 standard input # 1 file creation # 2 errors and warnings # 3 some systems may open it to /dev/tty # 4 used on the Kubota Titan # 6 checking for... messages and results # 5 compiler messages saved in config.log if test "$silent" = yes; then exec 6>/dev/null else exec 6>&1 fi exec 5>./config.log echo "\ This file contains any messages produced by compilers while running configure, to aid debugging if configure makes a mistake. " 1>&5 # Strip out --no-create and --no-recursion so they do not pile up. # Also quote any args containing shell metacharacters. ac_configure_args= for ac_arg do case "$ac_arg" in -no-create | --no-create | --no-creat | --no-crea | --no-cre \ | --no-cr | --no-c) ;; -no-recursion | --no-recursion | --no-recursio | --no-recursi \ | --no-recurs | --no-recur | --no-recu | --no-rec | --no-re | --no-r) ;; *" "*|*" "*|*[\[\]\~\#\$\^\&\*\(\)\{\}\\\|\;\<\>\?]*) ac_configure_args="$ac_configure_args '$ac_arg'" ;; *) ac_configure_args="$ac_configure_args $ac_arg" ;; esac done # NLS nuisances. # Only set LANG and LC_ALL to C if already set. # These must not be set unconditionally because not all systems understand # e.g. LANG=C (notably SCO). if test "${LC_ALL+set}" = set; then LC_ALL=C; export LC_ALL; fi if test "${LANG+set}" = set; then LANG=C; export LANG; fi # confdefs.h avoids OS command line length limits that DEFS can exceed. rm -rf conftest* confdefs.h # AIX cpp loses on an empty file, so make sure it contains at least a newline. echo > confdefs.h # A filename unique to this package, relative to the directory that # configure is in, which we can look for to find out if srcdir is correct. ac_unique_file=Makefile.in # Find the source files, if location was not specified. if test -z "$srcdir"; then ac_srcdir_defaulted=yes # Try the directory containing this script, then its parent. ac_prog=$0 ac_confdir=`echo $ac_prog|sed 's%/[^/][^/]*$%%'` test "x$ac_confdir" = "x$ac_prog" && ac_confdir=. srcdir=$ac_confdir if test ! -r $srcdir/$ac_unique_file; then srcdir=.. fi else ac_srcdir_defaulted=no fi if test ! -r $srcdir/$ac_unique_file; then if test "$ac_srcdir_defaulted" = yes; then { echo "configure: error: can not find sources in $ac_confdir or .." 1>&2; exit 1; } else { echo "configure: error: can not find sources in $srcdir" 1>&2; exit 1; } fi fi srcdir=`echo "${srcdir}" | sed 's%\([^/]\)/*$%\1%'` # Prefer explicitly selected file to automatically selected ones. if test -z "$CONFIG_SITE"; then if test "x$prefix" != xNONE; then CONFIG_SITE="$prefix/share/config.site $prefix/etc/config.site" else CONFIG_SITE="$ac_default_prefix/share/config.site $ac_default_prefix/etc/config.site" fi fi for ac_site_file in $CONFIG_SITE; do if test -r "$ac_site_file"; then echo "loading site script $ac_site_file" . "$ac_site_file" fi done if test -r "$cache_file"; then echo "loading cache $cache_file" . $cache_file else echo "creating cache $cache_file" > $cache_file fi ac_ext=c # CFLAGS is not in ac_cpp because -g, -O, etc. are not valid cpp options. ac_cpp='$CPP $CPPFLAGS' ac_compile='${CC-cc} -c $CFLAGS $CPPFLAGS conftest.$ac_ext 1>&5 2>&5' ac_link='${CC-cc} -o conftest $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS 1>&5 2>&5' if (echo "testing\c"; echo 1,2,3) | grep c >/dev/null; then # Stardent Vistra SVR4 grep lacks -e, says ghazi@caip.rutgers.edu. if (echo -n testing; echo 1,2,3) | sed s/-n/xn/ | grep xn >/dev/null; then ac_n= ac_c=' ' ac_t=' ' else ac_n=-n ac_c= ac_t= fi else ac_n= ac_c='\c' ac_t= fi all_subdirs=`cat $srcdir/*/PLUGIN/REQ $srcdir/*/PLUGIN/OPT /dev/null | tsort | xargs echo` req_subdirs=`cat $srcdir/*/PLUGIN/REQ /dev/null | tsort | xargs echo` opt_subdirs=`cat $srcdir/*/PLUGIN/OPT /dev/null | tsort | xargs echo` ac_aux_dir= for ac_dir in $srcdir $srcdir/.. $srcdir/../..; do if test -f $ac_dir/install-sh; then ac_aux_dir=$ac_dir ac_install_sh="$ac_aux_dir/install-sh -c" break elif test -f $ac_dir/install.sh; then ac_aux_dir=$ac_dir ac_install_sh="$ac_aux_dir/install.sh -c" break fi done if test -z "$ac_aux_dir"; then { echo "configure: error: can not find install-sh or install.sh in $srcdir $srcdir/.. $srcdir/../.." 1>&2; exit 1; } fi ac_config_guess=$ac_aux_dir/config.guess ac_config_sub=$ac_aux_dir/config.sub ac_configure=$ac_aux_dir/configure # This should be Cygnus configure. subdirs="$all_subdirs" existingdirs= for d in $all_subdirs; do if test -d $srcdir/$d ; then existingdirs="$existingdirs $d" if test -f $srcdir/$d/PLUGIN/greet ; then cat $srcdir/$d/PLUGIN/greet else echo === echo === Configuring plug-in component $d echo === fi fi done for d in $req_subdirs; do if test ! -d $srcdir/$d ; then echo "*******" echo "*******" echo "**\+/**" echo "**=*=**" ERROR: Missing required package: $d echo "**/+\**" echo "*******" echo "*******" exit 1 fi done for d in $opt_subdirs; do if test ! -d $srcdir/$d ; then echo "*****" echo "*===*" echo "*=*=*" WARNING: Missing suggested package: $d echo "*===*" echo "*****" fi done trap '' 1 2 15 cat > confcache <<\EOF # This file is a shell script that caches the results of configure # tests run on this system so they can be shared between configure # scripts and configure runs. It is not useful on other systems. # If it contains results you don't want to keep, you may remove or edit it. # # By default, configure uses ./config.cache as the cache file, # creating it if it does not exist already. You can give configure # the --cache-file=FILE option to use a different cache file; that is # what configure does when it calls configure scripts in # subdirectories, so they share the cache. # Giving --cache-file=/dev/null disables caching, for debugging configure. # config.status only pays attention to the cache file if you give it the # --recheck option to rerun configure. # EOF # Ultrix sh set writes to stderr and can't be redirected directly, # and sets the high bit in the cache file unless we assign to the vars. (set) 2>&1 | sed -n "s/^\([a-zA-Z0-9_]*_cv_[a-zA-Z0-9_]*\)=\(.*\)/\1=\${\1='\2'}/p" \ >> confcache if cmp -s $cache_file confcache; then : else if test -w $cache_file; then echo "updating cache $cache_file" cat confcache > $cache_file else echo "not updating unwritable cache $cache_file" fi fi rm -f confcache trap 'rm -fr conftest* confdefs* core core.* *.core $ac_clean_files; exit 1' 1 2 15 test "x$prefix" = xNONE && prefix=$ac_default_prefix # Let make expand exec_prefix. test "x$exec_prefix" = xNONE && exec_prefix='${prefix}' # Any assignment to VPATH causes Sun make to only execute # the first set of double-colon rules, so remove it if not needed. # If there is a colon in the path, we need to keep it. if test "x$srcdir" = x.; then ac_vpsub='/^[ ]*VPATH[ ]*=[^:]*$/d' fi trap 'rm -f $CONFIG_STATUS conftest*; exit 1' 1 2 15 # Transform confdefs.h into DEFS. # Protect against shell expansion while executing Makefile rules. # Protect against Makefile macro expansion. cat > conftest.defs <<\EOF s%#define \([A-Za-z_][A-Za-z0-9_]*\) \(.*\)%-D\1=\2%g s%[ `~#$^&*(){}\\|;'"<>?]%\\&%g s%\[%\\&%g s%\]%\\&%g s%\$%$$%g EOF DEFS=`sed -f conftest.defs confdefs.h | tr '\012' ' '` rm -f conftest.defs # Without the "./", some shells look in PATH for config.status. : ${CONFIG_STATUS=./config.status} echo creating $CONFIG_STATUS rm -f $CONFIG_STATUS cat > $CONFIG_STATUS </dev/null | sed 1q`: # # $0 $ac_configure_args # # Compiler output produced by configure, useful for debugging # configure, is in ./config.log if it exists. ac_cs_usage="Usage: $CONFIG_STATUS [--recheck] [--version] [--help]" for ac_option do case "\$ac_option" in -recheck | --recheck | --rechec | --reche | --rech | --rec | --re | --r) echo "running \${CONFIG_SHELL-/bin/sh} $0 $ac_configure_args --no-create --no-recursion" exec \${CONFIG_SHELL-/bin/sh} $0 $ac_configure_args --no-create --no-recursion ;; -version | --version | --versio | --versi | --vers | --ver | --ve | --v) echo "$CONFIG_STATUS generated by autoconf version 2.3" exit 0 ;; -help | --help | --hel | --he | --h) echo "\$ac_cs_usage"; exit 0 ;; *) echo "\$ac_cs_usage"; exit 1 ;; esac done ac_given_srcdir=$srcdir trap 'rm -fr `echo "Makefile" | sed "s/:[^ ]*//g"` conftest*; exit 1' 1 2 15 # Protect against being on the right side of a sed subst in config.status. sed 's/%@/@@/; s/@%/@@/; s/%g$/@g/; /@g$/s/[\\\\&%]/\\\\&/g; s/@@/%@/; s/@@/@%/; s/@g$/%g/' > conftest.subs <<\CEOF $ac_vpsub $extrasub s%@CFLAGS@%$CFLAGS%g s%@CPPFLAGS@%$CPPFLAGS%g s%@CXXFLAGS@%$CXXFLAGS%g s%@DEFS@%$DEFS%g s%@LDFLAGS@%$LDFLAGS%g s%@LIBS@%$LIBS%g s%@exec_prefix@%$exec_prefix%g s%@prefix@%$prefix%g s%@program_transform_name@%$program_transform_name%g s%@subdirs@%$subdirs%g s%@existingdirs@%$existingdirs%g CEOF EOF cat >> $CONFIG_STATUS <> $CONFIG_STATUS <<\EOF for ac_file in .. $CONFIG_FILES; do if test "x$ac_file" != x..; then # Support "outfile[:infile]", defaulting infile="outfile.in". case "$ac_file" in *:*) ac_file_in=`echo "$ac_file"|sed 's%.*:%%'` ac_file=`echo "$ac_file"|sed 's%:.*%%'` ;; *) ac_file_in="${ac_file}.in" ;; esac # Adjust relative srcdir, etc. for subdirectories. # Remove last slash and all that follows it. Not all systems have dirname. ac_dir=`echo $ac_file|sed 's%/[^/][^/]*$%%'` if test "$ac_dir" != "$ac_file" && test "$ac_dir" != .; then # The file is in a subdirectory. test ! -d "$ac_dir" && mkdir "$ac_dir" ac_dir_suffix="/`echo $ac_dir|sed 's%^\./%%'`" # A "../" for each directory in $ac_dir_suffix. ac_dots=`echo $ac_dir_suffix|sed 's%/[^/]*%../%g'` else ac_dir_suffix= ac_dots= fi case "$ac_given_srcdir" in .) srcdir=. if test -z "$ac_dots"; then top_srcdir=. else top_srcdir=`echo $ac_dots|sed 's%/$%%'`; fi ;; /*) srcdir="$ac_given_srcdir$ac_dir_suffix"; top_srcdir="$ac_given_srcdir" ;; *) # Relative path. srcdir="$ac_dots$ac_given_srcdir$ac_dir_suffix" top_srcdir="$ac_dots$ac_given_srcdir" ;; esac echo creating "$ac_file" rm -f "$ac_file" configure_input="Generated automatically from `echo $ac_file_in|sed 's%.*/%%'` by configure." case "$ac_file" in *Makefile*) ac_comsub="1i\\ # $configure_input" ;; *) ac_comsub= ;; esac sed -e "$ac_comsub s%@configure_input@%$configure_input%g s%@srcdir@%$srcdir%g s%@top_srcdir@%$top_srcdir%g " -f conftest.subs $ac_given_srcdir/$ac_file_in > $ac_file fi; done rm -f conftest.subs exit 0 EOF chmod +x $CONFIG_STATUS rm -fr confdefs* $ac_clean_files test "$no_create" = yes || ${CONFIG_SHELL-/bin/sh} $CONFIG_STATUS || exit 1 if test "$no_recursion" != yes; then # Remove --cache-file and --srcdir arguments so they do not pile up. ac_sub_configure_args= ac_prev= for ac_arg in $ac_configure_args; do if test -n "$ac_prev"; then ac_prev= continue fi case "$ac_arg" in -cache-file | --cache-file | --cache-fil | --cache-fi \ | --cache-f | --cache- | --cache | --cach | --cac | --ca | --c) ac_prev=cache_file ;; -cache-file=* | --cache-file=* | --cache-fil=* | --cache-fi=* \ | --cache-f=* | --cache-=* | --cache=* | --cach=* | --cac=* | --ca=* | --c=*) ;; -srcdir | --srcdir | --srcdi | --srcd | --src | --sr) ac_prev=srcdir ;; -srcdir=* | --srcdir=* | --srcdi=* | --srcd=* | --src=* | --sr=*) ;; *) ac_sub_configure_args="$ac_sub_configure_args $ac_arg" ;; esac done for ac_config_dir in $all_subdirs; do # Do not complain, so a configure script can configure whichever # parts of a large source tree are present. if test ! -d $srcdir/$ac_config_dir; then continue fi echo configuring in $ac_config_dir case "$srcdir" in .) ;; *) if test -d ./$ac_config_dir || mkdir ./$ac_config_dir; then :; else { echo "configure: error: can not create `pwd`/$ac_config_dir" 1>&2; exit 1; } fi ;; esac ac_popdir=`pwd` cd $ac_config_dir case "$srcdir" in .) # No --srcdir option. We are building in place. ac_sub_srcdir=$srcdir ;; /*) # Absolute path. ac_sub_srcdir=$srcdir/$ac_config_dir ;; *) # Relative path. ac_sub_srcdir=../$srcdir/$ac_config_dir ;; esac # Check for guested configure; otherwise get Cygnus style configure. if test -f $ac_sub_srcdir/configure; then ac_sub_configure=$ac_sub_srcdir/configure elif test -f $ac_sub_srcdir/configure.in; then ac_sub_configure=$ac_configure else echo "configure: warning: no configuration information is in $ac_config_dir" 1>&2 ac_sub_configure= fi # The recursion is here. if test -n "$ac_sub_configure"; then # Make the cache file name correct relative to the subdirectory. # A "../" for each directory in /$ac_config_dir. ac_dots=`echo $ac_config_dir|sed -e 's%^\./%%' -e 's%[^/]$%&/%' -e 's%[^/]*/%../%g'` case "$cache_file" in /*) ac_sub_cache_file=$cache_file ;; *) # Relative path. ac_sub_cache_file="$ac_dots$cache_file" ;; esac echo "running ${CONFIG_SHELL-/bin/sh} $ac_sub_configure $ac_sub_configure_args --cache-file=$ac_sub_cache_file --srcdir=$ac_sub_srcdir" # The eval makes quoting arguments work. if eval ${CONFIG_SHELL-/bin/sh} $ac_sub_configure $ac_sub_configure_args --cache-file=$ac_sub_cache_file --srcdir=$ac_sub_srcdir then : else { echo "configure: error: $ac_sub_configure failed for $ac_config_dir" 1>&2; exit 1; } fi fi cd $ac_popdir done fi dist/configure.in100660 765 144 2050 6071421124 12667 0ustar lordusersAC_INIT(Makefile.in) all_subdirs=`cat $srcdir/*/PLUGIN/REQ $srcdir/*/PLUGIN/OPT /dev/null | tsort | xargs echo` req_subdirs=`cat $srcdir/*/PLUGIN/REQ /dev/null | tsort | xargs echo` opt_subdirs=`cat $srcdir/*/PLUGIN/OPT /dev/null | tsort | xargs echo` AC_CONFIG_SUBDIRS($all_subdirs) existingdirs= for d in $all_subdirs; do if test -d $srcdir/$d ; then existingdirs="$existingdirs $d" if test -f $srcdir/$d/PLUGIN/greet ; then cat $srcdir/$d/PLUGIN/greet else echo === echo === Configuring plug-in component $d echo === fi fi done for d in $req_subdirs; do if test ! -d $srcdir/$d ; then echo "*******" echo "*******" echo "**\+/**" echo "**=*=**" ERROR: Missing required package: $d echo "**/+\**" echo "*******" echo "*******" exit 1 fi done for d in $opt_subdirs; do if test ! -d $srcdir/$d ; then echo "*****" echo "*===*" echo "*=*=*" WARNING: Missing suggested package: $d echo "*===*" echo "*****" fi done AC_SUBST(existingdirs) AC_OUTPUT(Makefile) dist/config.sub100570 765 144 41144 6071377322 12377 0ustar lordusers#! /bin/sh # Configuration validation subroutine script, version 1.1. # Copyright (C) 1991, 1992, 1993, 1994, 1995 Free Software Foundation, Inc. # This file is (in principle) common to ALL GNU software. # The presence of a machine in this file suggests that SOME GNU software # can handle that machine. It does not imply ALL GNU software can. # # This file is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by # the Free Software Foundation; either version 2 of the License, or # (at your option) any later version. # # This program is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with this program; if not, write to the Free Software # Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. # As a special exception to the GNU General Public License, if you # distribute this file as part of a program that contains a # configuration script generated by Autoconf, you may include it under # the same distribution terms that you use for the rest of that program. # Configuration subroutine to validate and canonicalize a configuration type. # Supply the specified configuration type as an argument. # If it is invalid, we print an error message on stderr and exit with code 1. # Otherwise, we print the canonical config type on stdout and succeed. # This file is supposed to be the same for all GNU packages # and recognize all the CPU types, system types and aliases # that are meaningful with *any* GNU software. # Each package is responsible for reporting which valid configurations # it does not support. The user should be able to distinguish # a failure to support a valid configuration from a meaningless # configuration. # The goal of this file is to map all the various variations of a given # machine specification into a single specification in the form: # CPU_TYPE-MANUFACTURER-OPERATING_SYSTEM # It is wrong to echo any other type of specification. if [ x$1 = x ] then echo Configuration name missing. 1>&2 echo "Usage: $0 CPU-MFR-OPSYS" 1>&2 echo "or $0 ALIAS" 1>&2 echo where ALIAS is a recognized configuration type. 1>&2 exit 1 fi # First pass through any local machine types. case $1 in *local*) echo $1 exit 0 ;; *) ;; esac # Separate what the user gave into CPU-COMPANY and OS (if any). basic_machine=`echo $1 | sed 's/-[^-]*$//'` if [ $basic_machine != $1 ] then os=`echo $1 | sed 's/.*-/-/'` else os=; fi ### Let's recognize common machines as not being operating systems so ### that things like config.sub decstation-3100 work. We also ### recognize some manufacturers as not being operating systems, so we ### can provide default operating systems below. case $os in -sun*os*) # Prevent following clause from handling this invalid input. ;; -dec* | -mips* | -sequent* | -encore* | -pc532* | -sgi* | -sony* | \ -att* | -7300* | -3300* | -delta* | -motorola* | -sun[234]* | \ -unicom* | -ibm* | -next | -hp | -isi* | -apollo | -altos* | \ -convergent* | -ncr* | -news | -32* | -3600* | -3100* | -hitachi* |\ -c[123]* | -convex* | -sun | -crds | -omron* | -dg | -ultra | -tti* | \ -harris | -dolphin | -highlevel | -gould | -cbm | -ns | -masscomp ) os= basic_machine=$1 ;; -hiux*) os=-hiuxwe2 ;; -sco4) os=-sco3.2v4 basic_machine=`echo $1 | sed -e 's/86-.*/86-unknown/'` ;; -sco3.2.[4-9]*) os=`echo $os | sed -e 's/sco3.2./sco3.2v/'` basic_machine=`echo $1 | sed -e 's/86-.*/86-unknown/'` ;; -sco3.2v[4-9]*) # Don't forget version if it is 3.2v4 or newer. basic_machine=`echo $1 | sed -e 's/86-.*/86-unknown/'` ;; -sco*) os=-sco3.2v2 basic_machine=`echo $1 | sed -e 's/86-.*/86-unknown/'` ;; -isc) os=-isc2.2 basic_machine=`echo $1 | sed -e 's/86-.*/86-unknown/'` ;; -clix*) basic_machine=clipper-intergraph ;; -isc*) basic_machine=`echo $1 | sed -e 's/86-.*/86-unknown/'` ;; -lynx*) os=-lynxos ;; -ptx*) basic_machine=`echo $1 | sed -e 's/86-.*/86-sequent/'` ;; -windowsnt*) os=`echo $os | sed -e 's/windowsnt/winnt/'` ;; esac # Decode aliases for certain CPU-COMPANY combinations. case $basic_machine in # Recognize the basic CPU types without company name. # Some are omitted here because they have special meanings below. tahoe | i[345]86 | i860 | m68k | m68000 | m88k | ns32k | arm | pyramid \ | tron | a29k | 580 | i960 | h8300 | hppa1.0 | hppa1.1 \ | alpha | we32k | ns16k | clipper | sparclite | i370 | sh \ | powerpc | sparc64 | 1750a | dsp16xx | mips64 | mipsel \ | pdp11 | mips64el | mips64orion | mips64orionel \ | sparc) basic_machine=$basic_machine-unknown ;; # Object if more than one company name word. *-*-*) echo Invalid configuration \`$1\': machine \`$basic_machine\' not recognized 1>&2 exit 1 ;; # Recognize the basic CPU types with company name. vax-* | tahoe-* | i[345]86-* | i860-* | m68k-* | m68000-* | m88k-* \ | sparc-* | ns32k-* | fx80-* | arm-* | c[123]* \ | mips-* | pyramid-* | tron-* | a29k-* | romp-* | rs6000-* \ | none-* | 580-* | cray2-* | h8300-* | i960-* | xmp-* | ymp-* \ | hppa1.0-* | hppa1.1-* | alpha-* | we32k-* | cydra-* | ns16k-* \ | pn-* | np1-* | xps100-* | clipper-* | orion-* | sparclite-* \ | pdp11-* | sh-* | powerpc-* | sparc64-* | mips64-* | mipsel-* \ | mips64el-* | mips64orion-* | mips64orionel-*) ;; # Recognize the various machine names and aliases which stand # for a CPU type and a company and sometimes even an OS. 3b1 | 7300 | 7300-att | att-7300 | pc7300 | safari | unixpc) basic_machine=m68000-att ;; 3b*) basic_machine=we32k-att ;; alliant | fx80) basic_machine=fx80-alliant ;; altos | altos3068) basic_machine=m68k-altos ;; am29k) basic_machine=a29k-none os=-bsd ;; amdahl) basic_machine=580-amdahl os=-sysv ;; amiga | amiga-*) basic_machine=m68k-cbm ;; amigados) basic_machine=m68k-cbm os=-amigados ;; amigaunix | amix) basic_machine=m68k-cbm os=-sysv4 ;; apollo68) basic_machine=m68k-apollo os=-sysv ;; balance) basic_machine=ns32k-sequent os=-dynix ;; convex-c1) basic_machine=c1-convex os=-bsd ;; convex-c2) basic_machine=c2-convex os=-bsd ;; convex-c32) basic_machine=c32-convex os=-bsd ;; convex-c34) basic_machine=c34-convex os=-bsd ;; convex-c38) basic_machine=c38-convex os=-bsd ;; cray | ymp) basic_machine=ymp-cray os=-unicos ;; cray2) basic_machine=cray2-cray os=-unicos ;; crds | unos) basic_machine=m68k-crds ;; da30 | da30-*) basic_machine=m68k-da30 ;; decstation | decstation-3100 | pmax | pmax-* | pmin | dec3100 | decstatn) basic_machine=mips-dec ;; delta | 3300 | motorola-3300 | motorola-delta \ | 3300-motorola | delta-motorola) basic_machine=m68k-motorola ;; delta88) basic_machine=m88k-motorola os=-sysv3 ;; dpx20 | dpx20-*) basic_machine=rs6000-bull os=-bosx ;; dpx2* | dpx2*-bull) basic_machine=m68k-bull os=-sysv3 ;; ebmon29k) basic_machine=a29k-amd os=-ebmon ;; elxsi) basic_machine=elxsi-elxsi os=-bsd ;; encore | umax | mmax) basic_machine=ns32k-encore ;; fx2800) basic_machine=i860-alliant ;; genix) basic_machine=ns32k-ns ;; gmicro) basic_machine=tron-gmicro os=-sysv ;; h3050r* | hiux*) basic_machine=hppa1.1-hitachi os=-hiuxwe2 ;; h8300hms) basic_machine=h8300-hitachi os=-hms ;; harris) basic_machine=m88k-harris os=-sysv3 ;; hp300-*) basic_machine=m68k-hp ;; hp300bsd) basic_machine=m68k-hp os=-bsd ;; hp300hpux) basic_machine=m68k-hp os=-hpux ;; hp9k2[0-9][0-9] | hp9k31[0-9]) basic_machine=m68000-hp ;; hp9k3[2-9][0-9]) basic_machine=m68k-hp ;; hp9k7[0-9][0-9] | hp7[0-9][0-9] | hp9k8[0-9]7 | hp8[0-9]7) basic_machine=hppa1.1-hp ;; hp9k8[0-9][0-9] | hp8[0-9][0-9]) basic_machine=hppa1.0-hp ;; i370-ibm* | ibm*) basic_machine=i370-ibm os=-mvs ;; # I'm not sure what "Sysv32" means. Should this be sysv3.2? i[345]86v32) basic_machine=`echo $1 | sed -e 's/86.*/86-unknown/'` os=-sysv32 ;; i[345]86v4*) basic_machine=`echo $1 | sed -e 's/86.*/86-unknown/'` os=-sysv4 ;; i[345]86v) basic_machine=`echo $1 | sed -e 's/86.*/86-unknown/'` os=-sysv ;; i[345]86sol2) basic_machine=`echo $1 | sed -e 's/86.*/86-unknown/'` os=-solaris2 ;; iris | iris4d) basic_machine=mips-sgi case $os in -irix*) ;; *) os=-irix4 ;; esac ;; isi68 | isi) basic_machine=m68k-isi os=-sysv ;; m88k-omron*) basic_machine=m88k-omron ;; magnum | m3230) basic_machine=mips-mips os=-sysv ;; merlin) basic_machine=ns32k-utek os=-sysv ;; miniframe) basic_machine=m68000-convergent ;; mips3*-*) basic_machine=`echo $basic_machine | sed -e 's/mips3/mips64/'` ;; mips3*) basic_machine=`echo $basic_machine | sed -e 's/mips3/mips64/'`-unknown ;; ncr3000) basic_machine=i486-ncr os=-sysv4 ;; news | news700 | news800 | news900) basic_machine=m68k-sony os=-newsos ;; news1000) basic_machine=m68030-sony os=-newsos ;; news-3600 | risc-news) basic_machine=mips-sony os=-newsos ;; next | m*-next ) basic_machine=m68k-next case $os in -nextstep* ) ;; -ns2*) os=-nextstep2 ;; *) os=-nextstep3 ;; esac ;; nh3000) basic_machine=m68k-harris os=-cxux ;; nh[45]000) basic_machine=m88k-harris os=-cxux ;; nindy960) basic_machine=i960-intel os=-nindy ;; np1) basic_machine=np1-gould ;; pa-hitachi) basic_machine=hppa1.1-hitachi os=-hiuxwe2 ;; paragon) basic_machine=i860-intel os=-osf ;; pbd) basic_machine=sparc-tti ;; pbb) basic_machine=m68k-tti ;; pc532 | pc532-*) basic_machine=ns32k-pc532 ;; pentium-*) # We will change tis to say i586 once there has been # time for various packages to start to recognize that. basic_machine=i486-`echo $basic_machine | sed 's/^[^-]*-//'` ;; pn) basic_machine=pn-gould ;; ps2) basic_machine=i386-ibm ;; rm[46]00) basic_machine=mips-siemens ;; rtpc | rtpc-*) basic_machine=romp-ibm ;; sequent) basic_machine=i386-sequent ;; sh) basic_machine=sh-hitachi os=-hms ;; sps7) basic_machine=m68k-bull os=-sysv2 ;; spur) basic_machine=spur-unknown ;; sun2) basic_machine=m68000-sun ;; sun2os3) basic_machine=m68000-sun os=-sunos3 ;; sun2os4) basic_machine=m68000-sun os=-sunos4 ;; sun3os3) basic_machine=m68k-sun os=-sunos3 ;; sun3os4) basic_machine=m68k-sun os=-sunos4 ;; sun4os3) basic_machine=sparc-sun os=-sunos3 ;; sun4os4) basic_machine=sparc-sun os=-sunos4 ;; sun4sol2) basic_machine=sparc-sun os=-solaris2 ;; sun3 | sun3-*) basic_machine=m68k-sun ;; sun4) basic_machine=sparc-sun ;; sun386 | sun386i | roadrunner) basic_machine=i386-sun ;; symmetry) basic_machine=i386-sequent os=-dynix ;; tower | tower-32) basic_machine=m68k-ncr ;; udi29k) basic_machine=a29k-amd os=-udi ;; ultra3) basic_machine=a29k-nyu os=-sym1 ;; vaxv) basic_machine=vax-dec os=-sysv ;; vms) basic_machine=vax-dec os=-vms ;; vxworks960) basic_machine=i960-wrs os=-vxworks ;; vxworks68) basic_machine=m68k-wrs os=-vxworks ;; xmp) basic_machine=xmp-cray os=-unicos ;; xps | xps100) basic_machine=xps100-honeywell ;; none) basic_machine=none-none os=-none ;; # Here we handle the default manufacturer of certain CPU types. It is in # some cases the only manufacturer, in others, it is the most popular. mips) basic_machine=mips-mips ;; romp) basic_machine=romp-ibm ;; rs6000) basic_machine=rs6000-ibm ;; vax) basic_machine=vax-dec ;; pdp11) basic_machine=pdp11-dec ;; we32k) basic_machine=we32k-att ;; sparc) basic_machine=sparc-sun ;; cydra) basic_machine=cydra-cydrome ;; orion) basic_machine=orion-highlevel ;; orion105) basic_machine=clipper-highlevel ;; *) echo Invalid configuration \`$1\': machine \`$basic_machine\' not recognized 1>&2 exit 1 ;; esac # Here we canonicalize certain aliases for manufacturers. case $basic_machine in *-digital*) basic_machine=`echo $basic_machine | sed 's/digital.*/dec/'` ;; *-commodore*) basic_machine=`echo $basic_machine | sed 's/commodore.*/cbm/'` ;; *) ;; esac # Decode manufacturer-specific aliases for certain operating systems. if [ x"$os" != x"" ] then case $os in # -solaris* is a basic system type, with this one exception. -solaris1 | -solaris1.*) os=`echo $os | sed -e 's|solaris1|sunos4|'` ;; -solaris) os=-solaris2 ;; -unixware* | svr4*) os=-sysv4 ;; -gnu/linux*) os=`echo $os | sed -e 's|gnu/linux|linux|'` ;; # First accept the basic system types. # The portable systems comes first. # Each alternative must end in a *, to match a version number. # -sysv* is not here because it comes later, after sysvr4. -gnu* | -bsd* | -mach* | -minix* | -genix* | -ultrix* | -irix* \ | -vms* | -sco* | -esix* | -isc* | -aix* | -sunos | -sunos[345]* \ | -hpux* | -unos* | -osf* | -luna* | -dgux* | -solaris* | -sym* \ | -amigados* | -msdos* | -newsos* | -unicos* | -aos* \ | -nindy* | -vxworks* | -ebmon* | -hms* | -mvs* | -clix* \ | -riscos* | -linux* | -uniplus* | -iris* | -rtu* | -xenix* \ | -hiux* | -386bsd* | -netbsd* | -freebsd* | -riscix* \ | -lynxos* | -bosx* | -nextstep* | -cxux* | -aout* | -elf* \ | -ptx* | -coff* | -ecoff* | -winnt* | -domain* | -vsta | -udi | -eabi) ;; -sunos5*) os=`echo $os | sed -e 's|sunos5|solaris2|'` ;; -sunos6*) os=`echo $os | sed -e 's|sunos6|solaris3|'` ;; -osfrose*) os=-osfrose ;; -osf*) os=-osf ;; -utek*) os=-bsd ;; -dynix*) os=-bsd ;; -acis*) os=-aos ;; -ctix* | -uts*) os=-sysv ;; # Preserve the version number of sinix5. -sinix5.*) os=`echo $os | sed -e 's|sinix|sysv|'` ;; -sinix*) os=-sysv4 ;; -triton*) os=-sysv3 ;; -oss*) os=-sysv3 ;; -svr4) os=-sysv4 ;; -svr3) os=-sysv3 ;; -sysvr4) os=-sysv4 ;; # This must come after -sysvr4. -sysv*) ;; -xenix) os=-xenix ;; -none) ;; *) # Get rid of the `-' at the beginning of $os. os=`echo $os | sed 's/[^-]*-//'` echo Invalid configuration \`$1\': system \`$os\' not recognized 1>&2 exit 1 ;; esac else # Here we handle the default operating systems that come with various machines. # The value should be what the vendor currently ships out the door with their # machine or put another way, the most popular os provided with the machine. # Note that if you're going to try to match "-MANUFACTURER" here (say, # "-sun"), then you have to tell the case statement up towards the top # that MANUFACTURER isn't an operating system. Otherwise, code above # will signal an error saying that MANUFACTURER isn't an operating # system, and we'll never get to this point. case $basic_machine in *-acorn) os=-riscix1.2 ;; pdp11-*) os=-none ;; *-dec | vax-*) os=-ultrix4.2 ;; m68*-apollo) os=-domain ;; i386-sun) os=-sunos4.0.2 ;; m68000-sun) os=-sunos3 # This also exists in the configure program, but was not the # default. # os=-sunos4 ;; *-tti) # must be before sparc entry or we get the wrong os. os=-sysv3 ;; sparc-* | *-sun) os=-sunos4.1.1 ;; *-ibm) os=-aix ;; *-hp) os=-hpux ;; *-hitachi) os=-hiux ;; i860-* | *-att | *-ncr | *-altos | *-motorola | *-convergent) os=-sysv ;; *-cbm) os=-amigados ;; *-dg) os=-dgux ;; *-dolphin) os=-sysv3 ;; m68k-ccur) os=-rtu ;; m88k-omron*) os=-luna ;; *-sequent) os=-ptx ;; *-crds) os=-unos ;; *-ns) os=-genix ;; i370-*) os=-mvs ;; *-next) os=-nextstep3 ;; *-gould) os=-sysv ;; *-highlevel) os=-bsd ;; *-encore) os=-bsd ;; *-sgi) os=-irix ;; *-siemens) os=-sysv4 ;; *-masscomp) os=-rtu ;; *) os=-none ;; esac fi # Here we handle the case where we know the os, and the CPU type, but not the # manufacturer. We pick the logical manufacturer. vendor=unknown case $basic_machine in *-unknown) case $os in -riscix*) vendor=acorn ;; -sunos*) vendor=sun ;; -lynxos*) vendor=lynx ;; -aix*) vendor=ibm ;; -hpux*) vendor=hp ;; -hiux*) vendor=hitachi ;; -unos*) vendor=crds ;; -dgux*) vendor=dg ;; -luna*) vendor=omron ;; -genix*) vendor=ns ;; -mvs*) vendor=ibm ;; -ptx*) vendor=sequent ;; -vxworks*) vendor=wrs ;; esac basic_machine=`echo $basic_machine | sed "s/unknown/$vendor/"` ;; esac echo $basic_machine$os dist/config.guess100570 765 144 33150 6071377322 12732 0ustar lordusers#! /bin/sh # Attempt to guess a canonical system name. # Copyright (C) 1992, 1993, 1994, 1995 Free Software Foundation, Inc. # # This file is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by # the Free Software Foundation; either version 2 of the License, or # (at your option) any later version. # # This program is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU # General Public License for more details. # # You should have received a copy of the GNU General Public License # along with this program; if not, write to the Free Software # Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. # # As a special exception to the GNU General Public License, if you # distribute this file as part of a program that contains a # configuration script generated by Autoconf, you may include it under # the same distribution terms that you use for the rest of that program. # Written by Per Bothner . # The master version of this file is at the FSF in /home/gd/gnu/lib. # # This script attempts to guess a canonical system name similar to # config.sub. If it succeeds, it prints the system name on stdout, and # exits with 0. Otherwise, it exits with 1. # # The plan is that this can be called by configure scripts if you # don't specify an explicit system type (host/target name). # # Only a few systems have been added to this list; please add others # (but try to keep the structure clean). # # This is needed to find uname on a Pyramid OSx when run in the BSD universe. # (ghazi@noc.rutgers.edu 8/24/94.) if (test -f /.attbin/uname) >/dev/null 2>&1 ; then PATH=$PATH:/.attbin ; export PATH fi UNAME_MACHINE=`(uname -m) 2>/dev/null` || UNAME_MACHINE=unknown UNAME_RELEASE=`(uname -r) 2>/dev/null` || UNAME_RELEASE=unknown UNAME_SYSTEM=`(uname -s) 2>/dev/null` || UNAME_SYSTEM=unknown UNAME_VERSION=`(uname -v) 2>/dev/null` || UNAME_VERSION=unknown trap 'rm -f dummy.c dummy.o dummy; exit 1' 1 2 15 # Note: order is significant - the case branches are not exclusive. case "${UNAME_MACHINE}:${UNAME_SYSTEM}:${UNAME_RELEASE}:${UNAME_VERSION}" in alpha:OSF1:V*:*) # After 1.2, OSF1 uses "V1.3" for uname -r. echo alpha-dec-osf`echo ${UNAME_RELEASE} | sed -e 's/^V//'` exit 0 ;; alpha:OSF1:*:*) # 1.2 uses "1.2" for uname -r. echo alpha-dec-osf${UNAME_RELEASE} exit 0 ;; amiga:NetBSD:*:*) echo m68k-cbm-netbsd${UNAME_RELEASE} exit 0 ;; arm:RISC*:1.[012]*:*|arm:riscix:1.[012]*:*) echo arm-acorn-riscix${UNAME_RELEASE} exit 0;; Pyramid*:OSx*:*:*) if test "`(/bin/universe) 2>/dev/null`" = att ; then echo pyramid-pyramid-sysv3 else echo pyramid-pyramid-bsd fi exit 0 ;; sun4*:SunOS:5.*:*) echo sparc-sun-solaris2`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'` exit 0 ;; sun4*:SunOS:6*:*) # According to config.sub, this is the proper way to canonicalize # SunOS6. Hard to guess exactly what SunOS6 will be like, but # it's likely to be more like Solaris than SunOS4. echo sparc-sun-solaris3`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'` exit 0 ;; sun4*:SunOS:*:*) case "`/usr/bin/arch -k`" in Series*|S4*) UNAME_RELEASE=`uname -v` ;; esac # Japanese Language versions have a version number like `4.1.3-JL'. echo sparc-sun-sunos`echo ${UNAME_RELEASE}|sed -e 's/-/_/'` exit 0 ;; sun3*:SunOS:*:*) echo m68k-sun-sunos${UNAME_RELEASE} exit 0 ;; RISC*:ULTRIX:*:*) echo mips-dec-ultrix${UNAME_RELEASE} exit 0 ;; VAX*:ULTRIX*:*:*) echo vax-dec-ultrix${UNAME_RELEASE} exit 0 ;; mips:*:5*:RISCos) echo mips-mips-riscos${UNAME_RELEASE} exit 0 ;; m88k:CX/UX:7*:*) echo m88k-harris-cxux7 exit 0 ;; m88k:*:4*:R4*) echo m88k-motorola-sysv4 exit 0 ;; m88k:*:3*:R3*) echo m88k-motorola-sysv3 exit 0 ;; AViiON:dgux:*:*) if [ ${TARGET_BINARY_INTERFACE}x = m88kdguxelfx \ -o ${TARGET_BINARY_INTERFACE}x = x ] ; then echo m88k-dg-dgux${UNAME_RELEASE} else echo m88k-dg-dguxbcs${UNAME_RELEASE} fi exit 0 ;; M88*:DolphinOS:*:*) # DolphinOS (SVR3) echo m88k-dolphin-sysv3 exit 0 ;; M88*:*:R3*:*) # Delta 88k system running SVR3 echo m88k-motorola-sysv3 exit 0 ;; XD88*:*:*:*) # Tektronix XD88 system running UTekV (SVR3) echo m88k-tektronix-sysv3 exit 0 ;; Tek43[0-9][0-9]:UTek:*:*) # Tektronix 4300 system running UTek (BSD) echo m68k-tektronix-bsd exit 0 ;; *:IRIX*:*:*) echo mips-sgi-irix`echo ${UNAME_RELEASE}|sed -e 's/-/_/g'` exit 0 ;; ????????:AIX?:[12].1:2) # AIX 2.2.1 or AIX 2.1.1 is RT/PC AIX. echo romp-ibm-aix # uname -m gives an 8 hex-code CPU id exit 0 ;; # Note that: echo "'`uname -s`'" gives 'AIX ' i[34]86:AIX:*:*) echo i386-ibm-aix exit 0 ;; *:AIX:2:3) if grep bos325 /usr/include/stdio.h >/dev/null 2>&1; then sed 's/^ //' << EOF >dummy.c #include main() { if (!__power_pc()) exit(1); puts("powerpc-ibm-aix3.2.5"); exit(0); } EOF ${CC-cc} dummy.c -o dummy && ./dummy && rm dummy.c dummy && exit 0 rm -f dummy.c dummy echo rs6000-ibm-aix3.2.5 elif grep bos324 /usr/include/stdio.h >/dev/null 2>&1; then echo rs6000-ibm-aix3.2.4 else echo rs6000-ibm-aix3.2 fi exit 0 ;; *:AIX:*:4) if /usr/sbin/lsattr -EHl proc0 | grep POWER >/dev/null 2>&1; then IBM_ARCH=rs6000 else IBM_ARCH=powerpc fi if grep bos410 /usr/include/stdio.h >/dev/null 2>&1; then IBM_REV=4.1 elif grep bos411 /usr/include/stdio.h >/dev/null 2>&1; then IBM_REV=4.1.1 else IBM_REV=4.${UNAME_RELEASE} fi echo ${IBM_ARCH}-ibm-aix${IBM_REV} exit 0 ;; *:AIX:*:*) echo rs6000-ibm-aix exit 0 ;; ibmrt:4.4BSD:*|romp-ibm:BSD:*) echo romp-ibm-bsd4.4 exit 0 ;; ibmrt:*BSD:*|romp-ibm:BSD:*) # covers RT/PC NetBSD and echo romp-ibm-bsd${UNAME_RELEASE} # 4.3 with uname added to exit 0 ;; # report: romp-ibm BSD 4.3 *:BOSX:*:*) echo rs6000-bull-bosx exit 0 ;; DPX/2?00:B.O.S.:*:*) echo m68k-bull-sysv3 exit 0 ;; 9000/[34]??:4.3bsd:1.*:*) echo m68k-hp-bsd exit 0 ;; hp300:4.4BSD:*:* | 9000/[34]??:4.3bsd:2.*:*) echo m68k-hp-bsd4.4 exit 0 ;; 9000/[3478]??:HP-UX:*:*) case "${UNAME_MACHINE}" in 9000/31? ) HP_ARCH=m68000 ;; 9000/[34]?? ) HP_ARCH=m68k ;; 9000/7?? | 9000/8?7 ) HP_ARCH=hppa1.1 ;; 9000/8?? ) HP_ARCH=hppa1.0 ;; esac HPUX_REV=`echo ${UNAME_RELEASE}|sed -e 's/[^.]*.[0B]*//'` echo ${HP_ARCH}-hp-hpux${HPUX_REV} exit 0 ;; 3050*:HI-UX:*:*) sed 's/^ //' << EOF >dummy.c #include int main () { long cpu = sysconf (_SC_CPU_VERSION); /* The order matters, because CPU_IS_HP_MC68K erroneously returns true for CPU_PA_RISC1_0. CPU_IS_PA_RISC returns correct results, however. */ if (CPU_IS_PA_RISC (cpu)) { switch (cpu) { case CPU_PA_RISC1_0: puts ("hppa1.0-hitachi-hiuxwe2"); break; case CPU_PA_RISC1_1: puts ("hppa1.1-hitachi-hiuxwe2"); break; case CPU_PA_RISC2_0: puts ("hppa2.0-hitachi-hiuxwe2"); break; default: puts ("hppa-hitachi-hiuxwe2"); break; } } else if (CPU_IS_HP_MC68K (cpu)) puts ("m68k-hitachi-hiuxwe2"); else puts ("unknown-hitachi-hiuxwe2"); exit (0); } EOF ${CC-cc} dummy.c -o dummy && ./dummy && rm dummy.c dummy && exit 0 rm -f dummy.c dummy echo unknown-hitachi-hiuxwe2 exit 0 ;; 9000/7??:4.3bsd:*:* | 9000/8?7:4.3bsd:*:* ) echo hppa1.1-hp-bsd exit 0 ;; 9000/8??:4.3bsd:*:*) echo hppa1.0-hp-bsd exit 0 ;; hp7??:OSF1:*:* | hp8?7:OSF1:*:* ) echo hppa1.1-hp-osf exit 0 ;; hp8??:OSF1:*:*) echo hppa1.0-hp-osf exit 0 ;; C1*:ConvexOS:*:* | convex:ConvexOS:C1*:*) echo c1-convex-bsd exit 0 ;; C2*:ConvexOS:*:* | convex:ConvexOS:C2*:*) if getsysinfo -f scalar_acc then echo c32-convex-bsd else echo c2-convex-bsd fi exit 0 ;; C34*:ConvexOS:*:* | convex:ConvexOS:C34*:*) echo c34-convex-bsd exit 0 ;; C38*:ConvexOS:*:* | convex:ConvexOS:C38*:*) echo c38-convex-bsd exit 0 ;; C4*:ConvexOS:*:* | convex:ConvexOS:C4*:*) echo c4-convex-bsd exit 0 ;; CRAY*X-MP:UNICOS:*:*) echo xmp-cray-unicos exit 0 ;; CRAY*Y-MP:UNICOS:*:*) echo ymp-cray-unicos exit 0 ;; CRAY-2:UNICOS:*:*) echo cray2-cray-unicos exit 0 ;; hp3[0-9][05]:NetBSD:*:*) echo m68k-hp-netbsd${UNAME_RELEASE} exit 0 ;; i[34]86:BSD/386:*:* | *:BSD/OS:*:*) echo ${UNAME_MACHINE}-unknown-bsdi${UNAME_RELEASE} exit 0 ;; *:FreeBSD:*:*) echo ${UNAME_MACHINE}-unknown-freebsd`echo ${UNAME_RELEASE}|sed -e 's/[-(].*//'` exit 0 ;; *:NetBSD:*:*) echo ${UNAME_MACHINE}-unknown-netbsd`echo ${UNAME_RELEASE}|sed -e 's/[-_].*/\./'` exit 0 ;; *:GNU:*:*) echo `echo ${UNAME_MACHINE}|sed -e 's,/.*$,,'`-unknown-gnu`echo ${UNAME_RELEASE}|sed -e 's,/.*$,,'` exit 0 ;; *:Linux:*:*) echo ${UNAME_MACHINE}-unknown-linux exit 0 ;; # ptx 4.0 does uname -s correctly, with DYNIX/ptx in there. earlier versions # are messed up and put the nodename in both sysname and nodename. i[34]86:DYNIX/ptx:4*:*) echo i386-sequent-sysv4 exit 0 ;; i[34]86:*:4.*:* | i[34]86:SYSTEM_V:4.*:*) if grep Novell /usr/include/link.h >/dev/null 2>/dev/null; then echo ${UNAME_MACHINE}-univel-sysv${UNAME_RELEASE} else echo ${UNAME_MACHINE}-unknown-sysv${UNAME_RELEASE} fi exit 0 ;; i[34]86:*:3.2:*) if test -f /usr/options/cb.name; then UNAME_REL=`sed -n 's/.*Version //p' /dev/null >/dev/null ; then UNAME_REL=`(/bin/uname -X|egrep Release|sed -e 's/.*= //')` (/bin/uname -X|egrep i80486 >/dev/null) && UNAME_MACHINE=i486 echo ${UNAME_MACHINE}-unknown-sco$UNAME_REL else echo ${UNAME_MACHINE}-unknown-sysv32 fi exit 0 ;; Intel:Mach:3*:*) echo i386-unknown-mach3 exit 0 ;; paragon:*:*:*) echo i860-intel-osf1 exit 0 ;; i860:*:4.*:*) # i860-SVR4 if grep Stardent /usr/include/sys/uadmin.h >/dev/null 2>&1 ; then echo i860-stardent-sysv${UNAME_RELEASE} # Stardent Vistra i860-SVR4 else # Add other i860-SVR4 vendors below as they are discovered. echo i860-unknown-sysv${UNAME_RELEASE} # Unknown i860-SVR4 fi exit 0 ;; mini*:CTIX:SYS*5:*) # "miniframe" echo m68010-convergent-sysv exit 0 ;; M680[234]0:*:R3V[567]*:*) test -r /sysV68 && echo 'm68k-motorola-sysv' && exit 0 ;; 3[34]??:*:4.0:3.0 | 3[34]??,*:*:4.0:3.0) uname -p 2>/dev/null | grep 86 >/dev/null \ && echo i486-ncr-sysv4.3 && exit 0 ;; 3[34]??:*:4.0:* | 3[34]??,*:*:4.0:*) uname -p 2>/dev/null | grep 86 >/dev/null \ && echo i486-ncr-sysv4 && exit 0 ;; m680[234]0:LynxOS:2.2*:*) echo m68k-lynx-lynxos${UNAME_RELEASE} exit 0 ;; mc68030:UNIX_System_V:4.*:*) echo m68k-atari-sysv4 exit 0 ;; i[34]86:LynxOS:2.2*:*) echo i386-lynx-lynxos${UNAME_RELEASE} exit 0 ;; TSUNAMI:LynxOS:2.2*:*) echo sparc-lynx-lynxos${UNAME_RELEASE} exit 0 ;; rs6000:LynxOS:2.2*:*) echo rs6000-lynx-lynxos${UNAME_RELEASE} exit 0 ;; RM*:SINIX-*:*:*) echo mips-sni-sysv4 exit 0 ;; *:SINIX-*:*:*) if uname -p 2>/dev/null >/dev/null ; then UNAME_MACHINE=`(uname -p) 2>/dev/null` echo ${UNAME_MACHINE}-sni-sysv4 else echo ns32k-sni-sysv fi exit 0 ;; esac #echo '(No uname command or uname output not recognized.)' 1>&2 #echo "${UNAME_MACHINE}:${UNAME_SYSTEM}:${UNAME_RELEASE}:${UNAME_VERSION}" 1>&2 cat >dummy.c < printf ("m68k-sony-newsos%s\n", #ifdef NEWSOS4 "4" #else "" #endif ); exit (0); #endif #endif #if defined (__arm) && defined (__acorn) && defined (__unix) printf ("arm-acorn-riscix"); exit (0); #endif #if defined (hp300) && !defined (hpux) printf ("m68k-hp-bsd\n"); exit (0); #endif #if defined (NeXT) #if !defined (__ARCHITECTURE__) #define __ARCHITECTURE__ "m68k" #endif int version; version=`(hostinfo | sed -n 's/.*NeXT Mach \([0-9]*\).*/\1/p') 2>/dev/null`; printf ("%s-next-nextstep%s\n", __ARCHITECTURE__, version==2 ? "2" : "3"); exit (0); #endif #if defined (MULTIMAX) || defined (n16) #if defined (UMAXV) printf ("ns32k-encore-sysv\n"); exit (0); #else #if defined (CMU) printf ("ns32k-encore-mach\n"); exit (0); #else printf ("ns32k-encore-bsd\n"); exit (0); #endif #endif #endif #if defined (__386BSD__) printf ("i386-unknown-bsd\n"); exit (0); #endif #if defined (sequent) #if defined (i386) printf ("i386-sequent-dynix\n"); exit (0); #endif #if defined (ns32000) printf ("ns32k-sequent-dynix\n"); exit (0); #endif #endif #if defined (_SEQUENT_) printf ("i386-sequent-ptx\n"); exit (0); #endif #if defined (vax) #if !defined (ultrix) printf ("vax-dec-bsd\n"); exit (0); #else printf ("vax-dec-ultrix\n"); exit (0); #endif #endif #if defined (alliant) && defined (i860) printf ("i860-alliant-bsd\n"); exit (0); #endif exit (1); } EOF ${CC-cc} dummy.c -o dummy 2>/dev/null && ./dummy && rm dummy.c dummy && exit 0 rm -f dummy.c dummy # Apollos put the system type in the environment. test -d /usr/apollo && { echo ${ISP}-apollo-${SYSTYPE}; exit 0; } # Convex versions that predate uname can use getsysinfo(1) if [ -x /usr/convex/getsysinfo ] then case `getsysinfo -f cpu_type` in c1*) echo c1-convex-bsd exit 0 ;; c2*) if getsysinfo -f scalar_acc then echo c32-convex-bsd else echo c2-convex-bsd fi exit 0 ;; c34*) echo c34-convex-bsd exit 0 ;; c38*) echo c38-convex-bsd exit 0 ;; c4*) echo c4-convex-bsd exit 0 ;; esac fi #echo '(Unable to guess system type)' 1>&2 exit 1 dist/doc/ 42770 765 144 0 6071534435 11044 5ustar lordusersdist/doc/rx.info100660 765 144 110710 6071534721 12504 0ustar lordusersThis is Info file rx.info, produced by Makeinfo-1.63 from the input file rx.texi.  File: rx.info, Node: Top, Next: Posix Entry Points, Prev: (dir), Up: (dir) Rx ** This document describes Rx. This document applies to version "Rx-diSpencer" of the library named librx.a * Menu: * Posix Entry Points:: * Performance Considerations:: Special considerations when using Rx. * Semantic Tests:: A test-suite for Posix matcher semantics. * General Functions:: Rx's native entry points. * Rx Theory:: A quick tour of the implementation. * Improvements to Make:: The Rx TODO list.  File: rx.info, Node: Posix Entry Points, Next: Performance Considerations, Prev: Top, Up: Top Regular Expression Matching =========================== This section is excerpted from *The GNU C Library* reference manual by Sandra Loosemore with Richard M. Stallman, Roland McGrath, and Andrew Oram. The GNU C library supports the standard POSIX.2 interface. Programs using this interface should include the header file `rx.h'. * Menu: * POSIX Regexp Compilation:: Using `regcomp' to prepare to match. * Flags for POSIX Regexps:: Syntax variations for `regcomp'. * Matching POSIX Regexps:: Using `regexec' to match the compiled pattern that you get from `regcomp'. * Regexp Subexpressions:: Finding which parts of the string were matched. * Subexpression Complications:: Find points of which parts were matched. * Regexp Cleanup:: Freeing storage; reporting errors.  File: rx.info, Node: POSIX Regexp Compilation, Next: Flags for POSIX Regexps, Up: Posix Entry Points POSIX Regular Expression Compilation ------------------------------------ Before you can actually match a regular expression, you must "compile" it. This is not true compilation--it produces a special data structure, not machine instructions. But it is like ordinary compilation in that its purpose is to enable you to "execute" the pattern fast. (*Note Matching POSIX Regexps::, for how to use the compiled regular expression for matching.) There is a special data type for compiled regular expressions: - Data Type: regex_t This type of object holds a compiled regular expression. It is actually a structure. It has just one field that your programs should look at: `re_nsub' This field holds the number of parenthetical subexpressions in the regular expression that was compiled. There are several other fields, but we don't describe them here, because only the functions in the library should use them. After you create a `regex_t' object, you can compile a regular expression into it by calling `regcomp'. - Function: int regcomp (regex_t *COMPILED, const char *PATTERN, int CFLAGS) The function `regcomp' "compiles" a regular expression into a data structure that you can use with `regexec' to match against a string. The compiled regular expression format is designed for efficient matching. `regcomp' stores it into `*COMPILED'. It's up to you to allocate an object of type `regex_t' and pass its address to `regcomp'. The argument CFLAGS lets you specify various options that control the syntax and semantics of regular expressions. *Note Flags for POSIX Regexps::. If you use the flag `REG_NOSUB', then `regcomp' omits from the compiled regular expression the information necessary to record how subexpressions actually match. In this case, you might as well pass `0' for the MATCHPTR and NMATCH arguments when you call `regexec'. If you don't use `REG_NOSUB', then the compiled regular expression does have the capacity to record how subexpressions match. Also, `regcomp' tells you how many subexpressions PATTERN has, by storing the number in `COMPILED->re_nsub'. You can use that value to decide how long an array to allocate to hold information about subexpression matches. `regcomp' returns `0' if it succeeds in compiling the regular expression; otherwise, it returns a nonzero error code (see the table below). You can use `regerror' to produce an error message string describing the reason for a nonzero value; see *Note Regexp Cleanup::. Here are the possible nonzero values that `regcomp' can return: `REG_BADBR' There was an invalid `\{...\}' construct in the regular expression. A valid `\{...\}' construct must contain either a single number, or two numbers in increasing order separated by a comma. `REG_BADPAT' There was a syntax error in the regular expression. `REG_BADRPT' A repetition operator such as `?' or `*' appeared in a bad position (with no preceding subexpression to act on). `REG_ECOLLATE' The regular expression referred to an invalid collating element (one not defined in the current locale for string collation). `REG_ECTYPE' The regular expression referred to an invalid character class name. `REG_EESCAPE' The regular expression ended with `\'. `REG_ESUBREG' There was an invalid number in the `\DIGIT' construct. `REG_EBRACK' There were unbalanced square brackets in the regular expression. `REG_EPAREN' An extended regular expression had unbalanced parentheses, or a basic regular expression had unbalanced `\(' and `\)'. `REG_EBRACE' The regular expression had unbalanced `\{' and `\}'. `REG_ERANGE' One of the endpoints in a range expression was invalid. `REG_ESPACE' `regcomp' ran out of memory.  File: rx.info, Node: Flags for POSIX Regexps, Next: Matching POSIX Regexps, Prev: POSIX Regexp Compilation, Up: Posix Entry Points Flags for POSIX Regular Expressions ----------------------------------- These are the bit flags that you can use in the CFLAGS operand when compiling a regular expression with `regcomp'. `REG_EXTENDED' Treat the pattern as an extended regular expression, rather than as a basic regular expression. `REG_ICASE' Ignore case when matching letters. `REG_NOSUB' Don't bother storing the contents of the MATCHES-PTR array. `REG_NEWLINE' Treat a newline in STRING as dividing STRING into multiple lines, so that `$' can match before the newline and `^' can match after. Also, don't permit `.' to match a newline, and don't permit `[^...]' to match a newline. Otherwise, newline acts like any other ordinary character.  File: rx.info, Node: Matching POSIX Regexps, Next: Regexp Subexpressions, Prev: Flags for POSIX Regexps, Up: Posix Entry Points Matching a Compiled POSIX Regular Expression -------------------------------------------- Once you have compiled a regular expression, as described in *Note POSIX Regexp Compilation::, you can match it against strings using `regexec'. A match anywhere inside the string counts as success, unless the regular expression contains anchor characters (`^' or `$'). - Function: int regexec (regex_t *COMPILED, char *STRING, size_t NMATCH, regmatch_t MATCHPTR [], int EFLAGS) This function tries to match the compiled regular expression `*COMPILED' against STRING. `regexec' returns `0' if the regular expression matches; otherwise, it returns a nonzero value. See the table below for what nonzero values mean. You can use `regerror' to produce an error message string describing the reason for a nonzero value; see *Note Regexp Cleanup::. The argument EFLAGS is a word of bit flags that enable various options. If you want to get information about what part of STRING actually matched the regular expression or its subexpressions, use the arguments MATCHPTR and NMATCH. Otherwise, pass `0' for NMATCH, and `NULL' for MATCHPTR. *Note Regexp Subexpressions::. You must match the regular expression with the same set of current locales that were in effect when you compiled the regular expression. The function `regexec' accepts the following flags in the EFLAGS argument: `REG_NOTBOL' Do not regard the beginning of the specified string as the beginning of a line; more generally, don't make any assumptions about what text might precede it. `REG_NOTEOL' Do not regard the end of the specified string as the end of a line; more generally, don't make any assumptions about what text might follow it. Here are the possible nonzero values that `regexec' can return: `REG_NOMATCH' The pattern didn't match the string. This isn't really an error. `REG_ESPACE' `regexec' ran out of memory.  File: rx.info, Node: Regexp Subexpressions, Next: Subexpression Complications, Prev: Matching POSIX Regexps, Up: Posix Entry Points Match Results with Subexpressions --------------------------------- When `regexec' matches parenthetical subexpressions of PATTERN, it records which parts of STRING they match. It returns that information by storing the offsets into an array whose elements are structures of type `regmatch_t'. The first element of the array (index `0') records the part of the string that matched the entire regular expression. Each other element of the array records the beginning and end of the part that matched a single parenthetical subexpression. - Data Type: regmatch_t This is the data type of the MATCHARRAY array that you pass to `regexec'. It containes two structure fields, as follows: `rm_so' The offset in STRING of the beginning of a substring. Add this value to STRING to get the address of that part. `rm_eo' The offset in STRING of the end of the substring. - Data Type: regoff_t `regoff_t' is an alias for another signed integer type. The fields of `regmatch_t' have type `regoff_t'. The `regmatch_t' elements correspond to subexpressions positionally; the first element (index `1') records where the first subexpression matched, the second element records the second subexpression, and so on. The order of the subexpressions is the order in which they begin. When you call `regexec', you specify how long the MATCHPTR array is, with the NMATCH argument. This tells `regexec' how many elements to store. If the actual regular expression has more than NMATCH subexpressions, then you won't get offset information about the rest of them. But this doesn't alter whether the pattern matches a particular string or not. If you don't want `regexec' to return any information about where the subexpressions matched, you can either supply `0' for NMATCH, or use the flag `REG_NOSUB' when you compile the pattern with `regcomp'.  File: rx.info, Node: Subexpression Complications, Next: Regexp Cleanup, Prev: Regexp Subexpressions, Up: Posix Entry Points Complications in Subexpression Matching --------------------------------------- Sometimes a subexpression matches a substring of no characters. This happens when `f\(o*\)' matches the string `fum'. (It really matches just the `f'.) In this case, both of the offsets identify the point in the string where the null substring was found. In this example, the offsets are both `1'. Sometimes the entire regular expression can match without using some of its subexpressions at all--for example, when `ba\(na\)*' matches the string `ba', the parenthetical subexpression is not used. When this happens, `regexec' stores `-1' in both fields of the element for that subexpression. Sometimes matching the entire regular expression can match a particular subexpression more than once--for example, when `ba\(na\)*' matches the string `bananana', the parenthetical subexpression matches three times. When this happens, `regexec' usually stores the offsets of the last part of the string that matched the subexpression. In the case of `bananana', these offsets are `6' and `8'. But the last match is not always the one that is chosen. It's more accurate to say that the last *opportunity* to match is the one that takes precedence. What this means is that when one subexpression appears within another, then the results reported for the inner subexpression reflect whatever happened on the last match of the outer subexpression. For an example, consider `\(ba\(na\)*s \)*' matching the string `bananas bas '. The last time the inner expression actually matches is near the end of the first word. But it is *considered* again in the second word, and fails to match there. `regexec' reports nonuse of the "na" subexpression. Another place where this rule applies is when the regular expression `\(ba\(na\)*s \|nefer\(ti\)* \)*' matches `bananas nefertiti'. The "na" subexpression does match in the first word, but it doesn't match in the second word because the other alternative is used there. Once again, the second repetition of the outer subexpression overrides the first, and within that second repetition, the "na" subexpression is not used. So `regexec' reports nonuse of the "na" subexpression.  File: rx.info, Node: Regexp Cleanup, Prev: Subexpression Complications, Up: Posix Entry Points POSIX Regexp Matching Cleanup ----------------------------- When you are finished using a compiled regular expression, you can free the storage it uses by calling `regfree'. - Function: void regfree (regex_t *COMPILED) Calling `regfree' frees all the storage that `*COMPILED' points to. This includes various internal fields of the `regex_t' structure that aren't documented in this manual. `regfree' does not free the object `*COMPILED' itself. You should always free the space in a `regex_t' structure with `regfree' before using the structure to compile another regular expression. When `regcomp' or `regexec' reports an error, you can use the function `regerror' to turn it into an error message string. - Function: size_t regerror (int ERRCODE, regex_t *COMPILED, char *BUFFER, size_t LENGTH) This function produces an error message string for the error code ERRCODE, and stores the string in LENGTH bytes of memory starting at BUFFER. For the COMPILED argument, supply the same compiled regular expression structure that `regcomp' or `regexec' was working with when it got the error. Alternatively, you can supply `NULL' for COMPILED; you will still get a meaningful error message, but it might not be as detailed. If the error message can't fit in LENGTH bytes (including a terminating null character), then `regerror' truncates it. The string that `regerror' stores is always null-terminated even if it has been truncated. The return value of `regerror' is the minimum length needed to store the entire error message. If this is less than LENGTH, then the error message was not truncated, and you can use it. Otherwise, you should call `regerror' again with a larger buffer. Here is a function which uses `regerror', but always dynamically allocates a buffer for the error message: char *get_regerror (int errcode, regex_t *compiled) { size_t length = regerror (errcode, compiled, NULL, 0); char *buffer = xmalloc (length); (void) regerror (errcode, compiled, buffer, length); return buffer; }  File: rx.info, Node: Performance Considerations, Next: Semantic Tests, Prev: Posix Entry Points, Up: Top Performance Considerations ========================== In order to use the Rx implementation of Posix regexp functions, you should include `' in your program. In order to use the Rx implementation of Posix regexp functions most effectively, it may help to know a bit about performance tuning in Rx. * Menu: * Avoiding Redundant Compilations :: Compiling Regexps is costly. * Controlling Memory Use:: You can limit Rx's memory use.  File: rx.info, Node: Avoiding Redundant Compilations, Next: Controlling Memory Use, Prev: Performance Considerations, Up: Performance Considerations Avoiding Redundant Compilations ------------------------------- Rx implements the Posix regexp entry points `regcomp', `regerror', `regexec', and `regfree'. Some special considerations apply when using the Rx versions. First, `regcomp' is a comparatively expensive operation. Addiitonally, `regexec' tends to perform better when a compiled expression is compiled once and used twice than when the pattern is compiled twice and used twice. Internally, Rx caches that information about a pattern which is constructed dynamicly by regexec. You can save on compilation costs and maximize the effectiveness of the Rx cache by re-using compiled expressions when it is convenient. For example, for reasons unrelated to Rx, it has long been the practice in GNU emacs to always save one compiled regular expression (the most recent). Before compiling a new expression, Emacs checks to see if the new pattern is the same as the one that is already compiled. This is the sort of optimization that Rx likes. (It may even win, in some cases, to cache more than a single compiled regexp.) In the long-run, Rx should be modified to cache compilations on its own. (*Note Improvements to Make::.) Sometimes programmers write programs with many regexps, all known staticly at compile time. This can be a problem with Rx because when the program starts up, or as it runs, all of those staticly known regexps have to be compiled, which may be too time consuming. One easy fix for programs with many static regexps is to use `flex' or another lexical-analyzer generator program instead. Lexer-generators are optimized for the case of compiling staticly known regexps. There are still reasons why staticly known regexps may not be convertable to `flex' input - for example, the Posix pattern language is more general than `flex''s. In the long-run, it may be a good idea to extend Rx to support static compilation of regexps. (*Note Improvements to Make::.)  File: rx.info, Node: Controlling Memory Use, Prev: Avoiding Redundant Compilations, Up: Performance Considerations Controlling Memory Use ---------------------- The size of the cache used by regexec is subject to control by programmers. Additionally, by using lower level entry points, programmers can create multiple, distinct Rx caches. This part of the code is subject to change and so is not yet documented. See "default_cache" in rxsuper.c and the parameters at the top of "rxbasic.c" to find some parameters you can tune or hints about creating alternative regexec caches.  File: rx.info, Node: Semantic Tests, Next: General Functions, Prev: Performance Considerations, Up: Top Semantic Tests ============== The distribution of *Rx* includes a test-suite, aimed at testing the conformance of an implementation of the Posix regexp functions to the standard. Currently, the tests come from two sources: a test-suite started by Henry Spencer, and the "Khadafy" test suite by Scott Anderson. The tests come in three files: * runtests.c - The driver program. * TESTS - The list of test cases * TESTS2C.sed - A script to convert test cases into C. * Menu: * Running the Tests:: Invoking the test suite. * Adding New Tests:: Extending the test suite.  File: rx.info, Node: Running the Tests, Next: Adding New Tests, Prev: Semantic Tests, Up: Semantic Tests Running the Tests ----------------- To build the test program, use `make runtests'. (Detailed configuration and build instructions can be found in INSTALL). Invoked with no arguments, `runtests' runs all test-cases. For each test case, a sequence number is printed. If there is a problem with that case, more information is printed. Output like this: #0 #1 #2 ... indicates a successful run. (The output format is likely to change in the future to make it easier to automate testing of Rx.) With a single numeric argument, runtests executes just that test and no others. This is handy when debugging. Sometimes a bug will occur when all tests are run, but disappear when just the problematic test is run. Usually this has to do with memory or cache corruption.  File: rx.info, Node: Adding New Tests, Prev: Running the Tests, Up: Semantic Tests Adding New Tests ---------------- This list of tests used by `runtests' is found in the file `TESTS'. Each line of that file is a list of colon separated fields similar to: 1:^(ab|cd)e:abcde The first field is the expected return value of regexec, or '2' meaning that the pattern is not valid. The second field is the regular expression being tested. The third field is the string to which the pattern is to be compared. The file `TESTS2C.sed' is used to convert the test cases into initializers for an array of structures (which is automated in the Makefile). To add a new test case, add lines to `TESTS' and recompile `runtests'.  File: rx.info, Node: General Functions, Next: Rx Theory, Prev: Semantic Tests, Up: Top General Functions ================= Here is a whirlwind tour of the lower-level Rx entry points and data structures. This is not intended to be complete, but rather to be a help to anyone reading the code itself. These are presented in the same order in which they might typically be used. * Menu: * Compiling Expressions:: Strings->Syntax Trees * Posixly Correct Matches:: Checking for a Posixly correct match. * Hairy Matches:: Matching with multi-part strings. * NFA Functions:: Forget Posix, eat NFAs raw.  File: rx.info, Node: Compiling Expressions, Next: Posixly Correct Matches, Prev: General Functions, Up: General Functions Compiling Expressions --------------------- - Internal Function: reg_errcode_t rx_parse (struct rexp_node ** rexp_p, __const__ char *pattern, int size, unsigned long syntax, int cset_size, unsigned char *translate); Compile a regular expression. REXP_P will return a syntax tree for the pattern. See `"rxnode.h"' to learn about syntax trees. Minimally, you should know that syntax trees are reference counted using `rx_save_rexp' and `rx_free_rexp'. PATTERN and SIZE are the expression and its length. The expression need not be 0-terminated. The bits set in SYNTAX control details of the language understood by the parser. The meaning of each bit is described in `"rxgnucomp.h"'. CSET_SIZE is usually 256. It should not be much larger than that or space-performance will suffer badly. TRANSLATE is a translation table - an array of characters. Characters in the pattern are looked up in TRANSLATE as they are read. Additionally, the pattern is compiled presuming that the same translation table will be applied to characters in the string being searched (meaning that the pattern is compiled in such a way that the effect will be as if every character in the target string is looked up in the translation table, even though the expense of that lookup is usually avoided). A return of 0 indicates success, otherwise, the error can be looked up in the `rx_error_msg' array. - Internal Function: int rx_posix_analyze_rexp (struct rexp_node *** subexps, int * n_subexps, struct rexp_node * node, int id) Staticly analyze a regular expression in preparation for matching. This function fills in some fields of the nodes of a syntax tree. `subexps' and `n_subexps' return a malloced array of pointers into the syntax tree, one per parenthesized subexpression. The caller must eventually free this array. `node' is the pattern to be analyzed. The tree returned by `rx_parse' is suitable for this. `id' should be 1. Ignore the return value. This function is not robust in the case that `malloc' or `realloc' returns 0 (which should be fixed). You might expect that `rx_posix_analyze_rexp' and `rx_parse' should be combined. They are not because the actions of `rx_posix_analyze_rexp' make sense for any syntax tree, regardless of how it is constructed. `rx_parse' is just one way to build a syntax tree - it is subject to replacement and trees can be built without using a parser at all (c.f. `"rxnode.h"').  File: rx.info, Node: Posixly Correct Matches, Next: Hairy Matches, Prev: Compiling Expressions, Up: General Functions Checking for Posixly Correct Matches with One-Piece Strings ----------------------------------------------------------- Looking for a match that is Posixly correct is conceptually tricky. Here is one way to think of it: The Posix expression languages, absent of any consideration of the meaning of "leftmost-longest", underspecify the return value of `regexec'. Consider the example: Pattern: foo\(.*\)\(.*\)bar Target string: fooxxxbar Initially it is ambiguous how this should be matched. The ambiguity is exposed by considering the possible bindings for the subexpressions `\1' and `\2': Possible outcomes: \1 == "xxx" \2 == "" \1 == "xx" \2 == "x" \1 == "x" \2 == "xx" \1 == "" \2 == "xxx" Posix tells us that the first outcome is the prefered one if the illustrated pattern is used alone (this is a consequence of the "leftmost longest" rule). But Posix also says the answer may be different if the pattern is a sub-pattern of a larger pattern. For example: Pattern: foo\(.*\)\(.*\)bar\2 Target string: fooxxxbarxx Now the only acceptable outcome is: \1 == "x" \2 == "xx" Three entry points provide a relatively simple interface to this moderately complicated situation: - Internal Function: struct rx_solutions * rx_basic_make_solutions (struct rx_registers * regs, struct rexp_node * expression, struct rexp_node ** subexps, int start, int end, struct rx_context_rules * rules, char * str) - Internal Function: void rx_basic_free_solutions (struct rx_solutions * solns); - Internal Function: enum rx_answers rx_next_solution (struct rx_solutions * solns) - Internal Function: void rx_basic_free_solutions (struct rx_solutions * solns) `make_solutions' returns a lazilly constructed stream of possible solutions for a given regexp and target string. `next_solution' constructs and returns the next solution from a list returned by `make_solutions'. Solutions are returned in order of preferability according to the Posix spec. So, the first solution is the leftmost-longest, and the last, the rightmost-shortest. REGS is where subexpression extents are tracked. EXPRESSION is the expression to solve. SUBEXPS is as returned by `rx_posix_analyze_rexp'. START and END are the integer addresses of the data to be compared to the pattern. A solution must match all characters starting at START and ending at `END - 1'. RULES are some bits that control the precise meaning of "^" and "$". See `"rxcontext.h"'. STR is the data to be matched. Note that anchoring operators can cause the matcher to look beyond START and END when testing a match. It may make sense to pass START > 0, for example. `make_solutions' returns a pointer to a solution list, or 0 if an allocation fails. `next_solution' returns a status which can be: `rx_yes' -- The pattern matches exactly. `rx_maybe' -- The pattern does not match, but might if more characters were added to the string. `rx_no' -- The pattern does not match and could not be made to match even by adding more characters. In addition, REGS is filled in by `next_solution'. Finally, `free_solutions' is used to release resources consumed by a solution list.  File: rx.info, Node: Hairy Matches, Next: NFA Functions, Prev: Posixly Correct Matches, Up: General Functions Hairy Matches and Suspendable Searches -------------------------------------- The functions `rx_basic_make_solutions' and `rx_basic_free_solutions' are used to find matches in a one-piece string. More general entry points are provided for multi-piece strings, which may or may not be entirely available in memory at one time: - Internal Function: struct rx_solutions * rx_make_solutions (struct rx_registers * regs, struct rx_unfaniverse * verse, struct rexp_node * expression, struct rexp_node ** subexps, int cset_size, int start, int end, rx_vmfn vmfn, rx_contextfn contextfn, void * closure) - Internal Function: void rx_free_solutions (struct rx_solutions * solns) These functions are similar to their `_basic_' analogues, but substitute some new parameters for the one piece string: VMFN is a function provided by the caller that returns at least part of the string being matched. CONTEXTFN is a function provided by the caller that does the work of the backreference operators, "^", and "$". CLOSURE is data provided by the caller, passed through to vmfn and contextfn. typedef enum rx_answers (*rx_vmfn) P((void * closure, unsigned char ** burst, int * len, int * offset, int start, int end, int need)); typedef enum rx_answers (*rx_contextfn) P((void * closure, int type, int start, int end, struct rx_registers * regs)); `vmfn' is asked for a range of characters from START to END. NEED is in that range and is the only position that must be returned by vmfn. BURST, LEN, and OFFSET are output parameters. OFFSET will get the integer address of burst, and LEN the number of valid characters in the burst. VMFN may be asked for the empty string from END to END and should be able to handle that case. If `vmfn' provides the needed character, it should return `rx_yes'. If `vmfn' can't provide the character now, but might be able to later, it should return `rx_maybe'. This will cause `rx_next_solution' to quickly return `rx_maybe'. If `rx_next_solution' is called again, it will retry the call to vmfn and possibly resume the search. If `vmfn' can not provide the needed character, it should return `rx_no'. `rx_next_solution' will not normally ask for out-of-range characters, so usually there is no good reason to return `rx_no', but the code is supposed to be robust in case you do. The `contextfn' also returns a `yes/no/maybe' status. It tests whether the characters in START to END satisfy the operator TYPE. `type' is an ascii value; one of the characters in "^$123456789".  File: rx.info, Node: NFA Functions, Prev: Hairy Matches, Up: General Functions NFA Functions and Super-NFA Functions ------------------------------------- Not really documented yet. rx.h, rxnfa.h -- translating an expression into an NFA. rxsuper.h -- optimizing the NFA into a DFA, or at least an NFA with fewer branch points. rxanal.h -- how to use DFAs to find longest matches, matches of a specific length, the shortest match, etc. rxsimp.h -- Translating a Posix "regular expression" into a true regular expression that describes a superset language.  File: rx.info, Node: Rx Theory, Next: Improvements to Make, Prev: General Functions, Up: Top Rx Theory ========= There are two match algorithms. One is for truly regular regexps (those that can be reduced to a dfa). The other is for non-regular regexps. The dfa algorithm implements the idea suggested in `Compilers' by Aho, Sethi and Ullman: [One] approach [to pattern matching regular expressions] is to use a DFA, but avoid constructing all of the transition table by using a technique called "lazy transition evaluation". Here, transitions are computed at run time [when] actually needed. [T]ransitions are stored in a cache. [....] If the cache becomes full, we can erase some previously computed transition to make room for the new transition. The implementation in Rx is generalized from that, but the above description covers what is used for Posix patterns. The non-dfa algorithm implements a "recursive decomposition" technique described in email by Henry Spencer. For a given pattern, this algorithm first checks to see if a simpler, superset language, DFA-pattern matches. If it does, then this algorithm does the detail-work to see if the non-DFA pattern matches. The detail work usually involves recursing on subpatterns. For example, a concatentation of two subexpressions matches a string if the string can be divided into two parts, each matching one subexpression, in the right order. More than one solution is often possible for a given pattern. This ambiguity is the subject of the "leftmost longest" rules in the spec, and the back-tracking oriented stream-of-solution functions `rx_make_solutions', `rx_next_solution' and `rx_free_solutions'. rxspencer.[ch] -- The non-DFA algorithm rxanal.[ch] rxsuper.[ch] rxnfa.[ch] -- The DFA algorithm  File: rx.info, Node: Improvements to Make, Prev: Rx Theory, Up: Top Improvements to Make ==================== An approximation of the GNU entry points is needed. This can not easily be an exact approximation because the GNU entry points have no function for freeing a compiled expression - they assume that compiled expressions can be freed using a single call to "free(3)". For that reason, do not use the names of the old GNU entry points ...use "rx_" instead. This version of Rx hasn't been profiled and tuned. It might be nice to have some C++ types to hide reference counting and provide handy constructors for rexp_nodes and nfa parts. Right now, the regexp->regular regexp mapping in rxsimp.[ch] is cached for a particular input expression. I think it would probably be a big win in many situations if it were cached, like unique nfas (c.f. rxunfa.[ch]), for equivalent expressions. It would speed up matching to lift the non-regular constructs to the top of the syntax tree. e.g., if you have: pattern: "abc$" syntax tree: (concat (cset "a") (concat (cset "b") (concat (cset "c") (operator "$")))) it involves a lot more work (in "next_solution) than if you have the equivalent: same pattern: "abc$" syntax tree: (concat (concat (cset "a") (concat (cset "b") (cset "c"))) (operator "$")) in fact, while doing that, notice that subtrees of nothing-but-concat can conceivably be optimized so that next_solution just uses "strcmp". For that matter, a tree which is a disjunction of conjunction (a tree of "|" whose leaves are trees of concat nodes) could be optimized to use a table method (Boyer/Moore?). Again, this would be for next_solution but might involve adding stuff to rx_posix_analyze_rexp. As more and more optimizations are added, programmers should be given simple control over which optimizations are important for a particular pattern. For example, packages in emacs often include a few regexps that are used heavily - it would be worth the trouble to arrange for those regexps to be compiled specially. Static compilation of regexps is mostly a job for lex, but posix patterns are more general and have a nicer interface. What's more, it'd be nice to have the benefits of static compilation even for dynamicly loaded regexps (e.g., imagine if when an emacs lisp file is byte-compiled, the regexps it contains are also compiled). A lot of optimizations apply to just syntax trees. One approach to static compilation is to make an ugly but fast-reading syntax for trees, and a printer for that syntax. Then, after optimiztaions, a pattern could be stored as a string and later read back in, already optimized. A similar trick could be done for the nfa built in rxnfa.c. next_solution already considers the length of a fixed-length subexpression, but many more lenght optimizations are possible. I'd guess these will pay off well.  Tag Table: Node: Top83 Node: Posix Entry Points638 Node: POSIX Regexp Compilation1546 Node: Flags for POSIX Regexps5626 Node: Matching POSIX Regexps6530 Node: Regexp Subexpressions8688 Node: Subexpression Complications10740 Node: Regexp Cleanup13096 Node: Performance Considerations15419 Node: Avoiding Redundant Compilations15989 Node: Controlling Memory Use18120 Node: Semantic Tests18719 Node: Running the Tests19445 Node: Adding New Tests20369 Node: General Functions21121 Node: Compiling Expressions21775 Node: Posixly Correct Matches24531 Node: Hairy Matches28136 Node: NFA Functions31078 Node: Rx Theory31742 Node: Improvements to Make33591  End Tag Table dist/doc/rx.texi100660 765 144 111765 6071535072 12535 0ustar lordusers\input texinfo @c -*-texinfo-*- @c %**start of header @setfilename rx.info @settitle Rx @setchapternewpage odd @c %**end of header @c This title page illustrates only one of the @c two methods of forming a title page. @titlepage @title Rx @subtitle A Dynamic Automata Library @author Tom Lord @center with excerpts from @emph{The GNU C Library} reference manual @center by Sandra Loosemore @center with @center Richard M. Stallman, Roland McGrath, and Andrew Oram @c The following two commands @c start the copyright page. @page @vskip 0pt plus 1filll Copyright @copyright{} 1995 Cygnus Support Permission is granted to make and distribute verbatim copies of this manual provided the copyright notice and this permission notice are preserved on all copies. Permission is granted to copy and distribute modified versions of this manual under the conditions for verbatim copying, provided that the entire resulting derived work is distributed under the terms of a permission notice identical to this one. Permission is granted to copy and distribute translations of this manual into another language, under the above conditions for modified versions, except that this permission notice may be stated in a translation approved by the author. @end titlepage @node Top, Posix Entry Points, (dir), (dir) @top Rx This document describes Rx. This document applies to version "Rx-diSpencer" of the library named librx.a @menu * Posix Entry Points:: The normal way to call regexp functions. * Performance Considerations:: Special considerations when using Rx. * Semantic Tests:: A test-suite for Posix matcher semantics. * General Functions:: Rx's native entry points. * Rx Theory:: A quick tour of the implementation. * Improvements to Make:: The Rx TODO list. @end menu @node Posix Entry Points, Performance Considerations, Top, Top @section Regular Expression Matching This section is excerpted from @emph{The GNU C Library} reference manual by Sandra Loosemore with Richard M. Stallman, Roland McGrath, and Andrew Oram. The GNU C library supports the standard POSIX.2 interface. Programs using this interface should include the header file @file{rxposix.h}. @menu * POSIX Regexp Compilation:: Using @code{regcomp} to prepare to match. * Flags for POSIX Regexps:: Syntax variations for @code{regcomp}. * Matching POSIX Regexps:: Using @code{regexec} to match the compiled pattern that you get from @code{regcomp}. * Regexp Subexpressions:: Finding which parts of the string were matched. * Subexpression Complications:: Find points of which parts were matched. * Regexp Cleanup:: Freeing storage; reporting errors. @end menu @node POSIX Regexp Compilation @subsection POSIX Regular Expression Compilation Before you can actually match a regular expression, you must @dfn{compile} it. This is not true compilation---it produces a special data structure, not machine instructions. But it is like ordinary compilation in that its purpose is to enable you to ``execute'' the pattern fast. (@xref{Matching POSIX Regexps}, for how to use the compiled regular expression for matching.) There is a special data type for compiled regular expressions: @comment regex.h @comment POSIX.2 @deftp {Data Type} regex_t This type of object holds a compiled regular expression. It is actually a structure. It has just one field that your programs should look at: @table @code @item re_nsub This field holds the number of parenthetical subexpressions in the regular expression that was compiled. @end table There are several other fields, but we don't describe them here, because only the functions in the library should use them. @end deftp After you create a @code{regex_t} object, you can compile a regular expression into it by calling @code{regcomp}. @comment regex.h @comment POSIX.2 @deftypefun int regcomp (regex_t *@var{compiled}, const char *@var{pattern}, int @var{cflags}) The function @code{regcomp} ``compiles'' a regular expression into a data structure that you can use with @code{regexec} to match against a string. The compiled regular expression format is designed for efficient matching. @code{regcomp} stores it into @code{*@var{compiled}}. It's up to you to allocate an object of type @code{regex_t} and pass its address to @code{regcomp}. The argument @var{cflags} lets you specify various options that control the syntax and semantics of regular expressions. @xref{Flags for POSIX Regexps}. If you use the flag @code{REG_NOSUB}, then @code{regcomp} omits from the compiled regular expression the information necessary to record how subexpressions actually match. In this case, you might as well pass @code{0} for the @var{matchptr} and @var{nmatch} arguments when you call @code{regexec}. If you don't use @code{REG_NOSUB}, then the compiled regular expression does have the capacity to record how subexpressions match. Also, @code{regcomp} tells you how many subexpressions @var{pattern} has, by storing the number in @code{@var{compiled}->re_nsub}. You can use that value to decide how long an array to allocate to hold information about subexpression matches. @code{regcomp} returns @code{0} if it succeeds in compiling the regular expression; otherwise, it returns a nonzero error code (see the table below). You can use @code{regerror} to produce an error message string describing the reason for a nonzero value; see @ref{Regexp Cleanup}. @end deftypefun Here are the possible nonzero values that @code{regcomp} can return: @table @code @comment regex.h @comment POSIX.2 @item REG_BADBR There was an invalid @samp{\@{@dots{}\@}} construct in the regular expression. A valid @samp{\@{@dots{}\@}} construct must contain either a single number, or two numbers in increasing order separated by a comma. @comment regex.h @comment POSIX.2 @item REG_BADPAT There was a syntax error in the regular expression. @comment regex.h @comment POSIX.2 @item REG_BADRPT A repetition operator such as @samp{?} or @samp{*} appeared in a bad position (with no preceding subexpression to act on). @comment regex.h @comment POSIX.2 @item REG_ECOLLATE The regular expression referred to an invalid collating element (one not defined in the current locale for string collation). @comment regex.h @comment POSIX.2 @item REG_ECTYPE The regular expression referred to an invalid character class name. @comment regex.h @comment POSIX.2 @item REG_EESCAPE The regular expression ended with @samp{\}. @comment regex.h @comment POSIX.2 @item REG_ESUBREG There was an invalid number in the @samp{\@var{digit}} construct. @comment regex.h @comment POSIX.2 @item REG_EBRACK There were unbalanced square brackets in the regular expression. @comment regex.h @comment POSIX.2 @item REG_EPAREN An extended regular expression had unbalanced parentheses, or a basic regular expression had unbalanced @samp{\(} and @samp{\)}. @comment regex.h @comment POSIX.2 @item REG_EBRACE The regular expression had unbalanced @samp{\@{} and @samp{\@}}. @comment regex.h @comment POSIX.2 @item REG_ERANGE One of the endpoints in a range expression was invalid. @comment regex.h @comment POSIX.2 @item REG_ESPACE @code{regcomp} ran out of memory. @end table @node Flags for POSIX Regexps @subsection Flags for POSIX Regular Expressions These are the bit flags that you can use in the @var{cflags} operand when compiling a regular expression with @code{regcomp}. @table @code @comment regex.h @comment POSIX.2 @item REG_EXTENDED Treat the pattern as an extended regular expression, rather than as a basic regular expression. @comment regex.h @comment POSIX.2 @item REG_ICASE Ignore case when matching letters. @comment regex.h @comment POSIX.2 @item REG_NOSUB Don't bother storing the contents of the @var{matches-ptr} array. @comment regex.h @comment POSIX.2 @item REG_NEWLINE Treat a newline in @var{string} as dividing @var{string} into multiple lines, so that @samp{$} can match before the newline and @samp{^} can match after. Also, don't permit @samp{.} to match a newline, and don't permit @samp{[^@dots{}]} to match a newline. Otherwise, newline acts like any other ordinary character. @end table @node Matching POSIX Regexps @subsection Matching a Compiled POSIX Regular Expression Once you have compiled a regular expression, as described in @ref{POSIX Regexp Compilation}, you can match it against strings using @code{regexec}. A match anywhere inside the string counts as success, unless the regular expression contains anchor characters (@samp{^} or @samp{$}). @comment regex.h @comment POSIX.2 @deftypefun int regexec (regex_t *@var{compiled}, char *@var{string}, size_t @var{nmatch}, regmatch_t @var{matchptr} @t{[]}, int @var{eflags}) This function tries to match the compiled regular expression @code{*@var{compiled}} against @var{string}. @code{regexec} returns @code{0} if the regular expression matches; otherwise, it returns a nonzero value. See the table below for what nonzero values mean. You can use @code{regerror} to produce an error message string describing the reason for a nonzero value; see @ref{Regexp Cleanup}. The argument @var{eflags} is a word of bit flags that enable various options. If you want to get information about what part of @var{string} actually matched the regular expression or its subexpressions, use the arguments @var{matchptr} and @var{nmatch}. Otherwise, pass @code{0} for @var{nmatch}, and @code{NULL} for @var{matchptr}. @xref{Regexp Subexpressions}. @end deftypefun You must match the regular expression with the same set of current locales that were in effect when you compiled the regular expression. The function @code{regexec} accepts the following flags in the @var{eflags} argument: @table @code @comment regex.h @comment POSIX.2 @item REG_NOTBOL Do not regard the beginning of the specified string as the beginning of a line; more generally, don't make any assumptions about what text might precede it. @comment regex.h @comment POSIX.2 @item REG_NOTEOL Do not regard the end of the specified string as the end of a line; more generally, don't make any assumptions about what text might follow it. @end table Here are the possible nonzero values that @code{regexec} can return: @table @code @comment regex.h @comment POSIX.2 @item REG_NOMATCH The pattern didn't match the string. This isn't really an error. @comment regex.h @comment POSIX.2 @item REG_ESPACE @code{regexec} ran out of memory. @end table @node Regexp Subexpressions @subsection Match Results with Subexpressions When @code{regexec} matches parenthetical subexpressions of @var{pattern}, it records which parts of @var{string} they match. It returns that information by storing the offsets into an array whose elements are structures of type @code{regmatch_t}. The first element of the array (index @code{0}) records the part of the string that matched the entire regular expression. Each other element of the array records the beginning and end of the part that matched a single parenthetical subexpression. @comment regex.h @comment POSIX.2 @deftp {Data Type} regmatch_t This is the data type of the @var{matcharray} array that you pass to @code{regexec}. It containes two structure fields, as follows: @table @code @item rm_so The offset in @var{string} of the beginning of a substring. Add this value to @var{string} to get the address of that part. @item rm_eo The offset in @var{string} of the end of the substring. @end table @end deftp @comment regex.h @comment POSIX.2 @deftp {Data Type} regoff_t @code{regoff_t} is an alias for another signed integer type. The fields of @code{regmatch_t} have type @code{regoff_t}. @end deftp The @code{regmatch_t} elements correspond to subexpressions positionally; the first element (index @code{1}) records where the first subexpression matched, the second element records the second subexpression, and so on. The order of the subexpressions is the order in which they begin. When you call @code{regexec}, you specify how long the @var{matchptr} array is, with the @var{nmatch} argument. This tells @code{regexec} how many elements to store. If the actual regular expression has more than @var{nmatch} subexpressions, then you won't get offset information about the rest of them. But this doesn't alter whether the pattern matches a particular string or not. If you don't want @code{regexec} to return any information about where the subexpressions matched, you can either supply @code{0} for @var{nmatch}, or use the flag @code{REG_NOSUB} when you compile the pattern with @code{regcomp}. @node Subexpression Complications @subsection Complications in Subexpression Matching Sometimes a subexpression matches a substring of no characters. This happens when @samp{f\(o*\)} matches the string @samp{fum}. (It really matches just the @samp{f}.) In this case, both of the offsets identify the point in the string where the null substring was found. In this example, the offsets are both @code{1}. Sometimes the entire regular expression can match without using some of its subexpressions at all---for example, when @samp{ba\(na\)*} matches the string @samp{ba}, the parenthetical subexpression is not used. When this happens, @code{regexec} stores @code{-1} in both fields of the element for that subexpression. Sometimes matching the entire regular expression can match a particular subexpression more than once---for example, when @samp{ba\(na\)*} matches the string @samp{bananana}, the parenthetical subexpression matches three times. When this happens, @code{regexec} usually stores the offsets of the last part of the string that matched the subexpression. In the case of @samp{bananana}, these offsets are @code{6} and @code{8}. But the last match is not always the one that is chosen. It's more accurate to say that the last @emph{opportunity} to match is the one that takes precedence. What this means is that when one subexpression appears within another, then the results reported for the inner subexpression reflect whatever happened on the last match of the outer subexpression. For an example, consider @samp{\(ba\(na\)*s \)*} matching the string @samp{bananas bas }. The last time the inner expression actually matches is near the end of the first word. But it is @emph{considered} again in the second word, and fails to match there. @code{regexec} reports nonuse of the ``na'' subexpression. Another place where this rule applies is when the regular expression @w{@samp{\(ba\(na\)*s \|nefer\(ti\)* \)*}} matches @samp{bananas nefertiti}. The ``na'' subexpression does match in the first word, but it doesn't match in the second word because the other alternative is used there. Once again, the second repetition of the outer subexpression overrides the first, and within that second repetition, the ``na'' subexpression is not used. So @code{regexec} reports nonuse of the ``na'' subexpression. @node Regexp Cleanup @subsection POSIX Regexp Matching Cleanup When you are finished using a compiled regular expression, you can free the storage it uses by calling @code{regfree}. @comment regex.h @comment POSIX.2 @deftypefun void regfree (regex_t *@var{compiled}) Calling @code{regfree} frees all the storage that @code{*@var{compiled}} points to. This includes various internal fields of the @code{regex_t} structure that aren't documented in this manual. @code{regfree} does not free the object @code{*@var{compiled}} itself. @end deftypefun You should always free the space in a @code{regex_t} structure with @code{regfree} before using the structure to compile another regular expression. When @code{regcomp} or @code{regexec} reports an error, you can use the function @code{regerror} to turn it into an error message string. @comment regex.h @comment POSIX.2 @deftypefun size_t regerror (int @var{errcode}, regex_t *@var{compiled}, char *@var{buffer}, size_t @var{length}) This function produces an error message string for the error code @var{errcode}, and stores the string in @var{length} bytes of memory starting at @var{buffer}. For the @var{compiled} argument, supply the same compiled regular expression structure that @code{regcomp} or @code{regexec} was working with when it got the error. Alternatively, you can supply @code{NULL} for @var{compiled}; you will still get a meaningful error message, but it might not be as detailed. If the error message can't fit in @var{length} bytes (including a terminating null character), then @code{regerror} truncates it. The string that @code{regerror} stores is always null-terminated even if it has been truncated. The return value of @code{regerror} is the minimum length needed to store the entire error message. If this is less than @var{length}, then the error message was not truncated, and you can use it. Otherwise, you should call @code{regerror} again with a larger buffer. Here is a function which uses @code{regerror}, but always dynamically allocates a buffer for the error message: @smallexample char *get_regerror (int errcode, regex_t *compiled) @{ size_t length = regerror (errcode, compiled, NULL, 0); char *buffer = xmalloc (length); (void) regerror (errcode, compiled, buffer, length); return buffer; @} @end smallexample @end deftypefun @node Performance Considerations, Semantic Tests, Posix Entry Points, Top @section Performance Considerations @cindex Performance Considerations In order to use the Rx implementation of Posix regexp functions most effectively, it may help to know a bit about performance tuning in Rx. @menu * Avoiding Redundant Compilations :: Compiling Regexps is costly. * Controlling Memory Use:: You can limit Rx's memory use. @end menu @node Avoiding Redundant Compilations , Controlling Memory Use, Performance Considerations, Performance Considerations @subsection Avoiding Redundant Compilations Rx implements the Posix regexp entry points @code{regcomp}, @code{regerror}, @code{regexec}, and @code{regfree}. Some special considerations apply when using the Rx versions. First, @code{regcomp} is a comparatively expensive operation. Addiitonally, @code{regexec} tends to perform better when a compiled expression is compiled once and used twice than when the pattern is compiled twice and used twice. Internally, Rx caches that information about a pattern which is constructed dynamicly by regexec. You can save on compilation costs and maximize the effectiveness of the Rx cache by re-using compiled expressions when it is convenient. For example, for reasons unrelated to Rx, it has long been the practice in GNU emacs to always save one compiled regular expression (the most recent). Before compiling a new expression, Emacs checks to see if the new pattern is the same as the one that is already compiled. This is the sort of optimization that Rx likes. (It may even win, in some cases, to cache more than a single compiled regexp.) In the long-run, Rx should be modified to cache compilations on its own. (@xref{Improvements to Make}.) Sometimes programmers write programs with many regexps, all known staticly at compile time. This can be a problem with Rx because when the program starts up, or as it runs, all of those staticly known regexps have to be compiled, which may be too time consuming. One easy fix for programs with many static regexps is to use @code{flex} or another lexical-analyzer generator program instead. Lexer-generators are optimized for the case of compiling staticly known regexps. There are still reasons why staticly known regexps may not be convertable to @code{flex} input -- for example, the Posix pattern language is more general than @code{flex}'s. In the long-run, it may be a good idea to extend Rx to support static compilation of regexps. (@xref{Improvements to Make}.) @node Controlling Memory Use, , Avoiding Redundant Compilations , Performance Considerations @subsection Controlling Memory Use The size of the cache used by regexec is subject to control by programmers. Additionally, by using lower level entry points, programmers can create multiple, distinct Rx caches. This part of the code is subject to change and so is not yet documented. See "default_cache" in rxsuper.c and the parameters at the top of "rxbasic.c" to find some parameters you can tune or hints about creating alternative regexec caches. @node Semantic Tests, General Functions, Performance Considerations, Top @section Semantic Tests @cindex Semantic tests The distribution of @emph{Rx} includes a test-suite, aimed at testing the conformance of an implementation of the Posix regexp functions to the standard. Currently, the tests come from two sources: a test-suite started by Henry Spencer, and the "Khadafy" test suite by Scott Anderson. The tests come in three files: @itemize @bullet @item runtests.c -- The driver program. @item TESTS -- The list of test cases @item TESTS2C.sed -- A script to convert test cases into C. @end itemize @menu * Running the Tests:: Invoking the test suite. * Adding New Tests:: Extending the test suite. @end menu @node Running the Tests, Adding New Tests, Semantic Tests, Semantic Tests @subsection Running the Tests To build the test program, use @code{make runtests}. (Detailed configuration and build instructions can be found in INSTALL). Invoked with no arguments, @code{runtests} runs all test-cases. For each test case, a sequence number is printed. If there is a problem with that case, more information is printed. Output like this: @example #0 #1 #2 ... @end example indicates a successful run. (The output format is likely to change in the future to make it easier to automate testing of Rx.) With a single numeric argument, runtests executes just that test and no others. This is handy when debugging. Sometimes a bug will occur when all tests are run, but disappear when just the problematic test is run. Usually this has to do with memory or cache corruption. @node Adding New Tests, , Running the Tests, Semantic Tests @subsection Adding New Tests This list of tests used by @code{runtests} is found in the file @code{TESTS}. Each line of that file is a list of colon separated fields similar to: @example 1:^(ab|cd)e:abcde @end example The first field is the expected return value of regexec, or '2' meaning that the pattern is not valid. The second field is the regular expression being tested. The third field is the string to which the pattern is to be compared. The file @code{TESTS2C.sed} is used to convert the test cases into initializers for an array of structures (which is automated in the Makefile). To add a new test case, add lines to @code{TESTS} and recompile @code{runtests}. @node General Functions, Rx Theory, Semantic Tests, Top @section General Functions @cindex Generalized entry points Here is a whirlwind tour of the lower-level Rx entry points and data structures. This is not intended to be complete, but rather to be a help to anyone reading the code itself. These are presented in the same order in which they might typically be used. @menu * Compiling Expressions:: Strings->Syntax Trees * Posixly Correct Matches:: Checking for a Posixly correct match. * Hairy Matches:: Matching with multi-part strings. * NFA Functions:: Forget Posix, eat NFAs raw. @end menu @node Compiling Expressions, Posixly Correct Matches, General Functions, General Functions @subsection Compiling Expressions @deftypefn {Internal Function} reg_errcode_t rx_parse (struct rexp_node ** rexp_p, __const__ char *pattern, int size, unsigned long syntax, int cset_size, unsigned char *translate); Compile a regular expression. @var{rexp_p} will return a syntax tree for the pattern. See @code{"rxnode.h"} to learn about syntax trees. Minimally, you should know that syntax trees are reference counted using @code{rx_save_rexp} and @code{rx_free_rexp}. @var{pattern} and @var{size} are the expression and its length. The expression need not be 0-terminated. The bits set in @var{syntax} control details of the language understood by the parser. The meaning of each bit is described in @code{"rxgnucomp.h"}. @var{cset_size} is usually 256. It should not be much larger than that or space-performance will suffer badly. @var{translate} is a translation table -- an array of characters. Characters in the pattern are looked up in @var{translate} as they are read. Additionally, the pattern is compiled presuming that the same translation table will be applied to characters in the string being searched (meaning that the pattern is compiled in such a way that the effect will be as if every character in the target string is looked up in the translation table, even though the expense of that lookup is usually avoided). A return of 0 indicates success, otherwise, the error can be looked up in the @code{rx_error_msg} array. @end deftypefn @deftypefn {Internal Function} int rx_posix_analyze_rexp (struct rexp_node *** subexps, int * n_subexps, struct rexp_node * node, int id) Staticly analyze a regular expression in preparation for matching. This function fills in some fields of the nodes of a syntax tree. @code{subexps} and @code{n_subexps} return a malloced array of pointers into the syntax tree, one per parenthesized subexpression. The caller must eventually free this array. @code{node} is the pattern to be analyzed. The tree returned by @code{rx_parse} is suitable for this. @code{id} should be 1. Ignore the return value. This function is not robust in the case that @code{malloc} or @code{realloc} returns 0 (which should be fixed). @end deftypefn You might expect that @code{rx_posix_analyze_rexp} and @code{rx_parse} should be combined. They are not because the actions of @code{rx_posix_analyze_rexp} make sense for any syntax tree, regardless of how it is constructed. @code{rx_parse} is just one way to build a syntax tree -- it is subject to replacement and trees can be built without using a parser at all (c.f. @code{"rxnode.h"}). @node Posixly Correct Matches, Hairy Matches, Compiling Expressions, General Functions @subsection Checking for Posixly Correct Matches with One-Piece Strings Looking for a match that is Posixly correct is conceptually tricky. Here is one way to think of it: The Posix expression languages, absent of any consideration of the meaning of "leftmost-longest", underspecify the return value of @code{regexec}. Consider the example: @example Pattern: foo\(.*\)\(.*\)bar Target string: fooxxxbar @end example Initially it is ambiguous how this should be matched. The ambiguity is exposed by considering the possible bindings for the subexpressions @code{\1} and @code{\2}: @example Possible outcomes: \1 == "xxx" \2 == "" \1 == "xx" \2 == "x" \1 == "x" \2 == "xx" \1 == "" \2 == "xxx" @end example Posix tells us that the first outcome is the prefered one if the illustrated pattern is used alone (this is a consequence of the "leftmost longest" rule). But Posix also says the answer may be different if the pattern is a sub-pattern of a larger pattern. For example: @example Pattern: foo\(.*\)\(.*\)bar\2 Target string: fooxxxbarxx @end example Now the only acceptable outcome is: @example \1 == "x" \2 == "xx" @end example Three entry points provide a relatively simple interface to this moderately complicated situation: @deftypefn {Internal Function} {struct rx_solutions *} rx_basic_make_solutions (struct rx_registers * regs, struct rexp_node * expression, struct rexp_node ** subexps, int start, int end, struct rx_context_rules * rules, char * str) @deftypefnx {Internal Function} void rx_basic_free_solutions (struct rx_solutions * solns); @deftypefnx {Internal Function} {enum rx_answers} rx_next_solution (struct rx_solutions * solns) @deftypefnx {Internal Function} void rx_basic_free_solutions (struct rx_solutions * solns) @code{make_solutions} returns a lazilly constructed stream of possible solutions for a given regexp and target string. @code{next_solution} constructs and returns the next solution from a list returned by @code{make_solutions}. Solutions are returned in order of preferability according to the Posix spec. So, the first solution is the leftmost-longest, and the last, the rightmost-shortest. @var{regs} is where subexpression extents are tracked. @var{expression} is the expression to solve. @var{subexps} is as returned by @code{rx_posix_analyze_rexp}. @var{start} and @var{end} are the integer addresses of the data to be compared to the pattern. A solution must match all characters starting at @var{start} and ending at @code{@var{end} - 1}. @var{rules} are some bits that control the precise meaning of "^" and "$". See @code{"rxcontext.h"}. @var{str} is the data to be matched. Note that anchoring operators can cause the matcher to look beyond @var{start} and @var{end} when testing a match. It may make sense to pass @var{start} > 0, for example. @code{make_solutions} returns a pointer to a solution list, or 0 if an allocation fails. @code{next_solution} returns a status which can be: @example @code{rx_yes} -- The pattern matches exactly. @code{rx_maybe} -- The pattern does not match, but might if more characters were added to the string. @code{rx_no} -- The pattern does not match and could not be made to match even by adding more characters. @end example In addition, @var{regs} is filled in by @code{next_solution}. Finally, @code{free_solutions} is used to release resources consumed by a solution list. @end deftypefn @node Hairy Matches, NFA Functions, Posixly Correct Matches, General Functions @subsection Hairy Matches and Suspendable Searches The functions @code{rx_basic_make_solutions} and @code{rx_basic_free_solutions} are used to find matches in a one-piece string. More general entry points are provided for multi-piece strings, which may or may not be entirely available in memory at one time: @deftypefn {Internal Function} {struct rx_solutions *} rx_make_solutions (struct rx_registers * regs, struct rx_unfaniverse * verse, struct rexp_node * expression, struct rexp_node ** subexps, int cset_size, int start, int end, rx_vmfn vmfn, rx_contextfn contextfn, void * closure) @deftypefnx {Internal Function} void rx_free_solutions (struct rx_solutions * solns) These functions are similar to their @code{_basic_} analogues, but substitute some new parameters for the one piece string: @var{vmfn} is a function provided by the caller that returns at least part of the string being matched. @var{contextfn} is a function provided by the caller that does the work of the backreference operators, "^", and "$". @var{closure} is data provided by the caller, passed through to vmfn and contextfn. @example typedef enum rx_answers (*rx_vmfn) P((void * closure, unsigned char ** burst, int * len, int * offset, int start, int end, int need)); typedef enum rx_answers (*rx_contextfn) P((void * closure, int type, int start, int end, struct rx_registers * regs)); @end example @code{vmfn} is asked for a range of characters from @var{start} to @var{end}. @var{need} is in that range and is the only position that must be returned by vmfn. @var{burst}, @var{len}, and @var{offset} are output parameters. @var{offset} will get the integer address of burst, and @var{len} the number of valid characters in the burst. @var{vmfn} may be asked for the empty string from @var{end} to @var{end} and should be able to handle that case. If @code{vmfn} provides the needed character, it should return @code{rx_yes}. If @code{vmfn} can't provide the character now, but might be able to later, it should return @code{rx_maybe}. This will cause @code{rx_next_solution} to quickly return @code{rx_maybe}. If @code{rx_next_solution} is called again, it will retry the call to vmfn and possibly resume the search. If @code{vmfn} can not provide the needed character, it should return @code{rx_no}. @code{rx_next_solution} will not normally ask for out-of-range characters, so usually there is no good reason to return @code{rx_no}, but the code is supposed to be robust in case you do. The @code{contextfn} also returns a @code{yes/no/maybe} status. It tests whether the characters in @var{start} to @var{end} satisfy the operator @var{type}. @code{type} is an ascii value; one of the characters in "^$123456789". @end deftypefn @node NFA Functions, , Hairy Matches, General Functions @subsection NFA Functions and Super-NFA Functions Not really documented yet. @example rx.h, rxnfa.h -- translating an expression into an NFA. rxsuper.h -- optimizing the NFA into a DFA, or at least an NFA with fewer branch points. rxanal.h -- how to use DFAs to find longest matches, matches of a specific length, the shortest match, etc. rxsimp.h -- Translating a Posix "regular expression" into a true regular expression that describes a superset language. @end example @node Rx Theory, Improvements to Make, General Functions, Top @section Rx Theory @cindex Rx Theory There are two match algorithms. One is for truly regular regexps (those that can be reduced to a dfa). The other is for non-regular regexps. The dfa algorithm implements the idea suggested in @cite{Compilers} by Aho, Sethi and Ullman: @quotation [One] approach [to pattern matching regular expressions] is to use a DFA, but avoid constructing all of the transition table by using a technique called "lazy transition evaluation". Here, transitions are computed at run time [when] actually needed. [T]ransitions are stored in a cache. [....] If the cache becomes full, we can erase some previously computed transition to make room for the new transition. @end quotation The implementation in Rx is generalized from that, but the above description covers what is used for Posix patterns. The non-dfa algorithm implements a "recursive decomposition" technique described in email by Henry Spencer. For a given pattern, this algorithm first checks to see if a simpler, superset language, DFA-pattern matches. If it does, then this algorithm does the detail-work to see if the non-DFA pattern matches. The detail work usually involves recursing on subpatterns. For example, a concatentation of two subexpressions matches a string if the string can be divided into two parts, each matching one subexpression, in the right order. More than one solution is often possible for a given pattern. This ambiguity is the subject of the "leftmost longest" rules in the spec, and the back-tracking oriented stream-of-solution functions @code{rx_make_solutions}, @code{rx_next_solution} and @code{rx_free_solutions}. @example rxspencer.[ch] -- The non-DFA algorithm rxanal.[ch] rxsuper.[ch] rxnfa.[ch] -- The DFA algorithm @end example @node Improvements to Make, , Rx Theory, Top @section Improvements to Make @cindex Improvements to make An approximation of the GNU entry points is needed. This can not easily be an exact approximation because the GNU entry points have no function for freeing a compiled expression -- they assume that compiled expressions can be freed using a single call to "free(3)". For that reason, do not use the names of the old GNU entry points ...use "rx_" instead. This version of Rx hasn't been profiled and tuned. It might be nice to have some C++ types to hide reference counting and provide handy constructors for rexp_nodes and nfa parts. Right now, the regexp->regular regexp mapping in rxsimp.[ch] is cached for a particular input expression. I think it would probably be a big win in many situations if it were cached, like unique nfas (c.f. rxunfa.[ch]), for equivalent expressions. It would speed up matching to lift the non-regular constructs to the top of the syntax tree. e.g., if you have: @example pattern: "abc$" syntax tree: (concat (cset "a") (concat (cset "b") (concat (cset "c") (operator "$")))) @end example it involves a lot more work (in "next_solution) than if you have the equivalent: @example same pattern: "abc$" syntax tree: (concat (concat (cset "a") (concat (cset "b") (cset "c"))) (operator "$")) @end example in fact, while doing that, notice that subtrees of nothing-but-concat can conceivably be optimized so that next_solution just uses "strcmp". For that matter, a tree which is a disjunction of conjunction (a tree of "|" whose leaves are trees of concat nodes) could be optimized to use a table method (Boyer/Moore?). Again, this would be for next_solution but might involve adding stuff to rx_posix_analyze_rexp. As more and more optimizations are added, programmers should be given simple control over which optimizations are important for a particular pattern. For example, packages in emacs often include a few regexps that are used heavily -- it would be worth the trouble to arrange for those regexps to be compiled specially. Static compilation of regexps is mostly a job for lex, but posix patterns are more general and have a nicer interface. What's more, it'd be nice to have the benefits of static compilation even for dynamicly loaded regexps (e.g., imagine if when an emacs lisp file is byte-compiled, the regexps it contains are also compiled). A lot of optimizations apply to just syntax trees. One approach to static compilation is to make an ugly but fast-reading syntax for trees, and a printer for that syntax. Then, after optimiztaions, a pattern could be stored as a string and later read back in, already optimized. A similar trick could be done for the nfa built in rxnfa.c. next_solution already considers the length of a fixed-length subexpression, but many more lenght optimizations are possible. I'd guess these will pay off well. @cindex TODO dist/doc/Makefile.in100660 765 144 2041 6071420245 13173 0ustar lordusers# Makefile for GNU SED, a batch editor. # Copyright (C) 1987, 1991 Free Software Foundation, Inc. # # This file is part of GNU SED. # # GNU SED is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by # the Free Software Foundation; either version 2, or (at your option) # any later version. # # GNU SED is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with GNU SED; see the file COPYING. If not, write to # the Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. SHELL = /bin/sh srcdir = @srcdir@ VPATH = @srcdir@ all: install: clean: distclean: realclean: SUBDIR=. manifest = COPYING Makefile.in texinfo.tex manifest: for file in $(manifest) ; \ do echo $(SUBDIR)/$$file ; \ done dist/doc/COPYING100660 765 144 43112 6071420111 12175 0ustar lordusers GNU GENERAL PUBLIC LICENSE Version 2, June 1991 Copyright (C) 1989, 1991, 1992, 1993 Free Software Foundation, Inc. 675 Mass Ave, Cambridge, MA 02139, USA Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed. Preamble The licenses for most software are designed to take away your freedom to share and change it. By contrast, the GNU General Public License is intended to guarantee your freedom to share and change free software--to make sure the software is free for all its users. This General Public License applies to most of the Free Software Foundation's software and to any other program whose authors commit to using it. (Some other Free Software Foundation software is covered by the GNU Library General Public License instead.) You can apply it to your programs, too. When we speak of free software, we are referring to freedom, not price. Our General Public Licenses are designed to make sure that you have the freedom to distribute copies of free software (and charge for this service if you wish), that you receive source code or can get it if you want it, that you can change the software or use pieces of it in new free programs; and that you know you can do these things. To protect your rights, we need to make restrictions that forbid anyone to deny you these rights or to ask you to surrender the rights. These restrictions translate to certain responsibilities for you if you distribute copies of the software, or if you modify it. For example, if you distribute copies of such a program, whether gratis or for a fee, you must give the recipients all the rights that you have. You must make sure that they, too, receive or can get the source code. And you must show them these terms so they know their rights. We protect your rights with two steps: (1) copyright the software, and (2) offer you this license which gives you legal permission to copy, distribute and/or modify the software. Also, for each author's protection and ours, we want to make certain that everyone understands that there is no warranty for this free software. If the software is modified by someone else and passed on, we want its recipients to know that what they have is not the original, so that any problems introduced by others will not reflect on the original authors' reputations. Finally, any free program is threatened constantly by software patents. We wish to avoid the danger that redistributors of a free program will individually obtain patent licenses, in effect making the program proprietary. To prevent this, we have made it clear that any patent must be licensed for everyone's free use or not licensed at all. The precise terms and conditions for copying, distribution and modification follow. GNU GENERAL PUBLIC LICENSE TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION 0. This License applies to any program or other work which contains a notice placed by the copyright holder saying it may be distributed under the terms of this General Public License. The "Program", below, refers to any such program or work, and a "work based on the Program" means either the Program or any derivative work under copyright law: that is to say, a work containing the Program or a portion of it, either verbatim or with modifications and/or translated into another language. (Hereinafter, translation is included without limitation in the term "modification".) Each licensee is addressed as "you". Activities other than copying, distribution and modification are not covered by this License; they are outside its scope. The act of running the Program is not restricted, and the output from the Program is covered only if its contents constitute a work based on the Program (independent of having been made by running the Program). Whether that is true depends on what the Program does. 1. You may copy and distribute verbatim copies of the Program's source code as you receive it, in any medium, provided that you conspicuously and appropriately publish on each copy an appropriate copyright notice and disclaimer of warranty; keep intact all the notices that refer to this License and to the absence of any warranty; and give any other recipients of the Program a copy of this License along with the Program. You may charge a fee for the physical act of transferring a copy, and you may at your option offer warranty protection in exchange for a fee. 2. You may modify your copy or copies of the Program or any portion of it, thus forming a work based on the Program, and copy and distribute such modifications or work under the terms of Section 1 above, provided that you also meet all of these conditions: a) You must cause the modified files to carry prominent notices stating that you changed the files and the date of any change. b) You must cause any work that you distribute or publish, that in whole or in part contains or is derived from the Program or any part thereof, to be licensed as a whole at no charge to all third parties under the terms of this License. c) If the modified program normally reads commands interactively when run, you must cause it, when started running for such interactive use in the most ordinary way, to print or display an announcement including an appropriate copyright notice and a notice that there is no warranty (or else, saying that you provide a warranty) and that users may redistribute the program under these conditions, and telling the user how to view a copy of this License. (Exception: if the Program itself is interactive but does not normally print such an announcement, your work based on the Program is not required to print an announcement.) These requirements apply to the modified work as a whole. If identifiable sections of that work are not derived from the Program, and can be reasonably considered independent and separate works in themselves, then this License, and its terms, do not apply to those sections when you distribute them as separate works. But when you distribute the same sections as part of a whole which is a work based on the Program, the distribution of the whole must be on the terms of this License, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it. Thus, it is not the intent of this section to claim rights or contest your rights to work written entirely by you; rather, the intent is to exercise the right to control the distribution of derivative or collective works based on the Program. In addition, mere aggregation of another work not based on the Program with the Program (or with a work based on the Program) on a volume of a storage or distribution medium does not bring the other work under the scope of this License. 3. You may copy and distribute the Program (or a work based on it, under Section 2) in object code or executable form under the terms of Sections 1 and 2 above provided that you also do one of the following: a) Accompany it with the complete corresponding machine-readable source code, which must be distributed under the terms of Sections 1 and 2 above on a medium customarily used for software interchange; or, b) Accompany it with a written offer, valid for at least three years, to give any third party, for a charge no more than your cost of physically performing source distribution, a complete machine-readable copy of the corresponding source code, to be distributed under the terms of Sections 1 and 2 above on a medium customarily used for software interchange; or, c) Accompany it with the information you received as to the offer to distribute corresponding source code. (This alternative is allowed only for noncommercial distribution and only if you received the program in object code or executable form with such an offer, in accord with Subsection b above.) The source code for a work means the preferred form of the work for making modifications to it. For an executable work, complete source code means all the source code for all modules it contains, plus any associated interface definition files, plus the scripts used to control compilation and installation of the executable. However, as a special exception, the source code distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable. If distribution of executable or object code is made by offering access to copy from a designated place, then offering equivalent access to copy the source code from the same place counts as distribution of the source code, even though third parties are not compelled to copy the source along with the object code. 4. You may not copy, modify, sublicense, or distribute the Program except as expressly provided under this License. Any attempt otherwise to copy, modify, sublicense or distribute the Program is void, and will automatically terminate your rights under this License. However, parties who have received copies, or rights, from you under this License will not have their licenses terminated so long as such parties remain in full compliance. 5. You are not required to accept this License, since you have not signed it. However, nothing else grants you permission to modify or distribute the Program or its derivative works. These actions are prohibited by law if you do not accept this License. Therefore, by modifying or distributing the Program (or any work based on the Program), you indicate your acceptance of this License to do so, and all its terms and conditions for copying, distributing or modifying the Program or works based on it. 6. Each time you redistribute the Program (or any work based on the Program), the recipient automatically receives a license from the original licensor to copy, distribute or modify the Program subject to these terms and conditions. You may not impose any further restrictions on the recipients' exercise of the rights granted herein. You are not responsible for enforcing compliance by third parties to this License. 7. If, as a consequence of a court judgment or allegation of patent infringement or for any other reason (not limited to patent issues), conditions are imposed on you (whether by court order, agreement or otherwise) that contradict the conditions of this License, they do not excuse you from the conditions of this License. If you cannot distribute so as to satisfy simultaneously your obligations under this License and any other pertinent obligations, then as a consequence you may not distribute the Program at all. For example, if a patent license would not permit royalty-free redistribution of the Program by all those who receive copies directly or indirectly through you, then the only way you could satisfy both it and this License would be to refrain entirely from distribution of the Program. If any portion of this section is held invalid or unenforceable under any particular circumstance, the balance of the section is intended to apply and the section as a whole is intended to apply in other circumstances. It is not the purpose of this section to induce you to infringe any patents or other property right claims or to contest validity of any such claims; this section has the sole purpose of protecting the integrity of the free software distribution system, which is implemented by public license practices. Many people have made generous contributions to the wide range of software distributed through that system in reliance on consistent application of that system; it is up to the author/donor to decide if he or she is willing to distribute software through any other system and a licensee cannot impose that choice. This section is intended to make thoroughly clear what is believed to be a consequence of the rest of this License. 8. If the distribution and/or use of the Program is restricted in certain countries either by patents or by copyrighted interfaces, the original copyright holder who places the Program under this License may add an explicit geographical distribution limitation excluding those countries, so that distribution is permitted only in or among countries not thus excluded. In such case, this License incorporates the limitation as if written in the body of this License. 9. The Free Software Foundation may publish revised and/or new versions of the General Public License from time to time. Such new versions will be similar in spirit to the present version, but may differ in detail to address new problems or concerns. Each version is given a distinguishing version number. If the Program specifies a version number of this License which applies to it and "any later version", you have the option of following the terms and conditions either of that version or of any later version published by the Free Software Foundation. If the Program does not specify a version number of this License, you may choose any version ever published by the Free Software Foundation. 10. If you wish to incorporate parts of the Program into other free programs whose distribution conditions are different, write to the author to ask for permission. For software which is copyrighted by the Free Software Foundation, write to the Free Software Foundation; we sometimes make exceptions for this. Our decision will be guided by the two goals of preserving the free status of all derivatives of our free software and of promoting the sharing and reuse of software generally. NO WARRANTY 11. BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION. 12. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. END OF TERMS AND CONDITIONS Appendix: How to Apply These Terms to Your New Programs If you develop a new program, and you want it to be of the greatest possible use to the public, the best way to achieve this is to make it free software which everyone can redistribute and change under these terms. To do so, attach the following notices to the program. It is safest to attach them to the start of each source file to most effectively convey the exclusion of warranty; and each file should have at least the "copyright" line and a pointer to where the full notice is found. Copyright (C) 19yy This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. Also add information on how to contact you by electronic and paper mail. If the program is interactive, make it output a short notice like this when it starts in an interactive mode: Gnomovision version 69, Copyright (C) 19yy name of author Gnomovision comes with ABSOLUTELY NO WARRANTY; for details type `show w'. This is free software, and you are welcome to redistribute it under certain conditions; type `show c' for details. The hypothetical commands `show w' and `show c' should show the appropriate parts of the General Public License. Of course, the commands you use may be called something other than `show w' and `show c'; they could even be mouse-clicks or menu items--whatever suits your program. You should also get your employer (if you work as a programmer) or your school, if any, to sign a "copyright disclaimer" for the program, if necessary. Here is a sample; alter the names: Yoyodyne, Inc., hereby disclaims all copyright interest in the program `Gnomovision' (which makes passes at compilers) written by James Hacker. , 1 April 1989 Ty Coon, President of Vice This General Public License does not permit incorporating your program into proprietary programs. If your program is a subroutine library, you may consider it more useful to permit linking proprietary applications with the library. If this is what you want to do, use the GNU Library General Public License instead of this License. dist/doc/texinfo.tex100440 765 144 435354 6071420233 13376 0ustar lordusers%% TeX macros to handle texinfo files % Copyright (C) 1985, 86, 88, 90, 91, 92, 93, 1994 Free Software Foundation, Inc. %This texinfo.tex file is free software; you can redistribute it and/or %modify it under the terms of the GNU General Public License as %published by the Free Software Foundation; either version 2, or (at %your option) any later version. %This texinfo.tex file is distributed in the hope that it will be %useful, but WITHOUT ANY WARRANTY; without even the implied warranty %of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU %General Public License for more details. %You should have received a copy of the GNU General Public License %along with this texinfo.tex file; see the file COPYING. If not, write %to the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, %USA. %In other words, you are welcome to use, share and improve this program. %You are forbidden to forbid anyone else to use, share and improve %what you give them. Help stamp out software-hoarding! % Send bug reports to bug-texinfo@prep.ai.mit.edu. % Please include a *precise* test case in each bug report. % Make it possible to create a .fmt file just by loading this file: % if the underlying format is not loaded, start by loading it now. % Added by gildea November 1993. \expandafter\ifx\csname fmtname\endcsname\relax\input plain\fi % This automatically updates the version number based on RCS. \def\deftexinfoversion$#1: #2 ${\def\texinfoversion{#2}} \deftexinfoversion$Revision: 2.145 $ \message{Loading texinfo package [Version \texinfoversion]:} % If in a .fmt file, print the version number % and turn on active characters that we couldn't do earlier because % they might have appeared in the input file name. \everyjob{\message{[Texinfo version \texinfoversion]}\message{} \catcode`+=\active \catcode`\_=\active} % Save some parts of plain tex whose names we will redefine. \let\ptextilde=\~ \let\ptexlbrace=\{ \let\ptexrbrace=\} \let\ptexdots=\dots \let\ptexdot=\. \let\ptexstar=\* \let\ptexend=\end \let\ptexbullet=\bullet \let\ptexb=\b \let\ptexc=\c \let\ptexi=\i \let\ptext=\t \let\ptexl=\l \let\ptexL=\L % Be sure we're in horizontal mode when doing a tie, since we make space % equivalent to this in @example-like environments. Otherwise, a space % at the beginning of a line will start with \penalty -- and % since \penalty is valid in vertical mode, we'd end up putting the % penalty on the vertical list instead of in the new paragraph. {\catcode`@ = 11 \gdef\tie{\leavevmode\penalty\@M\ } } \let\~ = \tie % And make it available as @~. \message{Basics,} \chardef\other=12 % If this character appears in an error message or help string, it % starts a new line in the output. \newlinechar = `^^J % Set up fixed words for English. \ifx\putwordChapter\undefined{\gdef\putwordChapter{Chapter}}\fi% \def\putwordInfo{Info}% \ifx\putwordSee\undefined{\gdef\putwordSee{See}}\fi% \ifx\putwordsee\undefined{\gdef\putwordsee{see}}\fi% \ifx\putwordfile\undefined{\gdef\putwordfile{file}}\fi% \ifx\putwordpage\undefined{\gdef\putwordpage{page}}\fi% \ifx\putwordsection\undefined{\gdef\putwordsection{section}}\fi% \ifx\putwordSection\undefined{\gdef\putwordSection{Section}}\fi% \ifx\putwordTableofContents\undefined{\gdef\putwordTableofContents{Table of Contents}}\fi% \ifx\putwordShortContents\undefined{\gdef\putwordShortContents{Short Contents}}\fi% \ifx\putwordAppendix\undefined{\gdef\putwordAppendix{Appendix}}\fi% % Ignore a token. % \def\gobble#1{} \hyphenation{ap-pen-dix} \hyphenation{mini-buf-fer mini-buf-fers} \hyphenation{eshell} % Margin to add to right of even pages, to left of odd pages. \newdimen \bindingoffset \bindingoffset=0pt \newdimen \normaloffset \normaloffset=\hoffset \newdimen\pagewidth \newdimen\pageheight \pagewidth=\hsize \pageheight=\vsize % Sometimes it is convenient to have everything in the transcript file % and nothing on the terminal. We don't just call \tracingall here, % since that produces some useless output on the terminal. % \def\gloggingall{\begingroup \globaldefs = 1 \loggingall \endgroup}% \def\loggingall{\tracingcommands2 \tracingstats2 \tracingpages1 \tracingoutput1 \tracinglostchars1 \tracingmacros2 \tracingparagraphs1 \tracingrestores1 \showboxbreadth\maxdimen\showboxdepth\maxdimen }% %---------------------Begin change----------------------- % %%%% For @cropmarks command. % Dimensions to add cropmarks at corners Added by P. A. MacKay, 12 Nov. 1986 % \newdimen\cornerlong \newdimen\cornerthick \newdimen \topandbottommargin \newdimen \outerhsize \newdimen \outervsize \cornerlong=1pc\cornerthick=.3pt % These set size of cropmarks \outerhsize=7in %\outervsize=9.5in % Alternative @smallbook page size is 9.25in \outervsize=9.25in \topandbottommargin=.75in % %---------------------End change----------------------- % \onepageout takes a vbox as an argument. Note that \pagecontents % does insertions itself, but you have to call it yourself. \chardef\PAGE=255 \output={\onepageout{\pagecontents\PAGE}} \def\onepageout#1{\hoffset=\normaloffset \ifodd\pageno \advance\hoffset by \bindingoffset \else \advance\hoffset by -\bindingoffset\fi {\escapechar=`\\\relax % makes sure backslash is used in output files. \shipout\vbox{{\let\hsize=\pagewidth \makeheadline} \pagebody{#1}% {\let\hsize=\pagewidth \makefootline}}}% \advancepageno \ifnum\outputpenalty>-20000 \else\dosupereject\fi} %%%% For @cropmarks command %%%% % Here is a modification of the main output routine for Near East Publications % This provides right-angle cropmarks at all four corners. % The contents of the page are centerlined into the cropmarks, % and any desired binding offset is added as an \hskip on either % site of the centerlined box. (P. A. MacKay, 12 November, 1986) % \def\croppageout#1{\hoffset=0pt % make sure this doesn't mess things up {\escapechar=`\\\relax % makes sure backslash is used in output files. \shipout \vbox to \outervsize{\hsize=\outerhsize \vbox{\line{\ewtop\hfill\ewtop}} \nointerlineskip \line{\vbox{\moveleft\cornerthick\nstop} \hfill \vbox{\moveright\cornerthick\nstop}} \vskip \topandbottommargin \centerline{\ifodd\pageno\hskip\bindingoffset\fi \vbox{ {\let\hsize=\pagewidth \makeheadline} \pagebody{#1} {\let\hsize=\pagewidth \makefootline}} \ifodd\pageno\else\hskip\bindingoffset\fi} \vskip \topandbottommargin plus1fill minus1fill \boxmaxdepth\cornerthick \line{\vbox{\moveleft\cornerthick\nsbot} \hfill \vbox{\moveright\cornerthick\nsbot}} \nointerlineskip \vbox{\line{\ewbot\hfill\ewbot}} }} \advancepageno \ifnum\outputpenalty>-20000 \else\dosupereject\fi} % % Do @cropmarks to get crop marks \def\cropmarks{\let\onepageout=\croppageout } \newinsert\margin \dimen\margin=\maxdimen \def\pagebody#1{\vbox to\pageheight{\boxmaxdepth=\maxdepth #1}} {\catcode`\@ =11 \gdef\pagecontents#1{\ifvoid\topins\else\unvbox\topins\fi % marginal hacks, juha@viisa.uucp (Juha Takala) \ifvoid\margin\else % marginal info is present \rlap{\kern\hsize\vbox to\z@{\kern1pt\box\margin \vss}}\fi \dimen@=\dp#1 \unvbox#1 \ifvoid\footins\else\vskip\skip\footins\footnoterule \unvbox\footins\fi \ifr@ggedbottom \kern-\dimen@ \vfil \fi} } % % Here are the rules for the cropmarks. Note that they are % offset so that the space between them is truly \outerhsize or \outervsize % (P. A. MacKay, 12 November, 1986) % \def\ewtop{\vrule height\cornerthick depth0pt width\cornerlong} \def\nstop{\vbox {\hrule height\cornerthick depth\cornerlong width\cornerthick}} \def\ewbot{\vrule height0pt depth\cornerthick width\cornerlong} \def\nsbot{\vbox {\hrule height\cornerlong depth\cornerthick width\cornerthick}} % Parse an argument, then pass it to #1. The argument is the rest of % the input line (except we remove a trailing comment). #1 should be a % macro which expects an ordinary undelimited TeX argument. % \def\parsearg#1{% \let\next = #1% \begingroup \obeylines \futurelet\temp\parseargx } % If the next token is an obeyed space (from an @example environment or % the like), remove it and recurse. Otherwise, we're done. \def\parseargx{% % \obeyedspace is defined far below, after the definition of \sepspaces. \ifx\obeyedspace\temp \expandafter\parseargdiscardspace \else \expandafter\parseargline \fi } % Remove a single space (as the delimiter token to the macro call). {\obeyspaces % \gdef\parseargdiscardspace {\futurelet\temp\parseargx}} {\obeylines % \gdef\parseargline#1^^M{% \endgroup % End of the group started in \parsearg. % % First remove any @c comment, then any @comment. % Result of each macro is put in \toks0. \argremovec #1\c\relax % \expandafter\argremovecomment \the\toks0 \comment\relax % % % Call the caller's macro, saved as \next in \parsearg. \expandafter\next\expandafter{\the\toks0}% }% } % Since all \c{,omment} does is throw away the argument, we can let TeX % do that for us. The \relax here is matched by the \relax in the call % in \parseargline; it could be more or less anything, its purpose is % just to delimit the argument to the \c. \def\argremovec#1\c#2\relax{\toks0 = {#1}} \def\argremovecomment#1\comment#2\relax{\toks0 = {#1}} % \argremovec{,omment} might leave us with trailing spaces, though; e.g., % @end itemize @c foo % will have two active spaces as part of the argument with the % `itemize'. Here we remove all active spaces from #1, and assign the % result to \toks0. % % This loses if there are any *other* active characters besides spaces % in the argument -- _ ^ +, for example -- since they get expanded. % Fortunately, Texinfo does not define any such commands. (If it ever % does, the catcode of the characters in questionwill have to be changed % here.) But this means we cannot call \removeactivespaces as part of % \argremovec{,omment}, since @c uses \parsearg, and thus the argument % that \parsearg gets might well have any character at all in it. % \def\removeactivespaces#1{% \begingroup \ignoreactivespaces \edef\temp{#1}% \global\toks0 = \expandafter{\temp}% \endgroup } % Change the active space to expand to nothing. % \begingroup \obeyspaces \gdef\ignoreactivespaces{\obeyspaces\let =\empty} \endgroup \def\flushcr{\ifx\par\lisppar \def\next##1{}\else \let\next=\relax \fi \next} %% These are used to keep @begin/@end levels from running away %% Call \inENV within environments (after a \begingroup) \newif\ifENV \ENVfalse \def\inENV{\ifENV\relax\else\ENVtrue\fi} \def\ENVcheck{% \ifENV\errmessage{Still within an environment. Type Return to continue.} \endgroup\fi} % This is not perfect, but it should reduce lossage % @begin foo is the same as @foo, for now. \newhelp\EMsimple{Type to continue.} \outer\def\begin{\parsearg\beginxxx} \def\beginxxx #1{% \expandafter\ifx\csname #1\endcsname\relax {\errhelp=\EMsimple \errmessage{Undefined command @begin #1}}\else \csname #1\endcsname\fi} % @end foo executes the definition of \Efoo. % \def\end{\parsearg\endxxx} \def\endxxx #1{% \removeactivespaces{#1}% \edef\endthing{\the\toks0}% % \expandafter\ifx\csname E\endthing\endcsname\relax \expandafter\ifx\csname \endthing\endcsname\relax % There's no \foo, i.e., no ``environment'' foo. \errhelp = \EMsimple \errmessage{Undefined command `@end \endthing'}% \else \unmatchedenderror\endthing \fi \else % Everything's ok; the right environment has been started. \csname E\endthing\endcsname \fi } % There is an environment #1, but it hasn't been started. Give an error. % \def\unmatchedenderror#1{% \errhelp = \EMsimple \errmessage{This `@end #1' doesn't have a matching `@#1'}% } % Define the control sequence \E#1 to give an unmatched @end error. % \def\defineunmatchedend#1{% \expandafter\def\csname E#1\endcsname{\unmatchedenderror{#1}}% } % Single-spacing is done by various environments (specifically, in % \nonfillstart and \quotations). \newskip\singlespaceskip \singlespaceskip = 12.5pt \def\singlespace{% % Why was this kern here? It messes up equalizing space above and below % environments. --karl, 6may93 %{\advance \baselineskip by -\singlespaceskip %\kern \baselineskip}% \setleading \singlespaceskip } %% Simple single-character @ commands % @@ prints an @ % Kludge this until the fonts are right (grr). \def\@{{\tt \char '100}} % This is turned off because it was never documented % and you can use @w{...} around a quote to suppress ligatures. %% Define @` and @' to be the same as ` and ' %% but suppressing ligatures. %\def\`{{`}} %\def\'{{'}} % Used to generate quoted braces. \def\mylbrace {{\tt \char '173}} \def\myrbrace {{\tt \char '175}} \let\{=\mylbrace \let\}=\myrbrace % @: forces normal size whitespace following. \def\:{\spacefactor=1000 } % @* forces a line break. \def\*{\hfil\break\hbox{}\ignorespaces} % @. is an end-of-sentence period. \def\.{.\spacefactor=3000 } % @enddots{} is an end-of-sentence ellipsis. \gdef\enddots{$\mathinner{\ldotp\ldotp\ldotp\ldotp}$\spacefactor=3000} % @! is an end-of-sentence bang. \gdef\!{!\spacefactor=3000 } % @? is an end-of-sentence query. \gdef\?{?\spacefactor=3000 } % @w prevents a word break. Without the \leavevmode, @w at the % beginning of a paragraph, when TeX is still in vertical mode, would % produce a whole line of output instead of starting the paragraph. \def\w#1{\leavevmode\hbox{#1}} % @group ... @end group forces ... to be all on one page, by enclosing % it in a TeX vbox. We use \vtop instead of \vbox to construct the box % to keep its height that of a normal line. According to the rules for % \topskip (p.114 of the TeXbook), the glue inserted is % max (\topskip - \ht (first item), 0). If that height is large, % therefore, no glue is inserted, and the space between the headline and % the text is small, which looks bad. % \def\group{\begingroup \ifnum\catcode13=\active \else \errhelp = \groupinvalidhelp \errmessage{@group invalid in context where filling is enabled}% \fi % % The \vtop we start below produces a box with normal height and large % depth; thus, TeX puts \baselineskip glue before it, and (when the % next line of text is done) \lineskip glue after it. (See p.82 of % the TeXbook.) Thus, space below is not quite equal to space % above. But it's pretty close. \def\Egroup{% \egroup % End the \vtop. \endgroup % End the \group. }% % \vtop\bgroup % We have to put a strut on the last line in case the @group is in % the midst of an example, rather than completely enclosing it. % Otherwise, the interline space between the last line of the group % and the first line afterwards is too small. But we can't put the % strut in \Egroup, since there it would be on a line by itself. % Hence this just inserts a strut at the beginning of each line. \everypar = {\strut}% % % Since we have a strut on every line, we don't need any of TeX's % normal interline spacing. \offinterlineskip % % OK, but now we have to do something about blank % lines in the input in @example-like environments, which normally % just turn into \lisppar, which will insert no space now that we've % turned off the interline space. Simplest is to make them be an % empty paragraph. \ifx\par\lisppar \edef\par{\leavevmode \par}% % % Reset ^^M's definition to new definition of \par. \obeylines \fi % % Do @comment since we are called inside an environment such as % @example, where each end-of-line in the input causes an % end-of-line in the output. We don't want the end-of-line after % the `@group' to put extra space in the output. Since @group % should appear on a line by itself (according to the Texinfo % manual), we don't worry about eating any user text. \comment } % % TeX puts in an \escapechar (i.e., `@') at the beginning of the help % message, so this ends up printing `@group can only ...'. % \newhelp\groupinvalidhelp{% group can only be used in environments such as @example,^^J% where each line of input produces a line of output.} % @need space-in-mils % forces a page break if there is not space-in-mils remaining. \newdimen\mil \mil=0.001in \def\need{\parsearg\needx} % Old definition--didn't work. %\def\needx #1{\par % %% This method tries to make TeX break the page naturally %% if the depth of the box does not fit. %{\baselineskip=0pt% %\vtop to #1\mil{\vfil}\kern -#1\mil\penalty 10000 %\prevdepth=-1000pt %}} \def\needx#1{% % Go into vertical mode, so we don't make a big box in the middle of a % paragraph. \par % % Don't add any leading before our big empty box, but allow a page % break, since the best break might be right here. \allowbreak \nointerlineskip \vtop to #1\mil{\vfil}% % % TeX does not even consider page breaks if a penalty added to the % main vertical list is 10000 or more. But in order to see if the % empty box we just added fits on the page, we must make it consider % page breaks. On the other hand, we don't want to actually break the % page after the empty box. So we use a penalty of 9999. % % There is an extremely small chance that TeX will actually break the % page at this \penalty, if there are no other feasible breakpoints in % sight. (If the user is using lots of big @group commands, which % almost-but-not-quite fill up a page, TeX will have a hard time doing % good page breaking, for example.) However, I could not construct an % example where a page broke at this \penalty; if it happens in a real % document, then we can reconsider our strategy. \penalty9999 % % Back up by the size of the box, whether we did a page break or not. \kern -#1\mil % % Do not allow a page break right after this kern. \nobreak } % @br forces paragraph break \let\br = \par % @dots{} output some dots \def\dots{$\ldots$} % @page forces the start of a new page \def\page{\par\vfill\supereject} % @exdent text.... % outputs text on separate line in roman font, starting at standard page margin % This records the amount of indent in the innermost environment. % That's how much \exdent should take out. \newskip\exdentamount % This defn is used inside fill environments such as @defun. \def\exdent{\parsearg\exdentyyy} \def\exdentyyy #1{{\hfil\break\hbox{\kern -\exdentamount{\rm#1}}\hfil\break}} % This defn is used inside nofill environments such as @example. \def\nofillexdent{\parsearg\nofillexdentyyy} \def\nofillexdentyyy #1{{\advance \leftskip by -\exdentamount \leftline{\hskip\leftskip{\rm#1}}}} %\hbox{{\rm#1}}\hfil\break}} % @include file insert text of that file as input. \def\include{\parsearg\includezzz} %Use \input\thisfile to avoid blank after \input, which may be an active %char (in which case the blank would become the \input argument). %The grouping keeps the value of \thisfile correct even when @include %is nested. \def\includezzz #1{\begingroup \def\thisfile{#1}\input\thisfile \endgroup} \def\thisfile{} % @center line outputs that line, centered \def\center{\parsearg\centerzzz} \def\centerzzz #1{{\advance\hsize by -\leftskip \advance\hsize by -\rightskip \centerline{#1}}} % @sp n outputs n lines of vertical space \def\sp{\parsearg\spxxx} \def\spxxx #1{\par \vskip #1\baselineskip} % @comment ...line which is ignored... % @c is the same as @comment % @ignore ... @end ignore is another way to write a comment \def\comment{\catcode 64=\other \catcode 123=\other \catcode 125=\other% \parsearg \commentxxx} \def\commentxxx #1{\catcode 64=0 \catcode 123=1 \catcode 125=2 } \let\c=\comment % Prevent errors for section commands. % Used in @ignore and in failing conditionals. \def\ignoresections{% \let\chapter=\relax \let\unnumbered=\relax \let\top=\relax \let\unnumberedsec=\relax \let\unnumberedsection=\relax \let\unnumberedsubsec=\relax \let\unnumberedsubsection=\relax \let\unnumberedsubsubsec=\relax \let\unnumberedsubsubsection=\relax \let\section=\relax \let\subsec=\relax \let\subsubsec=\relax \let\subsection=\relax \let\subsubsection=\relax \let\appendix=\relax \let\appendixsec=\relax \let\appendixsection=\relax \let\appendixsubsec=\relax \let\appendixsubsection=\relax \let\appendixsubsubsec=\relax \let\appendixsubsubsection=\relax \let\contents=\relax \let\smallbook=\relax \let\titlepage=\relax } % Used in nested conditionals, where we have to parse the Texinfo source % and so want to turn off most commands, in case they are used % incorrectly. % \def\ignoremorecommands{% \let\defcv = \relax \let\deffn = \relax \let\deffnx = \relax \let\defindex = \relax \let\defivar = \relax \let\defmac = \relax \let\defmethod = \relax \let\defop = \relax \let\defopt = \relax \let\defspec = \relax \let\deftp = \relax \let\deftypefn = \relax \let\deftypefun = \relax \let\deftypevar = \relax \let\deftypevr = \relax \let\defun = \relax \let\defvar = \relax \let\defvr = \relax \let\ref = \relax \let\xref = \relax \let\printindex = \relax \let\pxref = \relax \let\settitle = \relax \let\include = \relax \let\lowersections = \relax \let\down = \relax \let\raisesections = \relax \let\up = \relax \let\set = \relax \let\clear = \relax \let\item = \relax \let\message = \relax } % Ignore @ignore ... @end ignore. % \def\ignore{\doignore{ignore}} % Also ignore @ifinfo, @ifhtml, @html, @menu, and @direntry text. % \def\ifinfo{\doignore{ifinfo}} \def\ifhtml{\doignore{ifhtml}} \def\html{\doignore{html}} \def\menu{\doignore{menu}} \def\direntry{\doignore{direntry}} % Ignore text until a line `@end #1'. % \def\doignore#1{\begingroup % Don't complain about control sequences we have declared \outer. \ignoresections % % Define a command to swallow text until we reach `@end #1'. \long\def\doignoretext##1\end #1{\enddoignore}% % % Make sure that spaces turn into tokens that match what \doignoretext wants. \catcode32 = 10 % % And now expand that command. \doignoretext } % What we do to finish off ignored text. % \def\enddoignore{\endgroup\ignorespaces}% \newif\ifwarnedobs\warnedobsfalse \def\obstexwarn{% \ifwarnedobs\relax\else % We need to warn folks that they may have trouble with TeX 3.0. % This uses \immediate\write16 rather than \message to get newlines. \immediate\write16{} \immediate\write16{***WARNING*** for users of Unix TeX 3.0!} \immediate\write16{This manual trips a bug in TeX version 3.0 (tex hangs).} \immediate\write16{If you are running another version of TeX, relax.} \immediate\write16{If you are running Unix TeX 3.0, kill this TeX process.} \immediate\write16{ Then upgrade your TeX installation if you can.} \immediate\write16{If you are stuck with version 3.0, run the} \immediate\write16{ script ``tex3patch'' from the Texinfo distribution} \immediate\write16{ to use a workaround.} \immediate\write16{} \warnedobstrue \fi } % **In TeX 3.0, setting text in \nullfont hangs tex. For a % workaround (which requires the file ``dummy.tfm'' to be installed), % uncomment the following line: %%%%%\font\nullfont=dummy\let\obstexwarn=\relax % Ignore text, except that we keep track of conditional commands for % purposes of nesting, up to an `@end #1' command. % \def\nestedignore#1{% \obstexwarn % We must actually expand the ignored text to look for the @end % command, so that nested ignore constructs work. Thus, we put the % text into a \vbox and then do nothing with the result. To minimize % the change of memory overflow, we follow the approach outlined on % page 401 of the TeXbook: make the current font be a dummy font. % \setbox0 = \vbox\bgroup % Don't complain about control sequences we have declared \outer. \ignoresections % % Define `@end #1' to end the box, which will in turn undefine the % @end command again. \expandafter\def\csname E#1\endcsname{\egroup\ignorespaces}% % % We are going to be parsing Texinfo commands. Most cause no % trouble when they are used incorrectly, but some commands do % complicated argument parsing or otherwise get confused, so we % undefine them. % % We can't do anything about stray @-signs, unfortunately; % they'll produce `undefined control sequence' errors. \ignoremorecommands % % Set the current font to be \nullfont, a TeX primitive, and define % all the font commands to also use \nullfont. We don't use % dummy.tfm, as suggested in the TeXbook, because not all sites % might have that installed. Therefore, math mode will still % produce output, but that should be an extremely small amount of % stuff compared to the main input. % \nullfont \let\tenrm = \nullfont \let\tenit = \nullfont \let\tensl = \nullfont \let\tenbf = \nullfont \let\tentt = \nullfont \let\smallcaps = \nullfont \let\tensf = \nullfont % Similarly for index fonts (mostly for their use in % smallexample) \let\indrm = \nullfont \let\indit = \nullfont \let\indsl = \nullfont \let\indbf = \nullfont \let\indtt = \nullfont \let\indsc = \nullfont \let\indsf = \nullfont % % Don't complain when characters are missing from the fonts. \tracinglostchars = 0 % % Don't bother to do space factor calculations. \frenchspacing % % Don't report underfull hboxes. \hbadness = 10000 % % Do minimal line-breaking. \pretolerance = 10000 % % Do not execute instructions in @tex \def\tex{\doignore{tex}} } % @set VAR sets the variable VAR to an empty value. % @set VAR REST-OF-LINE sets VAR to the value REST-OF-LINE. % % Since we want to separate VAR from REST-OF-LINE (which might be % empty), we can't just use \parsearg; we have to insert a space of our % own to delimit the rest of the line, and then take it out again if we % didn't need it. % \def\set{\parsearg\setxxx} \def\setxxx#1{\setyyy#1 \endsetyyy} \def\setyyy#1 #2\endsetyyy{% \def\temp{#2}% \ifx\temp\empty \global\expandafter\let\csname SET#1\endcsname = \empty \else \setzzz{#1}#2\endsetzzz % Remove the trailing space \setxxx inserted. \fi } % Can't use \xdef to pre-expand #2 and save some time, since \temp or % \next or other control sequences that we've defined might get us into % an infinite loop. Consider `@set foo @cite{bar}'. \def\setzzz#1#2 \endsetzzz{\expandafter\gdef\csname SET#1\endcsname{#2}} % @clear VAR clears (i.e., unsets) the variable VAR. % \def\clear{\parsearg\clearxxx} \def\clearxxx#1{\global\expandafter\let\csname SET#1\endcsname=\relax} % @value{foo} gets the text saved in variable foo. % \def\value#1{\expandafter \ifx\csname SET#1\endcsname\relax {\{No value for ``#1''\}} \else \csname SET#1\endcsname \fi} % @ifset VAR ... @end ifset reads the `...' iff VAR has been defined % with @set. % \def\ifset{\parsearg\ifsetxxx} \def\ifsetxxx #1{% \expandafter\ifx\csname SET#1\endcsname\relax \expandafter\ifsetfail \else \expandafter\ifsetsucceed \fi } \def\ifsetsucceed{\conditionalsucceed{ifset}} \def\ifsetfail{\nestedignore{ifset}} \defineunmatchedend{ifset} % @ifclear VAR ... @end ifclear reads the `...' iff VAR has never been % defined with @set, or has been undefined with @clear. % \def\ifclear{\parsearg\ifclearxxx} \def\ifclearxxx #1{% \expandafter\ifx\csname SET#1\endcsname\relax \expandafter\ifclearsucceed \else \expandafter\ifclearfail \fi } \def\ifclearsucceed{\conditionalsucceed{ifclear}} \def\ifclearfail{\nestedignore{ifclear}} \defineunmatchedend{ifclear} % @iftex always succeeds; we read the text following, through @end % iftex). But `@end iftex' should be valid only after an @iftex. % \def\iftex{\conditionalsucceed{iftex}} \defineunmatchedend{iftex} % We can't just want to start a group at @iftex (for example) and end it % at @end iftex, since then @set commands inside the conditional have no % effect (they'd get reverted at the end of the group). So we must % define \Eiftex to redefine itself to be its previous value. (We can't % just define it to fail again with an ``unmatched end'' error, since % the @ifset might be nested.) % \def\conditionalsucceed#1{% \edef\temp{% % Remember the current value of \E#1. \let\nece{prevE#1} = \nece{E#1}% % % At the `@end #1', redefine \E#1 to be its previous value. \def\nece{E#1}{\let\nece{E#1} = \nece{prevE#1}}% }% \temp } % We need to expand lots of \csname's, but we don't want to expand the % control sequences after we've constructed them. % \def\nece#1{\expandafter\noexpand\csname#1\endcsname} % @asis just yields its argument. Used with @table, for example. % \def\asis#1{#1} % @math means output in math mode. % We don't use $'s directly in the definition of \math because control % sequences like \math are expanded when the toc file is written. Then, % we read the toc file back, the $'s will be normal characters (as they % should be, according to the definition of Texinfo). So we must use a % control sequence to switch into and out of math mode. % % This isn't quite enough for @math to work properly in indices, but it % seems unlikely it will ever be needed there. % \let\implicitmath = $ \def\math#1{\implicitmath #1\implicitmath} % @bullet and @minus need the same treatment as @math, just above. \def\bullet{\implicitmath\ptexbullet\implicitmath} \def\minus{\implicitmath-\implicitmath} \def\node{\ENVcheck\parsearg\nodezzz} \def\nodezzz#1{\nodexxx [#1,]} \def\nodexxx[#1,#2]{\gdef\lastnode{#1}} \let\nwnode=\node \let\lastnode=\relax \def\donoderef{\ifx\lastnode\relax\else \expandafter\expandafter\expandafter\setref{\lastnode}\fi \global\let\lastnode=\relax} \def\unnumbnoderef{\ifx\lastnode\relax\else \expandafter\expandafter\expandafter\unnumbsetref{\lastnode}\fi \global\let\lastnode=\relax} \def\appendixnoderef{\ifx\lastnode\relax\else \expandafter\expandafter\expandafter\appendixsetref{\lastnode}\fi \global\let\lastnode=\relax} \let\refill=\relax % @setfilename is done at the beginning of every texinfo file. % So open here the files we need to have open while reading the input. % This makes it possible to make a .fmt file for texinfo. \def\setfilename{% \readauxfile \opencontents \openindices \fixbackslash % Turn off hack to swallow `\input texinfo'. \global\let\setfilename=\comment % Ignore extra @setfilename cmds. \comment % Ignore the actual filename. } \outer\def\bye{\pagealignmacro\tracingstats=1\ptexend} \def\inforef #1{\inforefzzz #1,,,,**} \def\inforefzzz #1,#2,#3,#4**{\putwordSee{} \putwordInfo{} \putwordfile{} \file{\ignorespaces #3{}}, node \samp{\ignorespaces#1{}}} \message{fonts,} % Font-change commands. % Texinfo supports the sans serif font style, which plain TeX does not. % So we set up a \sf analogous to plain's \rm, etc. \newfam\sffam \def\sf{\fam=\sffam \tensf} \let\li = \sf % Sometimes we call it \li, not \sf. %% Try out Computer Modern fonts at \magstephalf \let\mainmagstep=\magstephalf % Set the font macro #1 to the font named #2, adding on the % specified font prefix (normally `cm'). \def\setfont#1#2{\font#1=\fontprefix#2} % Use cm as the default font prefix. % To specify the font prefix, you must define \fontprefix % before you read in texinfo.tex. \ifx\fontprefix\undefined \def\fontprefix{cm} \fi \ifx\bigger\relax \let\mainmagstep=\magstep1 \setfont\textrm{r12} \setfont\texttt{tt12} \else \setfont\textrm{r10 scaled \mainmagstep} \setfont\texttt{tt10 scaled \mainmagstep} \fi % Instead of cmb10, you many want to use cmbx10. % cmbx10 is a prettier font on its own, but cmb10 % looks better when embedded in a line with cmr10. \setfont\textbf{b10 scaled \mainmagstep} \setfont\textit{ti10 scaled \mainmagstep} \setfont\textsl{sl10 scaled \mainmagstep} \setfont\textsf{ss10 scaled \mainmagstep} \setfont\textsc{csc10 scaled \mainmagstep} \font\texti=cmmi10 scaled \mainmagstep \font\textsy=cmsy10 scaled \mainmagstep % A few fonts for @defun, etc. \setfont\defbf{bx10 scaled \magstep1} %was 1314 \setfont\deftt{tt10 scaled \magstep1} \def\df{\let\tentt=\deftt \let\tenbf = \defbf \bf} % Fonts for indices and small examples. % We actually use the slanted font rather than the italic, % because texinfo normally uses the slanted fonts for that. % Do not make many font distinctions in general in the index, since they % aren't very useful. \setfont\ninett{tt9} \setfont\indrm{r9} \setfont\indit{sl9} \let\indsl=\indit \let\indtt=\ninett \let\indsf=\indrm \let\indbf=\indrm \setfont\indsc{csc10 at 9pt} \font\indi=cmmi9 \font\indsy=cmsy9 % Fonts for headings \setfont\chaprm{bx12 scaled \magstep2} \setfont\chapit{ti12 scaled \magstep2} \setfont\chapsl{sl12 scaled \magstep2} \setfont\chaptt{tt12 scaled \magstep2} \setfont\chapsf{ss12 scaled \magstep2} \let\chapbf=\chaprm \setfont\chapsc{csc10 scaled\magstep3} \font\chapi=cmmi12 scaled \magstep2 \font\chapsy=cmsy10 scaled \magstep3 \setfont\secrm{bx12 scaled \magstep1} \setfont\secit{ti12 scaled \magstep1} \setfont\secsl{sl12 scaled \magstep1} \setfont\sectt{tt12 scaled \magstep1} \setfont\secsf{ss12 scaled \magstep1} \setfont\secbf{bx12 scaled \magstep1} \setfont\secsc{csc10 scaled\magstep2} \font\seci=cmmi12 scaled \magstep1 \font\secsy=cmsy10 scaled \magstep2 % \setfont\ssecrm{bx10 scaled \magstep1} % This size an font looked bad. % \setfont\ssecit{cmti10 scaled \magstep1} % The letters were too crowded. % \setfont\ssecsl{sl10 scaled \magstep1} % \setfont\ssectt{tt10 scaled \magstep1} % \setfont\ssecsf{ss10 scaled \magstep1} %\setfont\ssecrm{b10 scaled 1315} % Note the use of cmb rather than cmbx. %\setfont\ssecit{ti10 scaled 1315} % Also, the size is a little larger than %\setfont\ssecsl{sl10 scaled 1315} % being scaled magstep1. %\setfont\ssectt{tt10 scaled 1315} %\setfont\ssecsf{ss10 scaled 1315} %\let\ssecbf=\ssecrm \setfont\ssecrm{bx12 scaled \magstephalf} \setfont\ssecit{ti12 scaled \magstephalf} \setfont\ssecsl{sl12 scaled \magstephalf} \setfont\ssectt{tt12 scaled \magstephalf} \setfont\ssecsf{ss12 scaled \magstephalf} \setfont\ssecbf{bx12 scaled \magstephalf} \setfont\ssecsc{csc10 scaled \magstep1} \font\sseci=cmmi12 scaled \magstephalf \font\ssecsy=cmsy10 scaled \magstep1 % The smallcaps and symbol fonts should actually be scaled \magstep1.5, % but that is not a standard magnification. % Fonts for title page: \setfont\titlerm{bx12 scaled \magstep3} \let\authorrm = \secrm % In order for the font changes to affect most math symbols and letters, % we have to define the \textfont of the standard families. Since % texinfo doesn't allow for producing subscripts and superscripts, we % don't bother to reset \scriptfont and \scriptscriptfont (which would % also require loading a lot more fonts). % \def\resetmathfonts{% \textfont0 = \tenrm \textfont1 = \teni \textfont2 = \tensy \textfont\itfam = \tenit \textfont\slfam = \tensl \textfont\bffam = \tenbf \textfont\ttfam = \tentt \textfont\sffam = \tensf } % The font-changing commands redefine the meanings of \tenSTYLE, instead % of just \STYLE. We do this so that font changes will continue to work % in math mode, where it is the current \fam that is relevant in most % cases, not the current. Plain TeX does, for example, % \def\bf{\fam=\bffam \tenbf} By redefining \tenbf, we obviate the need % to redefine \bf itself. \def\textfonts{% \let\tenrm=\textrm \let\tenit=\textit \let\tensl=\textsl \let\tenbf=\textbf \let\tentt=\texttt \let\smallcaps=\textsc \let\tensf=\textsf \let\teni=\texti \let\tensy=\textsy \resetmathfonts} \def\chapfonts{% \let\tenrm=\chaprm \let\tenit=\chapit \let\tensl=\chapsl \let\tenbf=\chapbf \let\tentt=\chaptt \let\smallcaps=\chapsc \let\tensf=\chapsf \let\teni=\chapi \let\tensy=\chapsy \resetmathfonts} \def\secfonts{% \let\tenrm=\secrm \let\tenit=\secit \let\tensl=\secsl \let\tenbf=\secbf \let\tentt=\sectt \let\smallcaps=\secsc \let\tensf=\secsf \let\teni=\seci \let\tensy=\secsy \resetmathfonts} \def\subsecfonts{% \let\tenrm=\ssecrm \let\tenit=\ssecit \let\tensl=\ssecsl \let\tenbf=\ssecbf \let\tentt=\ssectt \let\smallcaps=\ssecsc \let\tensf=\ssecsf \let\teni=\sseci \let\tensy=\ssecsy \resetmathfonts} \def\indexfonts{% \let\tenrm=\indrm \let\tenit=\indit \let\tensl=\indsl \let\tenbf=\indbf \let\tentt=\indtt \let\smallcaps=\indsc \let\tensf=\indsf \let\teni=\indi \let\tensy=\indsy \resetmathfonts} % Set up the default fonts, so we can use them for creating boxes. % \textfonts % Count depth in font-changes, for error checks \newcount\fontdepth \fontdepth=0 % Fonts for short table of contents. \setfont\shortcontrm{r12} \setfont\shortcontbf{bx12} \setfont\shortcontsl{sl12} %% Add scribe-like font environments, plus @l for inline lisp (usually sans %% serif) and @ii for TeX italic % \smartitalic{ARG} outputs arg in italics, followed by an italic correction % unless the following character is such as not to need one. \def\smartitalicx{\ifx\next,\else\ifx\next-\else\ifx\next.\else\/\fi\fi\fi} \def\smartitalic#1{{\sl #1}\futurelet\next\smartitalicx} \let\i=\smartitalic \let\var=\smartitalic \let\dfn=\smartitalic \let\emph=\smartitalic \let\cite=\smartitalic \def\b#1{{\bf #1}} \let\strong=\b % We can't just use \exhyphenpenalty, because that only has effect at % the end of a paragraph. Restore normal hyphenation at the end of the % group within which \nohyphenation is presumably called. % \def\nohyphenation{\hyphenchar\font = -1 \aftergroup\restorehyphenation} \def\restorehyphenation{\hyphenchar\font = `- } \def\t#1{% {\tt \nohyphenation \rawbackslash \frenchspacing #1}% \null } \let\ttfont = \t %\def\samp #1{`{\tt \rawbackslash \frenchspacing #1}'\null} \def\samp #1{`\tclose{#1}'\null} \def\key #1{{\tt \nohyphenation \uppercase{#1}}\null} \def\ctrl #1{{\tt \rawbackslash \hat}#1} \let\file=\samp % @code is a modification of @t, % which makes spaces the same size as normal in the surrounding text. \def\tclose#1{% {% % Change normal interword space to be same as for the current font. \spaceskip = \fontdimen2\font % % Switch to typewriter. \tt % % But `\ ' produces the large typewriter interword space. \def\ {{\spaceskip = 0pt{} }}% % % Turn off hyphenation. \nohyphenation % \rawbackslash \frenchspacing #1% }% \null } % We *must* turn on hyphenation at `-' and `_' in \code. % Otherwise, it is too hard to avoid overful hboxes % in the Emacs manual, the Library manual, etc. % Unfortunately, TeX uses one parameter (\hyphenchar) to control % both hyphenation at - and hyphenation within words. % We must therefore turn them both off (\tclose does that) % and arrange explicitly to hyphenate an a dash. % -- rms. { \catcode`\-=\active \catcode`\_=\active \global\def\code{\begingroup \catcode`\-=\active \let-\codedash \catcode`\_=\active \let_\codeunder \codex} % The following is used by \doprintindex to insure that long function names % wrap around. It is necessary for - and _ to be active before the index is % read from the file, as \entry parses the arguments long before \code is % ever called. -- mycroft \global\def\indexbreaks{\catcode`\-=\active \let-\realdash \catcode`\_=\active \let_\realunder} } \def\realdash{-} \def\realunder{_} \def\codedash{-\discretionary{}{}{}} \def\codeunder{\normalunderscore\discretionary{}{}{}} \def\codex #1{\tclose{#1}\endgroup} %\let\exp=\tclose %Was temporary % @kbd is like @code, except that if the argument is just one @key command, % then @kbd has no effect. \def\xkey{\key} \def\kbdfoo#1#2#3\par{\def\one{#1}\def\three{#3}\def\threex{??}% \ifx\one\xkey\ifx\threex\three \key{#2}% \else\tclose{\look}\fi \else\tclose{\look}\fi} % Typeset a dimension, e.g., `in' or `pt'. The only reason for the % argument is to make the input look right: @dmn{pt} instead of % @dmn{}pt. % \def\dmn#1{\thinspace #1} \def\kbd#1{\def\look{#1}\expandafter\kbdfoo\look??\par} \def\l#1{{\li #1}\null} % \def\r#1{{\rm #1}} % roman font % Use of \lowercase was suggested. \def\sc#1{{\smallcaps#1}} % smallcaps font \def\ii#1{{\it #1}} % italic font \message{page headings,} \newskip\titlepagetopglue \titlepagetopglue = 1.5in \newskip\titlepagebottomglue \titlepagebottomglue = 2pc % First the title page. Must do @settitle before @titlepage. \def\titlefont#1{{\titlerm #1}} \newif\ifseenauthor \newif\iffinishedtitlepage \def\shorttitlepage{\parsearg\shorttitlepagezzz} \def\shorttitlepagezzz #1{\begingroup\hbox{}\vskip 1.5in \chaprm \centerline{#1}% \endgroup\page\hbox{}\page} \def\titlepage{\begingroup \parindent=0pt \textfonts \let\subtitlerm=\tenrm % I deinstalled the following change because \cmr12 is undefined. % This change was not in the ChangeLog anyway. --rms. % \let\subtitlerm=\cmr12 \def\subtitlefont{\subtitlerm \normalbaselineskip = 13pt \normalbaselines}% % \def\authorfont{\authorrm \normalbaselineskip = 16pt \normalbaselines}% % % Leave some space at the very top of the page. \vglue\titlepagetopglue % % Now you can print the title using @title. \def\title{\parsearg\titlezzz}% \def\titlezzz##1{\leftline{\titlefont{##1}} % print a rule at the page bottom also. \finishedtitlepagefalse \vskip4pt \hrule height 4pt width \hsize \vskip4pt}% % No rule at page bottom unless we print one at the top with @title. \finishedtitlepagetrue % % Now you can put text using @subtitle. \def\subtitle{\parsearg\subtitlezzz}% \def\subtitlezzz##1{{\subtitlefont \rightline{##1}}}% % % @author should come last, but may come many times. \def\author{\parsearg\authorzzz}% \def\authorzzz##1{\ifseenauthor\else\vskip 0pt plus 1filll\seenauthortrue\fi {\authorfont \leftline{##1}}}% % % Most title ``pages'' are actually two pages long, with space % at the top of the second. We don't want the ragged left on the second. \let\oldpage = \page \def\page{% \iffinishedtitlepage\else \finishtitlepage \fi \oldpage \let\page = \oldpage \hbox{}}% % \def\page{\oldpage \hbox{}} } \def\Etitlepage{% \iffinishedtitlepage\else \finishtitlepage \fi % It is important to do the page break before ending the group, % because the headline and footline are only empty inside the group. % If we use the new definition of \page, we always get a blank page % after the title page, which we certainly don't want. \oldpage \endgroup \HEADINGSon } \def\finishtitlepage{% \vskip4pt \hrule height 2pt width \hsize \vskip\titlepagebottomglue \finishedtitlepagetrue } %%% Set up page headings and footings. \let\thispage=\folio \newtoks \evenheadline % Token sequence for heading line of even pages \newtoks \oddheadline % Token sequence for heading line of odd pages \newtoks \evenfootline % Token sequence for footing line of even pages \newtoks \oddfootline % Token sequence for footing line of odd pages % Now make Tex use those variables \headline={{\textfonts\rm \ifodd\pageno \the\oddheadline \else \the\evenheadline \fi}} \footline={{\textfonts\rm \ifodd\pageno \the\oddfootline \else \the\evenfootline \fi}\HEADINGShook} \let\HEADINGShook=\relax % Commands to set those variables. % For example, this is what @headings on does % @evenheading @thistitle|@thispage|@thischapter % @oddheading @thischapter|@thispage|@thistitle % @evenfooting @thisfile|| % @oddfooting ||@thisfile \def\evenheading{\parsearg\evenheadingxxx} \def\oddheading{\parsearg\oddheadingxxx} \def\everyheading{\parsearg\everyheadingxxx} \def\evenfooting{\parsearg\evenfootingxxx} \def\oddfooting{\parsearg\oddfootingxxx} \def\everyfooting{\parsearg\everyfootingxxx} {\catcode`\@=0 % \gdef\evenheadingxxx #1{\evenheadingyyy #1@|@|@|@|\finish} \gdef\evenheadingyyy #1@|#2@|#3@|#4\finish{% \global\evenheadline={\rlap{\centerline{#2}}\line{#1\hfil#3}}} \gdef\oddheadingxxx #1{\oddheadingyyy #1@|@|@|@|\finish} \gdef\oddheadingyyy #1@|#2@|#3@|#4\finish{% \global\oddheadline={\rlap{\centerline{#2}}\line{#1\hfil#3}}} \gdef\everyheadingxxx #1{\everyheadingyyy #1@|@|@|@|\finish} \gdef\everyheadingyyy #1@|#2@|#3@|#4\finish{% \global\evenheadline={\rlap{\centerline{#2}}\line{#1\hfil#3}} \global\oddheadline={\rlap{\centerline{#2}}\line{#1\hfil#3}}} \gdef\evenfootingxxx #1{\evenfootingyyy #1@|@|@|@|\finish} \gdef\evenfootingyyy #1@|#2@|#3@|#4\finish{% \global\evenfootline={\rlap{\centerline{#2}}\line{#1\hfil#3}}} \gdef\oddfootingxxx #1{\oddfootingyyy #1@|@|@|@|\finish} \gdef\oddfootingyyy #1@|#2@|#3@|#4\finish{% \global\oddfootline={\rlap{\centerline{#2}}\line{#1\hfil#3}}} \gdef\everyfootingxxx #1{\everyfootingyyy #1@|@|@|@|\finish} \gdef\everyfootingyyy #1@|#2@|#3@|#4\finish{% \global\evenfootline={\rlap{\centerline{#2}}\line{#1\hfil#3}} \global\oddfootline={\rlap{\centerline{#2}}\line{#1\hfil#3}}} % }% unbind the catcode of @. % @headings double turns headings on for double-sided printing. % @headings single turns headings on for single-sided printing. % @headings off turns them off. % @headings on same as @headings double, retained for compatibility. % @headings after turns on double-sided headings after this page. % @headings doubleafter turns on double-sided headings after this page. % @headings singleafter turns on single-sided headings after this page. % By default, they are off. \def\headings #1 {\csname HEADINGS#1\endcsname} \def\HEADINGSoff{ \global\evenheadline={\hfil} \global\evenfootline={\hfil} \global\oddheadline={\hfil} \global\oddfootline={\hfil}} \HEADINGSoff % When we turn headings on, set the page number to 1. % For double-sided printing, put current file name in lower left corner, % chapter name on inside top of right hand pages, document % title on inside top of left hand pages, and page numbers on outside top % edge of all pages. \def\HEADINGSdouble{ %\pagealignmacro \global\pageno=1 \global\evenfootline={\hfil} \global\oddfootline={\hfil} \global\evenheadline={\line{\folio\hfil\thistitle}} \global\oddheadline={\line{\thischapter\hfil\folio}} } % For single-sided printing, chapter title goes across top left of page, % page number on top right. \def\HEADINGSsingle{ %\pagealignmacro \global\pageno=1 \global\evenfootline={\hfil} \global\oddfootline={\hfil} \global\evenheadline={\line{\thischapter\hfil\folio}} \global\oddheadline={\line{\thischapter\hfil\folio}} } \def\HEADINGSon{\HEADINGSdouble} \def\HEADINGSafter{\let\HEADINGShook=\HEADINGSdoublex} \let\HEADINGSdoubleafter=\HEADINGSafter \def\HEADINGSdoublex{% \global\evenfootline={\hfil} \global\oddfootline={\hfil} \global\evenheadline={\line{\folio\hfil\thistitle}} \global\oddheadline={\line{\thischapter\hfil\folio}} } \def\HEADINGSsingleafter{\let\HEADINGShook=\HEADINGSsinglex} \def\HEADINGSsinglex{% \global\evenfootline={\hfil} \global\oddfootline={\hfil} \global\evenheadline={\line{\thischapter\hfil\folio}} \global\oddheadline={\line{\thischapter\hfil\folio}} } % Subroutines used in generating headings % Produces Day Month Year style of output. \def\today{\number\day\space \ifcase\month\or January\or February\or March\or April\or May\or June\or July\or August\or September\or October\or November\or December\fi \space\number\year} % Use this if you want the Month Day, Year style of output. %\def\today{\ifcase\month\or %January\or February\or March\or April\or May\or June\or %July\or August\or September\or October\or November\or December\fi %\space\number\day, \number\year} % @settitle line... specifies the title of the document, for headings % It generates no output of its own \def\thistitle{No Title} \def\settitle{\parsearg\settitlezzz} \def\settitlezzz #1{\gdef\thistitle{#1}} \message{tables,} % @tabs -- simple alignment % These don't work. For one thing, \+ is defined as outer. % So these macros cannot even be defined. %\def\tabs{\parsearg\tabszzz} %\def\tabszzz #1{\settabs\+#1\cr} %\def\tabline{\parsearg\tablinezzz} %\def\tablinezzz #1{\+#1\cr} %\def\&{&} % Tables -- @table, @ftable, @vtable, @item(x), @kitem(x), @xitem(x). % default indentation of table text \newdimen\tableindent \tableindent=.8in % default indentation of @itemize and @enumerate text \newdimen\itemindent \itemindent=.3in % margin between end of table item and start of table text. \newdimen\itemmargin \itemmargin=.1in % used internally for \itemindent minus \itemmargin \newdimen\itemmax % Note @table, @vtable, and @vtable define @item, @itemx, etc., with % these defs. % They also define \itemindex % to index the item name in whatever manner is desired (perhaps none). \newif\ifitemxneedsnegativevskip \def\itemxpar{\par\ifitemxneedsnegativevskip\vskip-\parskip\nobreak\fi} \def\internalBitem{\smallbreak \parsearg\itemzzz} \def\internalBitemx{\itemxpar \parsearg\itemzzz} \def\internalBxitem "#1"{\def\xitemsubtopix{#1} \smallbreak \parsearg\xitemzzz} \def\internalBxitemx "#1"{\def\xitemsubtopix{#1} \itemxpar \parsearg\xitemzzz} \def\internalBkitem{\smallbreak \parsearg\kitemzzz} \def\internalBkitemx{\itemxpar \parsearg\kitemzzz} \def\kitemzzz #1{\dosubind {kw}{\code{#1}}{for {\bf \lastfunction}}% \itemzzz {#1}} \def\xitemzzz #1{\dosubind {kw}{\code{#1}}{for {\bf \xitemsubtopic}}% \itemzzz {#1}} \def\itemzzz #1{\begingroup % \advance\hsize by -\rightskip \advance\hsize by -\tableindent \setbox0=\hbox{\itemfont{#1}}% \itemindex{#1}% \nobreak % This prevents a break before @itemx. % % Be sure we are not still in the middle of a paragraph. %{\parskip = 0in %\par %}% % % If the item text does not fit in the space we have, put it on a line % by itself, and do not allow a page break either before or after that % line. We do not start a paragraph here because then if the next % command is, e.g., @kindex, the whatsit would get put into the % horizontal list on a line by itself, resulting in extra blank space. \ifdim \wd0>\itemmax % % Make this a paragraph so we get the \parskip glue and wrapping, % but leave it ragged-right. \begingroup \advance\leftskip by-\tableindent \advance\hsize by\tableindent \advance\rightskip by0pt plus1fil \leavevmode\unhbox0\par \endgroup % % We're going to be starting a paragraph, but we don't want the % \parskip glue -- logically it's part of the @item we just started. \nobreak \vskip-\parskip % % Stop a page break at the \parskip glue coming up. Unfortunately % we can't prevent a possible page break at the following % \baselineskip glue. \nobreak \endgroup \itemxneedsnegativevskipfalse \else % The item text fits into the space. Start a paragraph, so that the % following text (if any) will end up on the same line. Since that % text will be indented by \tableindent, we make the item text be in % a zero-width box. \noindent \rlap{\hskip -\tableindent\box0}\ignorespaces% \endgroup% \itemxneedsnegativevskiptrue% \fi } \def\item{\errmessage{@item while not in a table}} \def\itemx{\errmessage{@itemx while not in a table}} \def\kitem{\errmessage{@kitem while not in a table}} \def\kitemx{\errmessage{@kitemx while not in a table}} \def\xitem{\errmessage{@xitem while not in a table}} \def\xitemx{\errmessage{@xitemx while not in a table}} %% Contains a kludge to get @end[description] to work \def\description{\tablez{\dontindex}{1}{}{}{}{}} \def\table{\begingroup\inENV\obeylines\obeyspaces\tablex} {\obeylines\obeyspaces% \gdef\tablex #1^^M{% \tabley\dontindex#1 \endtabley}} \def\ftable{\begingroup\inENV\obeylines\obeyspaces\ftablex} {\obeylines\obeyspaces% \gdef\ftablex #1^^M{% \tabley\fnitemindex#1 \endtabley \def\Eftable{\endgraf\afterenvbreak\endgroup}% \let\Etable=\relax}} \def\vtable{\begingroup\inENV\obeylines\obeyspaces\vtablex} {\obeylines\obeyspaces% \gdef\vtablex #1^^M{% \tabley\vritemindex#1 \endtabley \def\Evtable{\endgraf\afterenvbreak\endgroup}% \let\Etable=\relax}} \def\dontindex #1{} \def\fnitemindex #1{\doind {fn}{\code{#1}}}% \def\vritemindex #1{\doind {vr}{\code{#1}}}% {\obeyspaces % \gdef\tabley#1#2 #3 #4 #5 #6 #7\endtabley{\endgroup% \tablez{#1}{#2}{#3}{#4}{#5}{#6}}} \def\tablez #1#2#3#4#5#6{% \aboveenvbreak % \begingroup % \def\Edescription{\Etable}% Neccessary kludge. \let\itemindex=#1% \ifnum 0#3>0 \advance \leftskip by #3\mil \fi % \ifnum 0#4>0 \tableindent=#4\mil \fi % \ifnum 0#5>0 \advance \rightskip by #5\mil \fi % \def\itemfont{#2}% \itemmax=\tableindent % \advance \itemmax by -\itemmargin % \advance \leftskip by \tableindent % \exdentamount=\tableindent \parindent = 0pt \parskip = \smallskipamount \ifdim \parskip=0pt \parskip=2pt \fi% \def\Etable{\endgraf\afterenvbreak\endgroup}% \let\item = \internalBitem % \let\itemx = \internalBitemx % \let\kitem = \internalBkitem % \let\kitemx = \internalBkitemx % \let\xitem = \internalBxitem % \let\xitemx = \internalBxitemx % } % This is the counter used by @enumerate, which is really @itemize \newcount \itemno \def\itemize{\parsearg\itemizezzz} \def\itemizezzz #1{% \begingroup % ended by the @end itemsize \itemizey {#1}{\Eitemize} } \def\itemizey #1#2{% \aboveenvbreak % \itemmax=\itemindent % \advance \itemmax by -\itemmargin % \advance \leftskip by \itemindent % \exdentamount=\itemindent \parindent = 0pt % \parskip = \smallskipamount % \ifdim \parskip=0pt \parskip=2pt \fi% \def#2{\endgraf\afterenvbreak\endgroup}% \def\itemcontents{#1}% \let\item=\itemizeitem} % Set sfcode to normal for the chars that usually have another value. % These are `.?!:;,' \def\frenchspacing{\sfcode46=1000 \sfcode63=1000 \sfcode33=1000 \sfcode58=1000 \sfcode59=1000 \sfcode44=1000 } % \splitoff TOKENS\endmark defines \first to be the first token in % TOKENS, and \rest to be the remainder. % \def\splitoff#1#2\endmark{\def\first{#1}\def\rest{#2}}% % Allow an optional argument of an uppercase letter, lowercase letter, % or number, to specify the first label in the enumerated list. No % argument is the same as `1'. % \def\enumerate{\parsearg\enumeratezzz} \def\enumeratezzz #1{\enumeratey #1 \endenumeratey} \def\enumeratey #1 #2\endenumeratey{% \begingroup % ended by the @end enumerate % % If we were given no argument, pretend we were given `1'. \def\thearg{#1}% \ifx\thearg\empty \def\thearg{1}\fi % % Detect if the argument is a single token. If so, it might be a % letter. Otherwise, the only valid thing it can be is a number. % (We will always have one token, because of the test we just made. % This is a good thing, since \splitoff doesn't work given nothing at % all -- the first parameter is undelimited.) \expandafter\splitoff\thearg\endmark \ifx\rest\empty % Only one token in the argument. It could still be anything. % A ``lowercase letter'' is one whose \lccode is nonzero. % An ``uppercase letter'' is one whose \lccode is both nonzero, and % not equal to itself. % Otherwise, we assume it's a number. % % We need the \relax at the end of the \ifnum lines to stop TeX from % continuing to look for a . % \ifnum\lccode\expandafter`\thearg=0\relax \numericenumerate % a number (we hope) \else % It's a letter. \ifnum\lccode\expandafter`\thearg=\expandafter`\thearg\relax \lowercaseenumerate % lowercase letter \else \uppercaseenumerate % uppercase letter \fi \fi \else % Multiple tokens in the argument. We hope it's a number. \numericenumerate \fi } % An @enumerate whose labels are integers. The starting integer is % given in \thearg. % \def\numericenumerate{% \itemno = \thearg \startenumeration{\the\itemno}% } % The starting (lowercase) letter is in \thearg. \def\lowercaseenumerate{% \itemno = \expandafter`\thearg \startenumeration{% % Be sure we're not beyond the end of the alphabet. \ifnum\itemno=0 \errmessage{No more lowercase letters in @enumerate; get a bigger alphabet}% \fi \char\lccode\itemno }% } % The starting (uppercase) letter is in \thearg. \def\uppercaseenumerate{% \itemno = \expandafter`\thearg \startenumeration{% % Be sure we're not beyond the end of the alphabet. \ifnum\itemno=0 \errmessage{No more uppercase letters in @enumerate; get a bigger alphabet} \fi \char\uccode\itemno }% } % Call itemizey, adding a period to the first argument and supplying the % common last two arguments. Also subtract one from the initial value in % \itemno, since @item increments \itemno. % \def\startenumeration#1{% \advance\itemno by -1 \itemizey{#1.}\Eenumerate\flushcr } % @alphaenumerate and @capsenumerate are abbreviations for giving an arg % to @enumerate. % \def\alphaenumerate{\enumerate{a}} \def\capsenumerate{\enumerate{A}} \def\Ealphaenumerate{\Eenumerate} \def\Ecapsenumerate{\Eenumerate} % Definition of @item while inside @itemize. \def\itemizeitem{% \advance\itemno by 1 {\let\par=\endgraf \smallbreak}% \ifhmode \errmessage{\in hmode at itemizeitem}\fi {\parskip=0in \hskip 0pt \hbox to 0pt{\hss \itemcontents\hskip \itemmargin}% \vadjust{\penalty 1200}}% \flushcr} % @multitable macros % Amy Hendrickson, 8/18/94 % % @multitable ... @endmultitable will make as many columns as desired. % Contents of each column will wrap at width given in preamble. Width % can be specified either with sample text given in a template line, % or in percent of \hsize, the current width of text on page. % Table can continue over pages but will only break between lines. % To make preamble: % % Either define widths of columns in terms of percent of \hsize: % @multitable @percentofhsize .2 .3 .5 % @item ... % % Numbers following @percentofhsize are the percent of the total % current hsize to be used for each column. You may use as many % columns as desired. % Or use a template: % @multitable {Column 1 template} {Column 2 template} {Column 3 template} % @item ... % using the widest term desired in each column. % Each new table line starts with @item, each subsequent new column % starts with @tab. Empty columns may be produced by supplying @tab's % with nothing between them for as many times as empty columns are needed, % ie, @tab@tab@tab will produce two empty columns. % @item, @tab, @multicolumn or @endmulticolumn do not need to be on their % own lines, but it will not hurt if they are. % Sample multitable: % @multitable {Column 1 template} {Column 2 template} {Column 3 template} % @item first col stuff @tab second col stuff @tab third col % @item % first col stuff % @tab % second col stuff % @tab % third col % @item first col stuff @tab second col stuff % @tab Many paragraphs of text may be used in any column. % % They will wrap at the width determined by the template. % @item@tab@tab This will be in third column. % @endmultitable % Default dimensions may be reset by user. % @intableparskip will set vertical space between paragraphs in table. % @intableparindent will set paragraph indent in table. % @spacebetweencols will set horizontal space to be left between columns. % @spacebetweenlines will set vertical space to be left between lines. %%%% % Dimensions \newdimen\intableparskip \newdimen\intableparindent \newdimen\spacebetweencols \newdimen\spacebetweenlines \intableparskip=0pt \intableparindent=6pt \spacebetweencols=12pt \spacebetweenlines=12pt %%%% % Macros used to set up halign preamble: \let\endsetuptable\relax \def\xendsetuptable{\endsetuptable} \let\percentofhsize\relax \def\xpercentofhsize{\percentofhsize} \newif\ifsetpercent \newcount\colcount \def\setuptable#1{\def\firstarg{#1}% \ifx\firstarg\xendsetuptable\let\go\relax% \else \ifx\firstarg\xpercentofhsize\global\setpercenttrue% \else \ifsetpercent \if#1.\else% \global\advance\colcount by1 % \expandafter\xdef\csname col\the\colcount\endcsname{.#1\hsize}% \fi \else \global\advance\colcount by1 \setbox0=\hbox{#1}% \expandafter\xdef\csname col\the\colcount\endcsname{\the\wd0}% \fi% \fi% \let\go\setuptable% \fi\go} %%%% % multitable syntax \def\tab{&} %%%% % @multitable ... @endmultitable definitions: \def\multitable#1\item{\bgroup \let\item\cr \tolerance=9500 \hbadness=9500 \parskip=\intableparskip \parindent=\intableparindent \overfullrule=0pt \global\colcount=0\relax% \def\Emultitable{\global\setpercentfalse\global\everycr{}\cr\egroup\egroup}% % To parse everything between @multitable and @item : \def\one{#1}\expandafter\setuptable\one\endsetuptable % Need to reset this to 0 after \setuptable. \global\colcount=0\relax% % % This preamble sets up a generic column definition, which will % be used as many times as user calls for columns. % \vtop will set a single line and will also let text wrap and % continue for many paragraphs if desired. \halign\bgroup&\global\advance\colcount by 1\relax% \vtop{\hsize=\expandafter\csname col\the\colcount\endcsname % In order to keep entries from bumping into each other % we will add a \leftskip of \spacebetweencols to all columns after % the first one. % If a template has been used, we will add \spacebetweencols % to the width of each template entry. % If user has set preamble in terms of percent of \hsize % we will use that dimension as the width of the column, and % the \leftskip will keep entries from bumping into each other. % Table will start at left margin and final column will justify at % right margin. \ifnum\colcount=1 \else \ifsetpercent \else % If user has set preamble in terms of percent of \hsize % we will advance \hsize by \spacebetweencols \advance\hsize by \spacebetweencols \fi % In either case we will make \leftskip=\spacebetweencols: \leftskip=\spacebetweencols \fi \noindent##}\cr% % \everycr will reset column counter, \colcount, at the end of % each line. Every column entry will cause \colcount to advance by one. % The table preamble % looks at the current \colcount to find the correct column width. \global\everycr{\noalign{\nointerlineskip\vskip\spacebetweenlines \filbreak%% keeps underfull box messages off when table breaks over pages. \global\colcount=0\relax}}} \message{indexing,} % Index generation facilities % Define \newwrite to be identical to plain tex's \newwrite % except not \outer, so it can be used within \newindex. {\catcode`\@=11 \gdef\newwrite{\alloc@7\write\chardef\sixt@@n}} % \newindex {foo} defines an index named foo. % It automatically defines \fooindex such that % \fooindex ...rest of line... puts an entry in the index foo. % It also defines \fooindfile to be the number of the output channel for % the file that accumulates this index. The file's extension is foo. % The name of an index should be no more than 2 characters long % for the sake of vms. \def\newindex #1{ \expandafter\newwrite \csname#1indfile\endcsname% Define number for output file \openout \csname#1indfile\endcsname \jobname.#1 % Open the file \expandafter\xdef\csname#1index\endcsname{% % Define \xxxindex \noexpand\doindex {#1}} } % @defindex foo == \newindex{foo} \def\defindex{\parsearg\newindex} % Define @defcodeindex, like @defindex except put all entries in @code. \def\newcodeindex #1{ \expandafter\newwrite \csname#1indfile\endcsname% Define number for output file \openout \csname#1indfile\endcsname \jobname.#1 % Open the file \expandafter\xdef\csname#1index\endcsname{% % Define \xxxindex \noexpand\docodeindex {#1}} } \def\defcodeindex{\parsearg\newcodeindex} % @synindex foo bar makes index foo feed into index bar. % Do this instead of @defindex foo if you don't want it as a separate index. \def\synindex #1 #2 {% \expandafter\let\expandafter\synindexfoo\expandafter=\csname#2indfile\endcsname \expandafter\let\csname#1indfile\endcsname=\synindexfoo \expandafter\xdef\csname#1index\endcsname{% % Define \xxxindex \noexpand\doindex {#2}}% } % @syncodeindex foo bar similar, but put all entries made for index foo % inside @code. \def\syncodeindex #1 #2 {% \expandafter\let\expandafter\synindexfoo\expandafter=\csname#2indfile\endcsname \expandafter\let\csname#1indfile\endcsname=\synindexfoo \expandafter\xdef\csname#1index\endcsname{% % Define \xxxindex \noexpand\docodeindex {#2}}% } % Define \doindex, the driver for all \fooindex macros. % Argument #1 is generated by the calling \fooindex macro, % and it is "foo", the name of the index. % \doindex just uses \parsearg; it calls \doind for the actual work. % This is because \doind is more useful to call from other macros. % There is also \dosubind {index}{topic}{subtopic} % which makes an entry in a two-level index such as the operation index. \def\doindex#1{\edef\indexname{#1}\parsearg\singleindexer} \def\singleindexer #1{\doind{\indexname}{#1}} % like the previous two, but they put @code around the argument. \def\docodeindex#1{\edef\indexname{#1}\parsearg\singlecodeindexer} \def\singlecodeindexer #1{\doind{\indexname}{\code{#1}}} \def\indexdummies{% % Take care of the plain tex accent commands. \def\"{\realbackslash "}% \def\`{\realbackslash `}% \def\'{\realbackslash '}% \def\^{\realbackslash ^}% \def\~{\realbackslash ~}% \def\={\realbackslash =}% \def\b{\realbackslash b}% \def\c{\realbackslash c}% \def\d{\realbackslash d}% \def\u{\realbackslash u}% \def\v{\realbackslash v}% \def\H{\realbackslash H}% % Take care of the plain tex special European modified letters. \def\oe{\realbackslash oe}% \def\ae{\realbackslash ae}% \def\aa{\realbackslash aa}% \def\OE{\realbackslash OE}% \def\AE{\realbackslash AE}% \def\AA{\realbackslash AA}% \def\o{\realbackslash o}% \def\O{\realbackslash O}% \def\l{\realbackslash l}% \def\L{\realbackslash L}% \def\ss{\realbackslash ss}% % Take care of texinfo commands likely to appear in an index entry. \def\_{{\realbackslash _}}% \def\w{\realbackslash w }% \def\bf{\realbackslash bf }% \def\rm{\realbackslash rm }% \def\sl{\realbackslash sl }% \def\sf{\realbackslash sf}% \def\tt{\realbackslash tt}% \def\gtr{\realbackslash gtr}% \def\less{\realbackslash less}% \def\hat{\realbackslash hat}% \def\char{\realbackslash char}% \def\TeX{\realbackslash TeX}% \def\dots{\realbackslash dots }% \def\copyright{\realbackslash copyright }% \def\tclose##1{\realbackslash tclose {##1}}% \def\code##1{\realbackslash code {##1}}% \def\samp##1{\realbackslash samp {##1}}% \def\t##1{\realbackslash r {##1}}% \def\r##1{\realbackslash r {##1}}% \def\i##1{\realbackslash i {##1}}% \def\b##1{\realbackslash b {##1}}% \def\cite##1{\realbackslash cite {##1}}% \def\key##1{\realbackslash key {##1}}% \def\file##1{\realbackslash file {##1}}% \def\var##1{\realbackslash var {##1}}% \def\kbd##1{\realbackslash kbd {##1}}% \def\dfn##1{\realbackslash dfn {##1}}% \def\emph##1{\realbackslash emph {##1}}% } % \indexnofonts no-ops all font-change commands. % This is used when outputting the strings to sort the index by. \def\indexdummyfont#1{#1} \def\indexdummytex{TeX} \def\indexdummydots{...} \def\indexnofonts{% % Just ignore accents. \let\"=\indexdummyfont \let\`=\indexdummyfont \let\'=\indexdummyfont \let\^=\indexdummyfont \let\~=\indexdummyfont \let\==\indexdummyfont \let\b=\indexdummyfont \let\c=\indexdummyfont \let\d=\indexdummyfont \let\u=\indexdummyfont \let\v=\indexdummyfont \let\H=\indexdummyfont % Take care of the plain tex special European modified letters. \def\oe{oe}% \def\ae{ae}% \def\aa{aa}% \def\OE{OE}% \def\AE{AE}% \def\AA{AA}% \def\o{o}% \def\O{O}% \def\l{l}% \def\L{L}% \def\ss{ss}% \let\w=\indexdummyfont \let\t=\indexdummyfont \let\r=\indexdummyfont \let\i=\indexdummyfont \let\b=\indexdummyfont \let\emph=\indexdummyfont \let\strong=\indexdummyfont \let\cite=\indexdummyfont \let\sc=\indexdummyfont %Don't no-op \tt, since it isn't a user-level command % and is used in the definitions of the active chars like <, >, |... %\let\tt=\indexdummyfont \let\tclose=\indexdummyfont \let\code=\indexdummyfont \let\file=\indexdummyfont \let\samp=\indexdummyfont \let\kbd=\indexdummyfont \let\key=\indexdummyfont \let\var=\indexdummyfont \let\TeX=\indexdummytex \let\dots=\indexdummydots } % To define \realbackslash, we must make \ not be an escape. % We must first make another character (@) an escape % so we do not become unable to do a definition. {\catcode`\@=0 \catcode`\\=\other @gdef@realbackslash{\}} \let\indexbackslash=0 %overridden during \printindex. \let\SETmarginindex=\relax %initialize! % workhorse for all \fooindexes % #1 is name of index, #2 is stuff to put there \def\doind #1#2{% % Put the index entry in the margin if desired. \ifx\SETmarginindex\relax\else% \insert\margin{\hbox{\vrule height8pt depth3pt width0pt #2}}% \fi% {\count10=\lastpenalty % {\indexdummies % Must do this here, since \bf, etc expand at this stage \escapechar=`\\% {\let\folio=0% Expand all macros now EXCEPT \folio \def\rawbackslashxx{\indexbackslash}% \indexbackslash isn't defined now % so it will be output as is; and it will print as backslash in the indx. % % Now process the index-string once, with all font commands turned off, % to get the string to sort the index by. {\indexnofonts \xdef\temp1{#2}% }% % Now produce the complete index entry. We process the index-string again, % this time with font commands expanded, to get what to print in the index. \edef\temp{% \write \csname#1indfile\endcsname{% \realbackslash entry {\temp1}{\folio}{#2}}}% \temp }% }\penalty\count10}} \def\dosubind #1#2#3{% {\count10=\lastpenalty % {\indexdummies % Must do this here, since \bf, etc expand at this stage \escapechar=`\\% {\let\folio=0% \def\rawbackslashxx{\indexbackslash}% % % Now process the index-string once, with all font commands turned off, % to get the string to sort the index by. {\indexnofonts \xdef\temp1{#2 #3}% }% % Now produce the complete index entry. We process the index-string again, % this time with font commands expanded, to get what to print in the index. \edef\temp{% \write \csname#1indfile\endcsname{% \realbackslash entry {\temp1}{\folio}{#2}{#3}}}% \temp }% }\penalty\count10}} % The index entry written in the file actually looks like % \entry {sortstring}{page}{topic} % or % \entry {sortstring}{page}{topic}{subtopic} % The texindex program reads in these files and writes files % containing these kinds of lines: % \initial {c} % before the first topic whose initial is c % \entry {topic}{pagelist} % for a topic that is used without subtopics % \primary {topic} % for the beginning of a topic that is used with subtopics % \secondary {subtopic}{pagelist} % for each subtopic. % Define the user-accessible indexing commands % @findex, @vindex, @kindex, @cindex. \def\findex {\fnindex} \def\kindex {\kyindex} \def\cindex {\cpindex} \def\vindex {\vrindex} \def\tindex {\tpindex} \def\pindex {\pgindex} \def\cindexsub {\begingroup\obeylines\cindexsub} {\obeylines % \gdef\cindexsub "#1" #2^^M{\endgroup % \dosubind{cp}{#2}{#1}}} % Define the macros used in formatting output of the sorted index material. % This is what you call to cause a particular index to get printed. % Write % @unnumbered Function Index % @printindex fn \def\printindex{\parsearg\doprintindex} \def\doprintindex#1{% \tex \dobreak \chapheadingskip {10000} \catcode`\%=\other\catcode`\&=\other\catcode`\#=\other \catcode`\$=\other \catcode`\~=\other \indexbreaks % % The following don't help, since the chars were translated % when the raw index was written, and their fonts were discarded % due to \indexnofonts. %\catcode`\"=\active %\catcode`\^=\active %\catcode`\_=\active %\catcode`\|=\active %\catcode`\<=\active %\catcode`\>=\active % % \def\indexbackslash{\rawbackslashxx} \indexfonts\rm \tolerance=9500 \advance\baselineskip -1pt \begindoublecolumns % % See if the index file exists and is nonempty. \openin 1 \jobname.#1s \ifeof 1 % \enddoublecolumns gets confused if there is no text in the index, % and it loses the chapter title and the aux file entries for the % index. The easiest way to prevent this problem is to make sure % there is some text. (Index is nonexistent) \else % % If the index file exists but is empty, then \openin leaves \ifeof % false. We have to make TeX try to read something from the file, so % it can discover if there is anything in it. \read 1 to \temp \ifeof 1 (Index is empty) \else \input \jobname.#1s \fi \fi \closein 1 \enddoublecolumns \Etex } % These macros are used by the sorted index file itself. % Change them to control the appearance of the index. % Same as \bigskipamount except no shrink. % \balancecolumns gets confused if there is any shrink. \newskip\initialskipamount \initialskipamount 12pt plus4pt \def\initial #1{% {\let\tentt=\sectt \let\tt=\sectt \let\sf=\sectt \ifdim\lastskip<\initialskipamount \removelastskip \penalty-200 \vskip \initialskipamount\fi \line{\secbf#1\hfill}\kern 2pt\penalty10000}} % This typesets a paragraph consisting of #1, dot leaders, and then #2 % flush to the right margin. It is used for index and table of contents % entries. The paragraph is indented by \leftskip. % \def\entry #1#2{\begingroup % % Start a new paragraph if necessary, so our assignments below can't % affect previous text. \par % % Do not fill out the last line with white space. \parfillskip = 0in % % No extra space above this paragraph. \parskip = 0in % % Do not prefer a separate line ending with a hyphen to fewer lines. \finalhyphendemerits = 0 % % \hangindent is only relevant when the entry text and page number % don't both fit on one line. In that case, bob suggests starting the % dots pretty far over on the line. Unfortunately, a large % indentation looks wrong when the entry text itself is broken across % lines. So we use a small indentation and put up with long leaders. % % \hangafter is reset to 1 (which is the value we want) at the start % of each paragraph, so we need not do anything with that. \hangindent=2em % % When the entry text needs to be broken, just fill out the first line % with blank space. \rightskip = 0pt plus1fil % % Start a ``paragraph'' for the index entry so the line breaking % parameters we've set above will have an effect. \noindent % % Insert the text of the index entry. TeX will do line-breaking on it. #1% % The following is kluged to not output a line of dots in the index if % there are no page numbers. The next person who breaks this will be % cursed by a Unix daemon. \def\tempa{{\rm }}% \def\tempb{#2}% \edef\tempc{\tempa}% \edef\tempd{\tempb}% \ifx\tempc\tempd\ \else% % % If we must, put the page number on a line of its own, and fill out % this line with blank space. (The \hfil is overwhelmed with the % fill leaders glue in \indexdotfill if the page number does fit.) \hfil\penalty50 \null\nobreak\indexdotfill % Have leaders before the page number. % % The `\ ' here is removed by the implicit \unskip that TeX does as % part of (the primitive) \par. Without it, a spurious underfull % \hbox ensues. \ #2% The page number ends the paragraph. \fi% \par \endgroup} % Like \dotfill except takes at least 1 em. \def\indexdotfill{\cleaders \hbox{$\mathsurround=0pt \mkern1.5mu ${\it .}$ \mkern1.5mu$}\hskip 1em plus 1fill} \def\primary #1{\line{#1\hfil}} \newskip\secondaryindent \secondaryindent=0.5cm \def\secondary #1#2{ {\parfillskip=0in \parskip=0in \hangindent =1in \hangafter=1 \noindent\hskip\secondaryindent\hbox{#1}\indexdotfill #2\par }} %% Define two-column mode, which is used in indexes. %% Adapted from the TeXbook, page 416. \catcode `\@=11 \newbox\partialpage \newdimen\doublecolumnhsize \def\begindoublecolumns{\begingroup % Grab any single-column material above us. \output = {\global\setbox\partialpage =\vbox{\unvbox255\kern -\topskip \kern \baselineskip}}% \eject % % Now switch to the double-column output routine. \output={\doublecolumnout}% % % Change the page size parameters. We could do this once outside this % routine, in each of @smallbook, @afourpaper, and the default 8.5x11 % format, but then we repeat the same computation. Repeating a couple % of assignments once per index is clearly meaningless for the % execution time, so we may as well do it once. % % First we halve the line length, less a little for the gutter between % the columns. We compute the gutter based on the line length, so it % changes automatically with the paper format. The magic constant % below is chosen so that the gutter has the same value (well, +- < % 1pt) as it did when we hard-coded it. % % We put the result in a separate register, \doublecolumhsize, so we % can restore it in \pagesofar, after \hsize itself has (potentially) % been clobbered. % \doublecolumnhsize = \hsize \advance\doublecolumnhsize by -.04154\hsize \divide\doublecolumnhsize by 2 \hsize = \doublecolumnhsize % % Double the \vsize as well. (We don't need a separate register here, % since nobody clobbers \vsize.) \vsize = 2\vsize \doublecolumnpagegoal } \def\enddoublecolumns{\eject \endgroup \pagegoal=\vsize \unvbox\partialpage} \def\doublecolumnsplit{\splittopskip=\topskip \splitmaxdepth=\maxdepth \global\dimen@=\pageheight \global\advance\dimen@ by-\ht\partialpage \global\setbox1=\vsplit255 to\dimen@ \global\setbox0=\vbox{\unvbox1} \global\setbox3=\vsplit255 to\dimen@ \global\setbox2=\vbox{\unvbox3} \ifdim\ht0>\dimen@ \setbox255=\vbox{\unvbox0\unvbox2} \global\setbox255=\copy5 \fi \ifdim\ht2>\dimen@ \setbox255=\vbox{\unvbox0\unvbox2} \global\setbox255=\copy5 \fi } \def\doublecolumnpagegoal{% \dimen@=\vsize \advance\dimen@ by-2\ht\partialpage \global\pagegoal=\dimen@ } \def\pagesofar{\unvbox\partialpage % \hsize=\doublecolumnhsize % have to restore this since output routine \wd0=\hsize \wd2=\hsize \hbox to\pagewidth{\box0\hfil\box2}} \def\doublecolumnout{% \setbox5=\copy255 {\vbadness=10000 \doublecolumnsplit} \ifvbox255 \setbox0=\vtop to\dimen@{\unvbox0} \setbox2=\vtop to\dimen@{\unvbox2} \onepageout\pagesofar \unvbox255 \penalty\outputpenalty \else \setbox0=\vbox{\unvbox5} \ifvbox0 \dimen@=\ht0 \advance\dimen@ by\topskip \advance\dimen@ by-\baselineskip \divide\dimen@ by2 \splittopskip=\topskip \splitmaxdepth=\maxdepth {\vbadness=10000 \loop \global\setbox5=\copy0 \setbox1=\vsplit5 to\dimen@ \setbox3=\vsplit5 to\dimen@ \ifvbox5 \global\advance\dimen@ by1pt \repeat \setbox0=\vbox to\dimen@{\unvbox1} \setbox2=\vbox to\dimen@{\unvbox3} \global\setbox\partialpage=\vbox{\pagesofar} \doublecolumnpagegoal } \fi \fi } \catcode `\@=\other \message{sectioning,} % Define chapters, sections, etc. \newcount \chapno \newcount \secno \secno=0 \newcount \subsecno \subsecno=0 \newcount \subsubsecno \subsubsecno=0 % This counter is funny since it counts through charcodes of letters A, B, ... \newcount \appendixno \appendixno = `\@ \def\appendixletter{\char\the\appendixno} \newwrite \contentsfile % This is called from \setfilename. \def\opencontents{\openout \contentsfile = \jobname.toc} % Each @chapter defines this as the name of the chapter. % page headings and footings can use it. @section does likewise \def\thischapter{} \def\thissection{} \def\seccheck#1{\if \pageno<0 % \errmessage{@#1 not allowed after generating table of contents}\fi % } \def\chapternofonts{% \let\rawbackslash=\relax% \let\frenchspacing=\relax% \def\result{\realbackslash result} \def\equiv{\realbackslash equiv} \def\expansion{\realbackslash expansion} \def\print{\realbackslash print} \def\TeX{\realbackslash TeX} \def\dots{\realbackslash dots} \def\copyright{\realbackslash copyright} \def\tt{\realbackslash tt} \def\bf{\realbackslash bf } \def\w{\realbackslash w} \def\less{\realbackslash less} \def\gtr{\realbackslash gtr} \def\hat{\realbackslash hat} \def\char{\realbackslash char} \def\tclose##1{\realbackslash tclose {##1}} \def\code##1{\realbackslash code {##1}} \def\samp##1{\realbackslash samp {##1}} \def\r##1{\realbackslash r {##1}} \def\b##1{\realbackslash b {##1}} \def\key##1{\realbackslash key {##1}} \def\file##1{\realbackslash file {##1}} \def\kbd##1{\realbackslash kbd {##1}} % These are redefined because @smartitalic wouldn't work inside xdef. \def\i##1{\realbackslash i {##1}} \def\cite##1{\realbackslash cite {##1}} \def\var##1{\realbackslash var {##1}} \def\emph##1{\realbackslash emph {##1}} \def\dfn##1{\realbackslash dfn {##1}} } \newcount\absseclevel % used to calculate proper heading level \newcount\secbase\secbase=0 % @raise/lowersections modify this count % @raisesections: treat @section as chapter, @subsection as section, etc. \def\raisesections{\global\advance\secbase by -1} \let\up=\raisesections % original BFox name % @lowersections: treat @chapter as section, @section as subsection, etc. \def\lowersections{\global\advance\secbase by 1} \let\down=\lowersections % original BFox name % Choose a numbered-heading macro % #1 is heading level if unmodified by @raisesections or @lowersections % #2 is text for heading \def\numhead#1#2{\absseclevel=\secbase\advance\absseclevel by #1 \ifcase\absseclevel \chapterzzz{#2} \or \seczzz{#2} \or \numberedsubseczzz{#2} \or \numberedsubsubseczzz{#2} \else \ifnum \absseclevel<0 \chapterzzz{#2} \else \numberedsubsubseczzz{#2} \fi \fi } % like \numhead, but chooses appendix heading levels \def\apphead#1#2{\absseclevel=\secbase\advance\absseclevel by #1 \ifcase\absseclevel \appendixzzz{#2} \or \appendixsectionzzz{#2} \or \appendixsubseczzz{#2} \or \appendixsubsubseczzz{#2} \else \ifnum \absseclevel<0 \appendixzzz{#2} \else \appendixsubsubseczzz{#2} \fi \fi } % like \numhead, but chooses numberless heading levels \def\unnmhead#1#2{\absseclevel=\secbase\advance\absseclevel by #1 \ifcase\absseclevel \unnumberedzzz{#2} \or \unnumberedseczzz{#2} \or \unnumberedsubseczzz{#2} \or \unnumberedsubsubseczzz{#2} \else \ifnum \absseclevel<0 \unnumberedzzz{#2} \else \unnumberedsubsubseczzz{#2} \fi \fi } \def\thischaptername{No Chapter Title} \outer\def\chapter{\parsearg\chapteryyy} \def\chapteryyy #1{\numhead0{#1}} % normally numhead0 calls chapterzzz \def\chapterzzz #1{\seccheck{chapter}% \secno=0 \subsecno=0 \subsubsecno=0 \global\advance \chapno by 1 \message{\putwordChapter \the\chapno}% \chapmacro {#1}{\the\chapno}% \gdef\thissection{#1}% \gdef\thischaptername{#1}% % We don't substitute the actual chapter name into \thischapter % because we don't want its macros evaluated now. \xdef\thischapter{\putwordChapter{} \the\chapno: \noexpand\thischaptername}% {\chapternofonts% \edef\temp{{\realbackslash chapentry {#1}{\the\chapno}{\noexpand\folio}}}% \escapechar=`\\% \write \contentsfile \temp % \donoderef % \global\let\section = \numberedsec \global\let\subsection = \numberedsubsec \global\let\subsubsection = \numberedsubsubsec }} \outer\def\appendix{\parsearg\appendixyyy} \def\appendixyyy #1{\apphead0{#1}} % normally apphead0 calls appendixzzz \def\appendixzzz #1{\seccheck{appendix}% \secno=0 \subsecno=0 \subsubsecno=0 \global\advance \appendixno by 1 \message{Appendix \appendixletter}% \chapmacro {#1}{\putwordAppendix{} \appendixletter}% \gdef\thissection{#1}% \gdef\thischaptername{#1}% \xdef\thischapter{\putwordAppendix{} \appendixletter: \noexpand\thischaptername}% {\chapternofonts% \edef\temp{{\realbackslash chapentry {#1}{\putwordAppendix{} \appendixletter}{\noexpand\folio}}}% \escapechar=`\\% \write \contentsfile \temp % \appendixnoderef % \global\let\section = \appendixsec \global\let\subsection = \appendixsubsec \global\let\subsubsection = \appendixsubsubsec }} \outer\def\top{\parsearg\unnumberedyyy} \outer\def\unnumbered{\parsearg\unnumberedyyy} \def\unnumberedyyy #1{\unnmhead0{#1}} % normally unnmhead0 calls unnumberedzzz \def\unnumberedzzz #1{\seccheck{unnumbered}% \secno=0 \subsecno=0 \subsubsecno=0 % % This used to be simply \message{#1}, but TeX fully expands the % argument to \message. Therefore, if #1 contained @-commands, TeX % expanded them. For example, in `@unnumbered The @cite{Book}', TeX % expanded @cite (which turns out to cause errors because \cite is meant % to be executed, not expanded). % % Anyway, we don't want the fully-expanded definition of @cite to appear % as a result of the \message, we just want `@cite' itself. We use % \the to achieve this: TeX expands \the only once, % simply yielding the contents of the . \toks0 = {#1}\message{(\the\toks0)}% % \unnumbchapmacro {#1}% \gdef\thischapter{#1}\gdef\thissection{#1}% {\chapternofonts% \edef\temp{{\realbackslash unnumbchapentry {#1}{\noexpand\folio}}}% \escapechar=`\\% \write \contentsfile \temp % \unnumbnoderef % \global\let\section = \unnumberedsec \global\let\subsection = \unnumberedsubsec \global\let\subsubsection = \unnumberedsubsubsec }} \outer\def\numberedsec{\parsearg\secyyy} \def\secyyy #1{\numhead1{#1}} % normally calls seczzz \def\seczzz #1{\seccheck{section}% \subsecno=0 \subsubsecno=0 \global\advance \secno by 1 % \gdef\thissection{#1}\secheading {#1}{\the\chapno}{\the\secno}% {\chapternofonts% \edef\temp{{\realbackslash secentry % {#1}{\the\chapno}{\the\secno}{\noexpand\folio}}}% \escapechar=`\\% \write \contentsfile \temp % \donoderef % \penalty 10000 % }} \outer\def\appenixsection{\parsearg\appendixsecyyy} \outer\def\appendixsec{\parsearg\appendixsecyyy} \def\appendixsecyyy #1{\apphead1{#1}} % normally calls appendixsectionzzz \def\appendixsectionzzz #1{\seccheck{appendixsection}% \subsecno=0 \subsubsecno=0 \global\advance \secno by 1 % \gdef\thissection{#1}\secheading {#1}{\appendixletter}{\the\secno}% {\chapternofonts% \edef\temp{{\realbackslash secentry % {#1}{\appendixletter}{\the\secno}{\noexpand\folio}}}% \escapechar=`\\% \write \contentsfile \temp % \appendixnoderef % \penalty 10000 % }} \outer\def\unnumberedsec{\parsearg\unnumberedsecyyy} \def\unnumberedsecyyy #1{\unnmhead1{#1}} % normally calls unnumberedseczzz \def\unnumberedseczzz #1{\seccheck{unnumberedsec}% \plainsecheading {#1}\gdef\thissection{#1}% {\chapternofonts% \edef\temp{{\realbackslash unnumbsecentry{#1}{\noexpand\folio}}}% \escapechar=`\\% \write \contentsfile \temp % \unnumbnoderef % \penalty 10000 % }} \outer\def\numberedsubsec{\parsearg\numberedsubsecyyy} \def\numberedsubsecyyy #1{\numhead2{#1}} % normally calls numberedsubseczzz \def\numberedsubseczzz #1{\seccheck{subsection}% \gdef\thissection{#1}\subsubsecno=0 \global\advance \subsecno by 1 % \subsecheading {#1}{\the\chapno}{\the\secno}{\the\subsecno}% {\chapternofonts% \edef\temp{{\realbackslash subsecentry % {#1}{\the\chapno}{\the\secno}{\the\subsecno}{\noexpand\folio}}}% \escapechar=`\\% \write \contentsfile \temp % \donoderef % \penalty 10000 % }} \outer\def\appendixsubsec{\parsearg\appendixsubsecyyy} \def\appendixsubsecyyy #1{\apphead2{#1}} % normally calls appendixsubseczzz \def\appendixsubseczzz #1{\seccheck{appendixsubsec}% \gdef\thissection{#1}\subsubsecno=0 \global\advance \subsecno by 1 % \subsecheading {#1}{\appendixletter}{\the\secno}{\the\subsecno}% {\chapternofonts% \edef\temp{{\realbackslash subsecentry % {#1}{\appendixletter}{\the\secno}{\the\subsecno}{\noexpand\folio}}}% \escapechar=`\\% \write \contentsfile \temp % \appendixnoderef % \penalty 10000 % }} \outer\def\unnumberedsubsec{\parsearg\unnumberedsubsecyyy} \def\unnumberedsubsecyyy #1{\unnmhead2{#1}} %normally calls unnumberedsubseczzz \def\unnumberedsubseczzz #1{\seccheck{unnumberedsubsec}% \plainsecheading {#1}\gdef\thissection{#1}% {\chapternofonts% \edef\temp{{\realbackslash unnumbsubsecentry{#1}{\noexpand\folio}}}% \escapechar=`\\% \write \contentsfile \temp % \unnumbnoderef % \penalty 10000 % }} \outer\def\numberedsubsubsec{\parsearg\numberedsubsubsecyyy} \def\numberedsubsubsecyyy #1{\numhead3{#1}} % normally numberedsubsubseczzz \def\numberedsubsubseczzz #1{\seccheck{subsubsection}% \gdef\thissection{#1}\global\advance \subsubsecno by 1 % \subsubsecheading {#1} {\the\chapno}{\the\secno}{\the\subsecno}{\the\subsubsecno}% {\chapternofonts% \edef\temp{{\realbackslash subsubsecentry % {#1} {\the\chapno}{\the\secno}{\the\subsecno}{\the\subsubsecno} {\noexpand\folio}}}% \escapechar=`\\% \write \contentsfile \temp % \donoderef % \penalty 10000 % }} \outer\def\appendixsubsubsec{\parsearg\appendixsubsubsecyyy} \def\appendixsubsubsecyyy #1{\apphead3{#1}} % normally appendixsubsubseczzz \def\appendixsubsubseczzz #1{\seccheck{appendixsubsubsec}% \gdef\thissection{#1}\global\advance \subsubsecno by 1 % \subsubsecheading {#1} {\appendixletter}{\the\secno}{\the\subsecno}{\the\subsubsecno}% {\chapternofonts% \edef\temp{{\realbackslash subsubsecentry{#1}% {\appendixletter} {\the\secno}{\the\subsecno}{\the\subsubsecno}{\noexpand\folio}}}% \escapechar=`\\% \write \contentsfile \temp % \appendixnoderef % \penalty 10000 % }} \outer\def\unnumberedsubsubsec{\parsearg\unnumberedsubsubsecyyy} \def\unnumberedsubsubsecyyy #1{\unnmhead3{#1}} %normally unnumberedsubsubseczzz \def\unnumberedsubsubseczzz #1{\seccheck{unnumberedsubsubsec}% \plainsecheading {#1}\gdef\thissection{#1}% {\chapternofonts% \edef\temp{{\realbackslash unnumbsubsubsecentry{#1}{\noexpand\folio}}}% \escapechar=`\\% \write \contentsfile \temp % \unnumbnoderef % \penalty 10000 % }} % These are variants which are not "outer", so they can appear in @ifinfo. % Actually, they should now be obsolete; ordinary section commands should work. \def\infotop{\parsearg\unnumberedzzz} \def\infounnumbered{\parsearg\unnumberedzzz} \def\infounnumberedsec{\parsearg\unnumberedseczzz} \def\infounnumberedsubsec{\parsearg\unnumberedsubseczzz} \def\infounnumberedsubsubsec{\parsearg\unnumberedsubsubseczzz} \def\infoappendix{\parsearg\appendixzzz} \def\infoappendixsec{\parsearg\appendixseczzz} \def\infoappendixsubsec{\parsearg\appendixsubseczzz} \def\infoappendixsubsubsec{\parsearg\appendixsubsubseczzz} \def\infochapter{\parsearg\chapterzzz} \def\infosection{\parsearg\sectionzzz} \def\infosubsection{\parsearg\subsectionzzz} \def\infosubsubsection{\parsearg\subsubsectionzzz} % These macros control what the section commands do, according % to what kind of chapter we are in (ordinary, appendix, or unnumbered). % Define them by default for a numbered chapter. \global\let\section = \numberedsec \global\let\subsection = \numberedsubsec \global\let\subsubsection = \numberedsubsubsec % Define @majorheading, @heading and @subheading % NOTE on use of \vbox for chapter headings, section headings, and % such: % 1) We use \vbox rather than the earlier \line to permit % overlong headings to fold. % 2) \hyphenpenalty is set to 10000 because hyphenation in a % heading is obnoxious; this forbids it. % 3) Likewise, headings look best if no \parindent is used, and % if justification is not attempted. Hence \raggedright. \def\majorheading{\parsearg\majorheadingzzz} \def\majorheadingzzz #1{% {\advance\chapheadingskip by 10pt \chapbreak }% {\chapfonts \vbox{\hyphenpenalty=10000\tolerance=5000 \parindent=0pt\raggedright \rm #1\hfill}}\bigskip \par\penalty 200} \def\chapheading{\parsearg\chapheadingzzz} \def\chapheadingzzz #1{\chapbreak % {\chapfonts \vbox{\hyphenpenalty=10000\tolerance=5000 \parindent=0pt\raggedright \rm #1\hfill}}\bigskip \par\penalty 200} \def\heading{\parsearg\secheadingi} \def\subheading{\parsearg\subsecheadingi} \def\subsubheading{\parsearg\subsubsecheadingi} % These macros generate a chapter, section, etc. heading only % (including whitespace, linebreaking, etc. around it), % given all the information in convenient, parsed form. %%% Args are the skip and penalty (usually negative) \def\dobreak#1#2{\par\ifdim\lastskip<#1\removelastskip\penalty#2\vskip#1\fi} \def\setchapterstyle #1 {\csname CHAPF#1\endcsname} %%% Define plain chapter starts, and page on/off switching for it % Parameter controlling skip before chapter headings (if needed) \newskip \chapheadingskip \chapheadingskip = 30pt plus 8pt minus 4pt \def\chapbreak{\dobreak \chapheadingskip {-4000}} \def\chappager{\par\vfill\supereject} \def\chapoddpage{\chappager \ifodd\pageno \else \hbox to 0pt{} \chappager\fi} \def\setchapternewpage #1 {\csname CHAPPAG#1\endcsname} \def\CHAPPAGoff{ \global\let\pchapsepmacro=\chapbreak \global\let\pagealignmacro=\chappager} \def\CHAPPAGon{ \global\let\pchapsepmacro=\chappager \global\let\pagealignmacro=\chappager \global\def\HEADINGSon{\HEADINGSsingle}} \def\CHAPPAGodd{ \global\let\pchapsepmacro=\chapoddpage \global\let\pagealignmacro=\chapoddpage \global\def\HEADINGSon{\HEADINGSdouble}} \CHAPPAGon \def\CHAPFplain{ \global\let\chapmacro=\chfplain \global\let\unnumbchapmacro=\unnchfplain} \def\chfplain #1#2{% \pchapsepmacro {% \chapfonts \vbox{\hyphenpenalty=10000\tolerance=5000 \parindent=0pt\raggedright \rm #2\enspace #1}% }% \bigskip \penalty5000 } \def\unnchfplain #1{% \pchapsepmacro % {\chapfonts \vbox{\hyphenpenalty=10000\tolerance=5000 \parindent=0pt\raggedright \rm #1\hfill}}\bigskip \par\penalty 10000 % } \CHAPFplain % The default \def\unnchfopen #1{% \chapoddpage {\chapfonts \vbox{\hyphenpenalty=10000\tolerance=5000 \parindent=0pt\raggedright \rm #1\hfill}}\bigskip \par\penalty 10000 % } \def\chfopen #1#2{\chapoddpage {\chapfonts \vbox to 3in{\vfil \hbox to\hsize{\hfil #2} \hbox to\hsize{\hfil #1} \vfil}}% \par\penalty 5000 % } \def\CHAPFopen{ \global\let\chapmacro=\chfopen \global\let\unnumbchapmacro=\unnchfopen} % Parameter controlling skip before section headings. \newskip \subsecheadingskip \subsecheadingskip = 17pt plus 8pt minus 4pt \def\subsecheadingbreak{\dobreak \subsecheadingskip {-500}} \newskip \secheadingskip \secheadingskip = 21pt plus 8pt minus 4pt \def\secheadingbreak{\dobreak \secheadingskip {-1000}} % @paragraphindent is defined for the Info formatting commands only. \let\paragraphindent=\comment % Section fonts are the base font at magstep2, which produces % a size a bit more than 14 points in the default situation. \def\secheading #1#2#3{\secheadingi {#2.#3\enspace #1}} \def\plainsecheading #1{\secheadingi {#1}} \def\secheadingi #1{{\advance \secheadingskip by \parskip % \secheadingbreak}% {\secfonts \vbox{\hyphenpenalty=10000\tolerance=5000 \parindent=0pt\raggedright \rm #1\hfill}}% \ifdim \parskip<10pt \kern 10pt\kern -\parskip\fi \penalty 10000 } % Subsection fonts are the base font at magstep1, % which produces a size of 12 points. \def\subsecheading #1#2#3#4{\subsecheadingi {#2.#3.#4\enspace #1}} \def\subsecheadingi #1{{\advance \subsecheadingskip by \parskip % \subsecheadingbreak}% {\subsecfonts \vbox{\hyphenpenalty=10000\tolerance=5000 \parindent=0pt\raggedright \rm #1\hfill}}% \ifdim \parskip<10pt \kern 10pt\kern -\parskip\fi \penalty 10000 } \def\subsubsecfonts{\subsecfonts} % Maybe this should change: % Perhaps make sssec fonts scaled % magstep half \def\subsubsecheading #1#2#3#4#5{\subsubsecheadingi {#2.#3.#4.#5\enspace #1}} \def\subsubsecheadingi #1{{\advance \subsecheadingskip by \parskip % \subsecheadingbreak}% {\subsubsecfonts \vbox{\hyphenpenalty=10000\tolerance=5000 \parindent=0pt\raggedright \rm #1\hfill}}% \ifdim \parskip<10pt \kern 10pt\kern -\parskip\fi \penalty 10000} \message{toc printing,} % Finish up the main text and prepare to read what we've written % to \contentsfile. \newskip\contentsrightmargin \contentsrightmargin=1in \def\startcontents#1{% \pagealignmacro \immediate\closeout \contentsfile \ifnum \pageno>0 \pageno = -1 % Request roman numbered pages. \fi % Don't need to put `Contents' or `Short Contents' in the headline. % It is abundantly clear what they are. \unnumbchapmacro{#1}\def\thischapter{}% \begingroup % Set up to handle contents files properly. \catcode`\\=0 \catcode`\{=1 \catcode`\}=2 \catcode`\@=11 \catcode`\^=7 % to see ^^e4 as \"a etc. juha@piuha.ydi.vtt.fi \raggedbottom % Worry more about breakpoints than the bottom. \advance\hsize by -\contentsrightmargin % Don't use the full line length. } % Normal (long) toc. \outer\def\contents{% \startcontents{\putwordTableofContents}% \input \jobname.toc \endgroup \vfill \eject } % And just the chapters. \outer\def\summarycontents{% \startcontents{\putwordShortContents}% % \let\chapentry = \shortchapentry \let\unnumbchapentry = \shortunnumberedentry % We want a true roman here for the page numbers. \secfonts \let\rm=\shortcontrm \let\bf=\shortcontbf \let\sl=\shortcontsl \rm \advance\baselineskip by 1pt % Open it up a little. \def\secentry ##1##2##3##4{} \def\unnumbsecentry ##1##2{} \def\subsecentry ##1##2##3##4##5{} \def\unnumbsubsecentry ##1##2{} \def\subsubsecentry ##1##2##3##4##5##6{} \def\unnumbsubsubsecentry ##1##2{} \input \jobname.toc \endgroup \vfill \eject } \let\shortcontents = \summarycontents % These macros generate individual entries in the table of contents. % The first argument is the chapter or section name. % The last argument is the page number. % The arguments in between are the chapter number, section number, ... % Chapter-level things, for both the long and short contents. \def\chapentry#1#2#3{\dochapentry{#2\labelspace#1}{#3}} % See comments in \dochapentry re vbox and related settings \def\shortchapentry#1#2#3{% \tocentry{\shortchaplabel{#2}\labelspace #1}{\doshortpageno{#3}}% } % Typeset the label for a chapter or appendix for the short contents. % The arg is, e.g. `Appendix A' for an appendix, or `3' for a chapter. % We could simplify the code here by writing out an \appendixentry % command in the toc file for appendices, instead of using \chapentry % for both, but it doesn't seem worth it. \setbox0 = \hbox{\shortcontrm \putwordAppendix } \newdimen\shortappendixwidth \shortappendixwidth = \wd0 \def\shortchaplabel#1{% % We typeset #1 in a box of constant width, regardless of the text of % #1, so the chapter titles will come out aligned. \setbox0 = \hbox{#1}% \dimen0 = \ifdim\wd0 > \shortappendixwidth \shortappendixwidth \else 0pt \fi % % This space should be plenty, since a single number is .5em, and the % widest letter (M) is 1em, at least in the Computer Modern fonts. % (This space doesn't include the extra space that gets added after % the label; that gets put in in \shortchapentry above.) \advance\dimen0 by 1.1em \hbox to \dimen0{#1\hfil}% } \def\unnumbchapentry#1#2{\dochapentry{#1}{#2}} \def\shortunnumberedentry#1#2{\tocentry{#1}{\doshortpageno{#2}}} % Sections. \def\secentry#1#2#3#4{\dosecentry{#2.#3\labelspace#1}{#4}} \def\unnumbsecentry#1#2{\dosecentry{#1}{#2}} % Subsections. \def\subsecentry#1#2#3#4#5{\dosubsecentry{#2.#3.#4\labelspace#1}{#5}} \def\unnumbsubsecentry#1#2{\dosubsecentry{#1}{#2}} % And subsubsections. \def\subsubsecentry#1#2#3#4#5#6{% \dosubsubsecentry{#2.#3.#4.#5\labelspace#1}{#6}} \def\unnumbsubsubsecentry#1#2{\dosubsubsecentry{#1}{#2}} % This parameter controls the indentation of the various levels. \newdimen\tocindent \tocindent = 3pc % Now for the actual typesetting. In all these, #1 is the text and #2 is the % page number. % % If the toc has to be broken over pages, we would want to be at chapters % if at all possible; hence the \penalty. \def\dochapentry#1#2{% \penalty-300 \vskip\baselineskip \begingroup \chapentryfonts \tocentry{#1}{\dopageno{#2}}% \endgroup \nobreak\vskip .25\baselineskip } \def\dosecentry#1#2{\begingroup \secentryfonts \leftskip=\tocindent \tocentry{#1}{\dopageno{#2}}% \endgroup} \def\dosubsecentry#1#2{\begingroup \subsecentryfonts \leftskip=2\tocindent \tocentry{#1}{\dopageno{#2}}% \endgroup} \def\dosubsubsecentry#1#2{\begingroup \subsubsecentryfonts \leftskip=3\tocindent \tocentry{#1}{\dopageno{#2}}% \endgroup} % Final typesetting of a toc entry; we use the same \entry macro as for % the index entries, but we want to suppress hyphenation here. (We % can't do that in the \entry macro, since index entries might consist % of hyphenated-identifiers-that-do-not-fit-on-a-line-and-nothing-else.) % \def\tocentry#1#2{\begingroup \hyphenpenalty = 10000 \entry{#1}{#2}% \endgroup} % Space between chapter (or whatever) number and the title. \def\labelspace{\hskip1em \relax} \def\dopageno#1{{\rm #1}} \def\doshortpageno#1{{\rm #1}} \def\chapentryfonts{\secfonts \rm} \def\secentryfonts{\textfonts} \let\subsecentryfonts = \textfonts \let\subsubsecentryfonts = \textfonts \message{environments,} % Since these characters are used in examples, it should be an even number of % \tt widths. Each \tt character is 1en, so two makes it 1em. % Furthermore, these definitions must come after we define our fonts. \newbox\dblarrowbox \newbox\longdblarrowbox \newbox\pushcharbox \newbox\bullbox \newbox\equivbox \newbox\errorbox \let\ptexequiv = \equiv %{\tentt %\global\setbox\dblarrowbox = \hbox to 1em{\hfil$\Rightarrow$\hfil} %\global\setbox\longdblarrowbox = \hbox to 1em{\hfil$\mapsto$\hfil} %\global\setbox\pushcharbox = \hbox to 1em{\hfil$\dashv$\hfil} %\global\setbox\equivbox = \hbox to 1em{\hfil$\ptexequiv$\hfil} % Adapted from the manmac format (p.420 of TeXbook) %\global\setbox\bullbox = \hbox to 1em{\kern.15em\vrule height .75ex width .85ex % depth .1ex\hfil} %} \def\point{$\star$} \def\result{\leavevmode\raise.15ex\hbox to 1em{\hfil$\Rightarrow$\hfil}} \def\expansion{\leavevmode\raise.1ex\hbox to 1em{\hfil$\mapsto$\hfil}} \def\print{\leavevmode\lower.1ex\hbox to 1em{\hfil$\dashv$\hfil}} \def\equiv{\leavevmode\lower.1ex\hbox to 1em{\hfil$\ptexequiv$\hfil}} % Adapted from the TeXbook's \boxit. {\tentt \global\dimen0 = 3em}% Width of the box. \dimen2 = .55pt % Thickness of rules % The text. (`r' is open on the right, `e' somewhat less so on the left.) \setbox0 = \hbox{\kern-.75pt \tensf error\kern-1.5pt} \global\setbox\errorbox=\hbox to \dimen0{\hfil \hsize = \dimen0 \advance\hsize by -5.8pt % Space to left+right. \advance\hsize by -2\dimen2 % Rules. \vbox{ \hrule height\dimen2 \hbox{\vrule width\dimen2 \kern3pt % Space to left of text. \vtop{\kern2.4pt \box0 \kern2.4pt}% Space above/below. \kern3pt\vrule width\dimen2}% Space to right. \hrule height\dimen2} \hfil} % The @error{} command. \def\error{\leavevmode\lower.7ex\copy\errorbox} % @tex ... @end tex escapes into raw Tex temporarily. % One exception: @ is still an escape character, so that @end tex works. % But \@ or @@ will get a plain tex @ character. \def\tex{\begingroup \catcode `\\=0 \catcode `\{=1 \catcode `\}=2 \catcode `\$=3 \catcode `\&=4 \catcode `\#=6 \catcode `\^=7 \catcode `\_=8 \catcode `\~=13 \let~=\tie \catcode `\%=14 \catcode 43=12 \catcode`\"=12 \catcode`\==12 \catcode`\|=12 \catcode`\<=12 \catcode`\>=12 \escapechar=`\\ % \let\~=\ptextilde \let\{=\ptexlbrace \let\}=\ptexrbrace \let\.=\ptexdot \let\*=\ptexstar \let\dots=\ptexdots \def\@{@}% \let\bullet=\ptexbullet \let\b=\ptexb \let\c=\ptexc \let\i=\ptexi \let\t=\ptext \let\l=\ptexl \let\L=\ptexL % \let\Etex=\endgroup} % Define @lisp ... @endlisp. % @lisp does a \begingroup so it can rebind things, % including the definition of @endlisp (which normally is erroneous). % Amount to narrow the margins by for @lisp. \newskip\lispnarrowing \lispnarrowing=0.4in % This is the definition that ^^M gets inside @lisp, @example, and other % such environments. \null is better than a space, since it doesn't % have any width. \def\lisppar{\null\endgraf} % Make each space character in the input produce a normal interword % space in the output. Don't allow a line break at this space, as this % is used only in environments like @example, where each line of input % should produce a line of output anyway. % {\obeyspaces % \gdef\sepspaces{\obeyspaces\let =\tie}} % Define \obeyedspace to be our active space, whatever it is. This is % for use in \parsearg. {\sepspaces% \global\let\obeyedspace= } % This space is always present above and below environments. \newskip\envskipamount \envskipamount = 0pt % Make spacing and below environment symmetrical. We use \parskip here % to help in doing that, since in @example-like environments \parskip % is reset to zero; thus the \afterenvbreak inserts no space -- but the % start of the next paragraph will insert \parskip % \def\aboveenvbreak{{\advance\envskipamount by \parskip \endgraf \ifdim\lastskip<\envskipamount \removelastskip \penalty-50 \vskip\envskipamount \fi}} \let\afterenvbreak = \aboveenvbreak % \nonarrowing is a flag. If "set", @lisp etc don't narrow margins. \let\nonarrowing=\relax %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % \cartouche: draw rectangle w/rounded corners around argument \font\circle=lcircle10 \newdimen\circthick \newdimen\cartouter\newdimen\cartinner \newskip\normbskip\newskip\normpskip\newskip\normlskip \circthick=\fontdimen8\circle % \def\ctl{{\circle\char'013\hskip -6pt}}% 6pt from pl file: 1/2charwidth \def\ctr{{\hskip 6pt\circle\char'010}} \def\cbl{{\circle\char'012\hskip -6pt}} \def\cbr{{\hskip 6pt\circle\char'011}} \def\carttop{\hbox to \cartouter{\hskip\lskip \ctl\leaders\hrule height\circthick\hfil\ctr \hskip\rskip}} \def\cartbot{\hbox to \cartouter{\hskip\lskip \cbl\leaders\hrule height\circthick\hfil\cbr \hskip\rskip}} % \newskip\lskip\newskip\rskip \long\def\cartouche{% \begingroup \lskip=\leftskip \rskip=\rightskip \leftskip=0pt\rightskip=0pt %we want these *outside*. \cartinner=\hsize \advance\cartinner by-\lskip \advance\cartinner by-\rskip \cartouter=\hsize \advance\cartouter by 18pt % allow for 3pt kerns on either % side, and for 6pt waste from % each corner char \normbskip=\baselineskip \normpskip=\parskip \normlskip=\lineskip % Flag to tell @lisp, etc., not to narrow margin. \let\nonarrowing=\comment \vbox\bgroup \baselineskip=0pt\parskip=0pt\lineskip=0pt \carttop \hbox\bgroup \hskip\lskip \vrule\kern3pt \vbox\bgroup \hsize=\cartinner \kern3pt \begingroup \baselineskip=\normbskip \lineskip=\normlskip \parskip=\normpskip \vskip -\parskip \def\Ecartouche{% \endgroup \kern3pt \egroup \kern3pt\vrule \hskip\rskip \egroup \cartbot \egroup \endgroup }} % This macro is called at the beginning of all the @example variants, % inside a group. \def\nonfillstart{% \aboveenvbreak \inENV % This group ends at the end of the body \hfuzz = 12pt % Don't be fussy \sepspaces % Make spaces be word-separators rather than space tokens. \singlespace \let\par = \lisppar % don't ignore blank lines \obeylines % each line of input is a line of output \parskip = 0pt \parindent = 0pt \emergencystretch = 0pt % don't try to avoid overfull boxes % @cartouche defines \nonarrowing to inhibit narrowing % at next level down. \ifx\nonarrowing\relax \advance \leftskip by \lispnarrowing \exdentamount=\lispnarrowing \let\exdent=\nofillexdent \let\nonarrowing=\relax \fi } % To ending an @example-like environment, we first end the paragraph % (via \afterenvbreak's vertical glue), and then the group. That way we % keep the zero \parskip that the environments set -- \parskip glue % will be inserted at the beginning of the next paragraph in the % document, after the environment. % \def\nonfillfinish{\afterenvbreak\endgroup}% % This macro is \def\lisp{\begingroup \nonfillstart \let\Elisp = \nonfillfinish \tt \rawbackslash % have \ input char produce \ char from current font \gobble } % Define the \E... control sequence only if we are inside the % environment, so the error checking in \end will work. % % We must call \lisp last in the definition, since it reads the % return following the @example (or whatever) command. % \def\example{\begingroup \def\Eexample{\nonfillfinish\endgroup}\lisp} \def\smallexample{\begingroup \def\Esmallexample{\nonfillfinish\endgroup}\lisp} \def\smalllisp{\begingroup \def\Esmalllisp{\nonfillfinish\endgroup}\lisp} % @smallexample and @smalllisp. This is not used unless the @smallbook % command is given. Originally contributed by Pavel@xerox. % \def\smalllispx{\begingroup \nonfillstart \let\Esmalllisp = \nonfillfinish \let\Esmallexample = \nonfillfinish % % Smaller interline space and fonts for small examples. \setleading{10pt}% \indexfonts \tt \rawbackslash % make \ output the \ character from the current font (tt) \gobble } % This is @display; same as @lisp except use roman font. % \def\display{\begingroup \nonfillstart \let\Edisplay = \nonfillfinish \gobble } % This is @format; same as @display except don't narrow margins. % \def\format{\begingroup \let\nonarrowing = t \nonfillstart \let\Eformat = \nonfillfinish \gobble } % @flushleft (same as @format) and @flushright. % \def\flushleft{\begingroup \let\nonarrowing = t \nonfillstart \let\Eflushleft = \nonfillfinish \gobble } \def\flushright{\begingroup \let\nonarrowing = t \nonfillstart \let\Eflushright = \nonfillfinish \advance\leftskip by 0pt plus 1fill \gobble} % @quotation does normal linebreaking (hence we can't use \nonfillstart) % and narrows the margins. % \def\quotation{% \begingroup\inENV %This group ends at the end of the @quotation body {\parskip=0pt \aboveenvbreak}% because \aboveenvbreak inserts \parskip \singlespace \parindent=0pt % We have retained a nonzero parskip for the environment, since we're % doing normal filling. So to avoid extra space below the environment... \def\Equotation{\parskip = 0pt \nonfillfinish}% % % @cartouche defines \nonarrowing to inhibit narrowing at next level down. \ifx\nonarrowing\relax \advance\leftskip by \lispnarrowing \advance\rightskip by \lispnarrowing \exdentamount = \lispnarrowing \let\nonarrowing = \relax \fi } \message{defuns,} % Define formatter for defuns % First, allow user to change definition object font (\df) internally \def\setdeffont #1 {\csname DEF#1\endcsname} \newskip\defbodyindent \defbodyindent=.4in \newskip\defargsindent \defargsindent=50pt \newskip\deftypemargin \deftypemargin=12pt \newskip\deflastargmargin \deflastargmargin=18pt \newcount\parencount % define \functionparens, which makes ( and ) and & do special things. % \functionparens affects the group it is contained in. \def\activeparens{% \catcode`\(=\active \catcode`\)=\active \catcode`\&=\active \catcode`\[=\active \catcode`\]=\active} % Make control sequences which act like normal parenthesis chars. \let\lparen = ( \let\rparen = ) {\activeparens % Now, smart parens don't turn on until &foo (see \amprm) % Be sure that we always have a definition for `(', etc. For example, % if the fn name has parens in it, \boldbrax will not be in effect yet, % so TeX would otherwise complain about undefined control sequence. \global\let(=\lparen \global\let)=\rparen \global\let[=\lbrack \global\let]=\rbrack \gdef\functionparens{\boldbrax\let&=\amprm\parencount=0 } \gdef\boldbrax{\let(=\opnr\let)=\clnr\let[=\lbrb\let]=\rbrb} % This is used to turn on special parens % but make & act ordinary (given that it's active). \gdef\boldbraxnoamp{\let(=\opnr\let)=\clnr\let[=\lbrb\let]=\rbrb\let&=\ampnr} % Definitions of (, ) and & used in args for functions. % This is the definition of ( outside of all parentheses. \gdef\oprm#1 {{\rm\char`\(}#1 \bf \let(=\opnested % \global\advance\parencount by 1 } % % This is the definition of ( when already inside a level of parens. \gdef\opnested{\char`\(\global\advance\parencount by 1 } % \gdef\clrm{% Print a paren in roman if it is taking us back to depth of 0. % also in that case restore the outer-level definition of (. \ifnum \parencount=1 {\rm \char `\)}\sl \let(=\oprm \else \char `\) \fi \global\advance \parencount by -1 } % If we encounter &foo, then turn on ()-hacking afterwards \gdef\amprm#1 {{\rm\}\let(=\oprm \let)=\clrm\ } % \gdef\normalparens{\boldbrax\let&=\ampnr} } % End of definition inside \activeparens %% These parens (in \boldbrax) actually are a little bolder than the %% contained text. This is especially needed for [ and ] \def\opnr{{\sf\char`\(}} \def\clnr{{\sf\char`\)}} \def\ampnr{\&} \def\lbrb{{\bf\char`\[}} \def\rbrb{{\bf\char`\]}} % First, defname, which formats the header line itself. % #1 should be the function name. % #2 should be the type of definition, such as "Function". \def\defname #1#2{% % Get the values of \leftskip and \rightskip as they were % outside the @def... \dimen2=\leftskip \advance\dimen2 by -\defbodyindent \dimen3=\rightskip \advance\dimen3 by -\defbodyindent \noindent % \setbox0=\hbox{\hskip \deflastargmargin{\rm #2}\hskip \deftypemargin}% \dimen0=\hsize \advance \dimen0 by -\wd0 % compute size for first line \dimen1=\hsize \advance \dimen1 by -\defargsindent %size for continuations \parshape 2 0in \dimen0 \defargsindent \dimen1 % % Now output arg 2 ("Function" or some such) % ending at \deftypemargin from the right margin, % but stuck inside a box of width 0 so it does not interfere with linebreaking {% Adjust \hsize to exclude the ambient margins, % so that \rightline will obey them. \advance \hsize by -\dimen2 \advance \hsize by -\dimen3 \rlap{\rightline{{\rm #2}\hskip \deftypemargin}}}% % Make all lines underfull and no complaints: \tolerance=10000 \hbadness=10000 \advance\leftskip by -\defbodyindent \exdentamount=\defbodyindent {\df #1}\enskip % Generate function name } % Actually process the body of a definition % #1 should be the terminating control sequence, such as \Edefun. % #2 should be the "another name" control sequence, such as \defunx. % #3 should be the control sequence that actually processes the header, % such as \defunheader. \def\defparsebody #1#2#3{\begingroup\inENV% Environment for definitionbody \medbreak % % Define the end token that this defining construct specifies % so that it will exit this group. \def#1{\endgraf\endgroup\medbreak}% \def#2{\begingroup\obeylines\activeparens\spacesplit#3}% \parindent=0in \advance\leftskip by \defbodyindent \advance \rightskip by \defbodyindent \exdentamount=\defbodyindent \begingroup % \catcode 61=\active % 61 is `=' \obeylines\activeparens\spacesplit#3} \def\defmethparsebody #1#2#3#4 {\begingroup\inENV % \medbreak % % Define the end token that this defining construct specifies % so that it will exit this group. \def#1{\endgraf\endgroup\medbreak}% \def#2##1 {\begingroup\obeylines\activeparens\spacesplit{#3{##1}}}% \parindent=0in \advance\leftskip by \defbodyindent \advance \rightskip by \defbodyindent \exdentamount=\defbodyindent \begingroup\obeylines\activeparens\spacesplit{#3{#4}}} \def\defopparsebody #1#2#3#4#5 {\begingroup\inENV % \medbreak % % Define the end token that this defining construct specifies % so that it will exit this group. \def#1{\endgraf\endgroup\medbreak}% \def#2##1 ##2 {\def#4{##1}% \begingroup\obeylines\activeparens\spacesplit{#3{##2}}}% \parindent=0in \advance\leftskip by \defbodyindent \advance \rightskip by \defbodyindent \exdentamount=\defbodyindent \begingroup\obeylines\activeparens\spacesplit{#3{#5}}} % These parsing functions are similar to the preceding ones % except that they do not make parens into active characters. % These are used for "variables" since they have no arguments. \def\defvarparsebody #1#2#3{\begingroup\inENV% Environment for definitionbody \medbreak % % Define the end token that this defining construct specifies % so that it will exit this group. \def#1{\endgraf\endgroup\medbreak}% \def#2{\begingroup\obeylines\spacesplit#3}% \parindent=0in \advance\leftskip by \defbodyindent \advance \rightskip by \defbodyindent \exdentamount=\defbodyindent \begingroup % \catcode 61=\active % \obeylines\spacesplit#3} % This is used for \def{tp,vr}parsebody. It could probably be used for % some of the others, too, with some judicious conditionals. % \def\parsebodycommon#1#2#3{% \begingroup\inENV % \medbreak % % Define the end token that this defining construct specifies % so that it will exit this group. \def#1{\endgraf\endgroup\medbreak}% \def#2##1 {\begingroup\obeylines\spacesplit{#3{##1}}}% \parindent=0in \advance\leftskip by \defbodyindent \advance \rightskip by \defbodyindent \exdentamount=\defbodyindent \begingroup\obeylines } \def\defvrparsebody#1#2#3#4 {% \parsebodycommon{#1}{#2}{#3}% \spacesplit{#3{#4}}% } % This loses on `@deftp {Data Type} {struct termios}' -- it thinks the % type is just `struct', because we lose the braces in `{struct % termios}' when \spacesplit reads its undelimited argument. Sigh. % \let\deftpparsebody=\defvrparsebody % % So, to get around this, we put \empty in with the type name. That % way, TeX won't find exactly `{...}' as an undelimited argument, and % won't strip off the braces. % \def\deftpparsebody #1#2#3#4 {% \parsebodycommon{#1}{#2}{#3}% \spacesplit{\parsetpheaderline{#3{#4}}}\empty } % Fine, but then we have to eventually remove the \empty *and* the % braces (if any). That's what this does, putting the result in \tptemp. % \def\removeemptybraces\empty#1\relax{\def\tptemp{#1}}% % After \spacesplit has done its work, this is called -- #1 is the final % thing to call, #2 the type name (which starts with \empty), and #3 % (which might be empty) the arguments. % \def\parsetpheaderline#1#2#3{% \removeemptybraces#2\relax #1{\tptemp}{#3}% }% \def\defopvarparsebody #1#2#3#4#5 {\begingroup\inENV % \medbreak % % Define the end token that this defining construct specifies % so that it will exit this group. \def#1{\endgraf\endgroup\medbreak}% \def#2##1 ##2 {\def#4{##1}% \begingroup\obeylines\spacesplit{#3{##2}}}% \parindent=0in \advance\leftskip by \defbodyindent \advance \rightskip by \defbodyindent \exdentamount=\defbodyindent \begingroup\obeylines\spacesplit{#3{#5}}} % Split up #2 at the first space token. % call #1 with two arguments: % the first is all of #2 before the space token, % the second is all of #2 after that space token. % If #2 contains no space token, all of it is passed as the first arg % and the second is passed as empty. {\obeylines \gdef\spacesplit#1#2^^M{\endgroup\spacesplitfoo{#1}#2 \relax\spacesplitfoo}% \long\gdef\spacesplitfoo#1#2 #3#4\spacesplitfoo{% \ifx\relax #3% #1{#2}{}\else #1{#2}{#3#4}\fi}} % So much for the things common to all kinds of definitions. % Define @defun. % First, define the processing that is wanted for arguments of \defun % Use this to expand the args and terminate the paragraph they make up \def\defunargs #1{\functionparens \sl % Expand, preventing hyphenation at `-' chars. % Note that groups don't affect changes in \hyphenchar. \hyphenchar\tensl=0 #1% \hyphenchar\tensl=45 \ifnum\parencount=0 \else \errmessage{unbalanced parens in @def arguments}\fi% \interlinepenalty=10000 \advance\rightskip by 0pt plus 1fil \endgraf\penalty 10000\vskip -\parskip\penalty 10000% } \def\deftypefunargs #1{% % Expand, preventing hyphenation at `-' chars. % Note that groups don't affect changes in \hyphenchar. % Use \boldbraxnoamp, not \functionparens, so that & is not special. \boldbraxnoamp \tclose{#1}% avoid \code because of side effects on active chars \interlinepenalty=10000 \advance\rightskip by 0pt plus 1fil \endgraf\penalty 10000\vskip -\parskip\penalty 10000% } % Do complete processing of one @defun or @defunx line already parsed. % @deffn Command forward-char nchars \def\deffn{\defmethparsebody\Edeffn\deffnx\deffnheader} \def\deffnheader #1#2#3{\doind {fn}{\code{#2}}% \begingroup\defname {#2}{#1}\defunargs{#3}\endgroup % \catcode 61=\other % Turn off change made in \defparsebody } % @defun == @deffn Function \def\defun{\defparsebody\Edefun\defunx\defunheader} \def\defunheader #1#2{\doind {fn}{\code{#1}}% Make entry in function index \begingroup\defname {#1}{Function}% \defunargs {#2}\endgroup % \catcode 61=\other % Turn off change made in \defparsebody } % @deftypefun int foobar (int @var{foo}, float @var{bar}) \def\deftypefun{\defparsebody\Edeftypefun\deftypefunx\deftypefunheader} % #1 is the data type. #2 is the name and args. \def\deftypefunheader #1#2{\deftypefunheaderx{#1}#2 \relax} % #1 is the data type, #2 the name, #3 the args. \def\deftypefunheaderx #1#2 #3\relax{% \doind {fn}{\code{#2}}% Make entry in function index \begingroup\defname {\defheaderxcond#1\relax$$$#2}{Function}% \deftypefunargs {#3}\endgroup % \catcode 61=\other % Turn off change made in \defparsebody } % @deftypefn {Library Function} int foobar (int @var{foo}, float @var{bar}) \def\deftypefn{\defmethparsebody\Edeftypefn\deftypefnx\deftypefnheader} % \defheaderxcond#1\relax$$$ % puts #1 in @code, followed by a space, but does nothing if #1 is null. \def\defheaderxcond#1#2$$${\ifx#1\relax\else\code{#1#2} \fi} % #1 is the classification. #2 is the data type. #3 is the name and args. \def\deftypefnheader #1#2#3{\deftypefnheaderx{#1}{#2}#3 \relax} % #1 is the classification, #2 the data type, #3 the name, #4 the args. \def\deftypefnheaderx #1#2#3 #4\relax{% \doind {fn}{\code{#3}}% Make entry in function index \begingroup \normalparens % notably, turn off `&' magic, which prevents % at least some C++ text from working \defname {\defheaderxcond#2\relax$$$#3}{#1}% \deftypefunargs {#4}\endgroup % \catcode 61=\other % Turn off change made in \defparsebody } % @defmac == @deffn Macro \def\defmac{\defparsebody\Edefmac\defmacx\defmacheader} \def\defmacheader #1#2{\doind {fn}{\code{#1}}% Make entry in function index \begingroup\defname {#1}{Macro}% \defunargs {#2}\endgroup % \catcode 61=\other % Turn off change made in \defparsebody } % @defspec == @deffn Special Form \def\defspec{\defparsebody\Edefspec\defspecx\defspecheader} \def\defspecheader #1#2{\doind {fn}{\code{#1}}% Make entry in function index \begingroup\defname {#1}{Special Form}% \defunargs {#2}\endgroup % \catcode 61=\other % Turn off change made in \defparsebody } % This definition is run if you use @defunx % anywhere other than immediately after a @defun or @defunx. \def\deffnx #1 {\errmessage{@deffnx in invalid context}} \def\defunx #1 {\errmessage{@defunx in invalid context}} \def\defmacx #1 {\errmessage{@defmacx in invalid context}} \def\defspecx #1 {\errmessage{@defspecx in invalid context}} \def\deftypefnx #1 {\errmessage{@deftypefnx in invalid context}} \def\deftypeunx #1 {\errmessage{@deftypeunx in invalid context}} % @defmethod, and so on % @defop {Funny Method} foo-class frobnicate argument \def\defop #1 {\def\defoptype{#1}% \defopparsebody\Edefop\defopx\defopheader\defoptype} \def\defopheader #1#2#3{% \dosubind {fn}{\code{#2}}{on #1}% Make entry in function index \begingroup\defname {#2}{\defoptype{} on #1}% \defunargs {#3}\endgroup % } % @defmethod == @defop Method \def\defmethod{\defmethparsebody\Edefmethod\defmethodx\defmethodheader} \def\defmethodheader #1#2#3{% \dosubind {fn}{\code{#2}}{on #1}% entry in function index \begingroup\defname {#2}{Method on #1}% \defunargs {#3}\endgroup % } % @defcv {Class Option} foo-class foo-flag \def\defcv #1 {\def\defcvtype{#1}% \defopvarparsebody\Edefcv\defcvx\defcvarheader\defcvtype} \def\defcvarheader #1#2#3{% \dosubind {vr}{\code{#2}}{of #1}% Make entry in var index \begingroup\defname {#2}{\defcvtype{} of #1}% \defvarargs {#3}\endgroup % } % @defivar == @defcv {Instance Variable} \def\defivar{\defvrparsebody\Edefivar\defivarx\defivarheader} \def\defivarheader #1#2#3{% \dosubind {vr}{\code{#2}}{of #1}% Make entry in var index \begingroup\defname {#2}{Instance Variable of #1}% \defvarargs {#3}\endgroup % } % These definitions are run if you use @defmethodx, etc., % anywhere other than immediately after a @defmethod, etc. \def\defopx #1 {\errmessage{@defopx in invalid context}} \def\defmethodx #1 {\errmessage{@defmethodx in invalid context}} \def\defcvx #1 {\errmessage{@defcvx in invalid context}} \def\defivarx #1 {\errmessage{@defivarx in invalid context}} % Now @defvar % First, define the processing that is wanted for arguments of @defvar. % This is actually simple: just print them in roman. % This must expand the args and terminate the paragraph they make up \def\defvarargs #1{\normalparens #1% \interlinepenalty=10000 \endgraf\penalty 10000\vskip -\parskip\penalty 10000} % @defvr Counter foo-count \def\defvr{\defvrparsebody\Edefvr\defvrx\defvrheader} \def\defvrheader #1#2#3{\doind {vr}{\code{#2}}% \begingroup\defname {#2}{#1}\defvarargs{#3}\endgroup} % @defvar == @defvr Variable \def\defvar{\defvarparsebody\Edefvar\defvarx\defvarheader} \def\defvarheader #1#2{\doind {vr}{\code{#1}}% Make entry in var index \begingroup\defname {#1}{Variable}% \defvarargs {#2}\endgroup % } % @defopt == @defvr {User Option} \def\defopt{\defvarparsebody\Edefopt\defoptx\defoptheader} \def\defoptheader #1#2{\doind {vr}{\code{#1}}% Make entry in var index \begingroup\defname {#1}{User Option}% \defvarargs {#2}\endgroup % } % @deftypevar int foobar \def\deftypevar{\defvarparsebody\Edeftypevar\deftypevarx\deftypevarheader} % #1 is the data type. #2 is the name. \def\deftypevarheader #1#2{% \doind {vr}{\code{#2}}% Make entry in variables index \begingroup\defname {\defheaderxcond#1\relax$$$#2}{Variable}% \interlinepenalty=10000 \endgraf\penalty 10000\vskip -\parskip\penalty 10000 \endgroup} % @deftypevr {Global Flag} int enable \def\deftypevr{\defvrparsebody\Edeftypevr\deftypevrx\deftypevrheader} \def\deftypevrheader #1#2#3{\doind {vr}{\code{#3}}% \begingroup\defname {\defheaderxcond#2\relax$$$#3}{#1} \interlinepenalty=10000 \endgraf\penalty 10000\vskip -\parskip\penalty 10000 \endgroup} % This definition is run if you use @defvarx % anywhere other than immediately after a @defvar or @defvarx. \def\defvrx #1 {\errmessage{@defvrx in invalid context}} \def\defvarx #1 {\errmessage{@defvarx in invalid context}} \def\defoptx #1 {\errmessage{@defoptx in invalid context}} \def\deftypevarx #1 {\errmessage{@deftypevarx in invalid context}} \def\deftypevrx #1 {\errmessage{@deftypevrx in invalid context}} % Now define @deftp % Args are printed in bold, a slight difference from @defvar. \def\deftpargs #1{\bf \defvarargs{#1}} % @deftp Class window height width ... \def\deftp{\deftpparsebody\Edeftp\deftpx\deftpheader} \def\deftpheader #1#2#3{\doind {tp}{\code{#2}}% \begingroup\defname {#2}{#1}\deftpargs{#3}\endgroup} % This definition is run if you use @deftpx, etc % anywhere other than immediately after a @deftp, etc. \def\deftpx #1 {\errmessage{@deftpx in invalid context}} \message{cross reference,} % Define cross-reference macros \newwrite \auxfile \newif\ifhavexrefs % True if xref values are known. \newif\ifwarnedxrefs % True if we warned once that they aren't known. % \setref{foo} defines a cross-reference point named foo. \def\setref#1{% \dosetq{#1-title}{Ytitle}% \dosetq{#1-pg}{Ypagenumber}% \dosetq{#1-snt}{Ysectionnumberandtype}} \def\unnumbsetref#1{% \dosetq{#1-title}{Ytitle}% \dosetq{#1-pg}{Ypagenumber}% \dosetq{#1-snt}{Ynothing}} \def\appendixsetref#1{% \dosetq{#1-title}{Ytitle}% \dosetq{#1-pg}{Ypagenumber}% \dosetq{#1-snt}{Yappendixletterandtype}} % \xref, \pxref, and \ref generate cross-references to specified points. % For \xrefX, #1 is the node name, #2 the name of the Info % cross-reference, #3 the printed node name, #4 the name of the Info % file, #5 the name of the printed manual. All but the node name can be % omitted. % \def\pxref#1{\putwordsee{} \xrefX[#1,,,,,,,]} \def\xref#1{\putwordSee{} \xrefX[#1,,,,,,,]} \def\ref#1{\xrefX[#1,,,,,,,]} \def\xrefX[#1,#2,#3,#4,#5,#6]{\begingroup \def\printedmanual{\ignorespaces #5}% \def\printednodename{\ignorespaces #3}% \setbox1=\hbox{\printedmanual}% \setbox0=\hbox{\printednodename}% \ifdim \wd0 = 0pt % No printed node name was explicitly given. \ifx\SETxref-automatic-section-title\relax % % Use the actual chapter/section title appear inside % the square brackets. Use the real section title if we have it. \ifdim \wd1>0pt% % It is in another manual, so we don't have it. \def\printednodename{\ignorespaces #1}% \else \ifhavexrefs % We know the real title if we have the xref values. \def\printednodename{\refx{#1-title}}% \else % Otherwise just copy the Info node name. \def\printednodename{\ignorespaces #1}% \fi% \fi \def\printednodename{#1-title}% \else % Use the node name inside the square brackets. \def\printednodename{\ignorespaces #1}% \fi \fi % % If we use \unhbox0 and \unhbox1 to print the node names, TeX does not % insert empty discretionaries after hyphens, which means that it will % not find a line break at a hyphen in a node names. Since some manuals % are best written with fairly long node names, containing hyphens, this % is a loss. Therefore, we give the text of the node name again, so it % is as if TeX is seeing it for the first time. \ifdim \wd1 > 0pt \putwordsection{} ``\printednodename'' in \cite{\printedmanual}% \else % _ (for example) has to be the character _ for the purposes of the % control sequence corresponding to the node, but it has to expand % into the usual \leavevmode...\vrule stuff for purposes of % printing. So we \turnoffactive for the \refx-snt, back on for the % printing, back off for the \refx-pg. {\turnoffactive \refx{#1-snt}{}}% \space [\printednodename],\space \turnoffactive \putwordpage\tie\refx{#1-pg}{}% \fi \endgroup} % \dosetq is the interface for calls from other macros % Use \turnoffactive so that punctuation chars such as underscore % work in node names. \def\dosetq #1#2{{\let\folio=0 \turnoffactive \auxhat% \edef\next{\write\auxfile{\internalsetq {#1}{#2}}}% \next}} % \internalsetq {foo}{page} expands into % CHARACTERS 'xrdef {foo}{...expansion of \Ypage...} % When the aux file is read, ' is the escape character \def\internalsetq #1#2{'xrdef {#1}{\csname #2\endcsname}} % Things to be expanded by \internalsetq \def\Ypagenumber{\folio} \def\Ytitle{\thissection} \def\Ynothing{} \def\Ysectionnumberandtype{% \ifnum\secno=0 \putwordChapter\xreftie\the\chapno % \else \ifnum \subsecno=0 \putwordSection\xreftie\the\chapno.\the\secno % \else \ifnum \subsubsecno=0 % \putwordSection\xreftie\the\chapno.\the\secno.\the\subsecno % \else % \putwordSection\xreftie\the\chapno.\the\secno.\the\subsecno.\the\subsubsecno % \fi \fi \fi } \def\Yappendixletterandtype{% \ifnum\secno=0 \putwordAppendix\xreftie'char\the\appendixno{}% \else \ifnum \subsecno=0 \putwordSection\xreftie'char\the\appendixno.\the\secno % \else \ifnum \subsubsecno=0 % \putwordSection\xreftie'char\the\appendixno.\the\secno.\the\subsecno % \else % \putwordSection\xreftie'char\the\appendixno.\the\secno.\the\subsecno.\the\subsubsecno % \fi \fi \fi } \gdef\xreftie{'tie} % Use TeX 3.0's \inputlineno to get the line number, for better error % messages, but if we're using an old version of TeX, don't do anything. % \ifx\inputlineno\thisisundefined \let\linenumber = \empty % Non-3.0. \else \def\linenumber{\the\inputlineno:\space} \fi % Define \refx{NAME}{SUFFIX} to reference a cross-reference string named NAME. % If its value is nonempty, SUFFIX is output afterward. \def\refx#1#2{% \expandafter\ifx\csname X#1\endcsname\relax % If not defined, say something at least. $\langle$un\-de\-fined$\rangle$% \ifhavexrefs \message{\linenumber Undefined cross reference `#1'.}% \else \ifwarnedxrefs\else \global\warnedxrefstrue \message{Cross reference values unknown; you must run TeX again.}% \fi \fi \else % It's defined, so just use it. \csname X#1\endcsname \fi #2% Output the suffix in any case. } % Read the last existing aux file, if any. No error if none exists. % This is the macro invoked by entries in the aux file. \def\xrdef #1#2{ {\catcode`\'=\other\expandafter \gdef \csname X#1\endcsname {#2}}} \def\readauxfile{% \begingroup \catcode `\^^@=\other \catcode `\=\other \catcode `\=\other \catcode `\^^C=\other \catcode `\^^D=\other \catcode `\^^E=\other \catcode `\^^F=\other \catcode `\^^G=\other \catcode `\^^H=\other \catcode `\ =\other \catcode `\^^L=\other \catcode `\=\other \catcode `\=\other \catcode `\=\other \catcode `\=\other \catcode `\=\other \catcode `\=\other \catcode `\=\other \catcode `\=\other \catcode `\=\other \catcode `\=\other \catcode `\=\other \catcode `\=\other \catcode 26=\other \catcode `\^^[=\other \catcode `\^^\=\other \catcode `\^^]=\other \catcode `\^^^=\other \catcode `\^^_=\other \catcode `\@=\other \catcode `\^=\other \catcode `\~=\other \catcode `\[=\other \catcode `\]=\other \catcode`\"=\other \catcode`\_=\other \catcode`\|=\other \catcode`\<=\other \catcode`\>=\other \catcode `\$=\other \catcode `\#=\other \catcode `\&=\other % `\+ does not work, so use 43. \catcode 43=\other % Make the characters 128-255 be printing characters {% \count 1=128 \def\loop{% \catcode\count 1=\other \advance\count 1 by 1 \ifnum \count 1<256 \loop \fi }% }% % the aux file uses ' as the escape. % Turn off \ as an escape so we do not lose on % entries which were dumped with control sequences in their names. % For example, 'xrdef {$\leq $-fun}{page ...} made by @defun ^^ % Reference to such entries still does not work the way one would wish, % but at least they do not bomb out when the aux file is read in. \catcode `\{=1 \catcode `\}=2 \catcode `\%=\other \catcode `\'=0 \catcode`\^=7 % to make ^^e4 etc usable in xref tags \catcode `\\=\other \openin 1 \jobname.aux \ifeof 1 \else \closein 1 \input \jobname.aux \global\havexrefstrue \global\warnedobstrue \fi % Open the new aux file. Tex will close it automatically at exit. \openout \auxfile=\jobname.aux \endgroup} % Footnotes. \newcount \footnoteno % The trailing space in the following definition for supereject is % vital for proper filling; pages come out unaligned when you do a % pagealignmacro call if that space before the closing brace is % removed. \def\supereject{\par\penalty -20000\footnoteno =0 } % @footnotestyle is meaningful for info output only.. \let\footnotestyle=\comment \let\ptexfootnote=\footnote {\catcode `\@=11 % % Auto-number footnotes. Otherwise like plain. \gdef\footnote{% \global\advance\footnoteno by \@ne \edef\thisfootno{$^{\the\footnoteno}$}% % % In case the footnote comes at the end of a sentence, preserve the % extra spacing after we do the footnote number. \let\@sf\empty \ifhmode\edef\@sf{\spacefactor\the\spacefactor}\/\fi % % Remove inadvertent blank space before typesetting the footnote number. \unskip \thisfootno\@sf \footnotezzz }% % Don't bother with the trickery in plain.tex to not require the % footnote text as a parameter. Our footnotes don't need to be so general. % \long\gdef\footnotezzz#1{\insert\footins{% % We want to typeset this text as a normal paragraph, even if the % footnote reference occurs in (for example) a display environment. % So reset some parameters. \interlinepenalty\interfootnotelinepenalty \splittopskip\ht\strutbox % top baseline for broken footnotes \splitmaxdepth\dp\strutbox \floatingpenalty\@MM \leftskip\z@skip \rightskip\z@skip \spaceskip\z@skip \xspaceskip\z@skip \parindent\defaultparindent % % Hang the footnote text off the number. \hang \textindent{\thisfootno}% % % Don't crash into the line above the footnote text. Since this % expands into a box, it must come within the paragraph, lest it % provide a place where TeX can split the footnote. \footstrut #1\strut}% } }%end \catcode `\@=11 % Set the baselineskip to #1, and the lineskip and strut size % correspondingly. There is no deep meaning behind these magic numbers % used as factors; they just match (closely enough) what Knuth defined. % \def\lineskipfactor{.08333} \def\strutheightpercent{.70833} \def\strutdepthpercent {.29167} % \def\setleading#1{% \normalbaselineskip = #1\relax \normallineskip = \lineskipfactor\normalbaselineskip \normalbaselines \setbox\strutbox =\hbox{% \vrule width0pt height\strutheightpercent\baselineskip depth \strutdepthpercent \baselineskip }% } % @| inserts a changebar to the left of the current line. It should % surround any changed text. This approach does *not* work if the % change spans more than two lines of output. To handle that, we would % have adopt a much more difficult approach (putting marks into the main % vertical list for the beginning and end of each change). % \def\|{% % \vadjust can only be used in horizontal mode. \leavevmode % % Append this vertical mode material after the current line in the output. \vadjust{% % We want to insert a rule with the height and depth of the current % leading; that is exactly what \strutbox is supposed to record. \vskip-\baselineskip % % \vadjust-items are inserted at the left edge of the type. So % the \llap here moves out into the left-hand margin. \llap{% % % For a thicker or thinner bar, change the `1pt'. \vrule height\baselineskip width1pt % % This is the space between the bar and the text. \hskip 12pt }% }% } % For a final copy, take out the rectangles % that mark overfull boxes (in case you have decided % that the text looks ok even though it passes the margin). % \def\finalout{\overfullrule=0pt} % End of control word definitions. \message{and turning on texinfo input format.} \def\openindices{% \newindex{cp}% \newcodeindex{fn}% \newcodeindex{vr}% \newcodeindex{tp}% \newcodeindex{ky}% \newcodeindex{pg}% } % Set some numeric style parameters, for 8.5 x 11 format. %\hsize = 6.5in \newdimen\defaultparindent \defaultparindent = 15pt \parindent = \defaultparindent \parskip 18pt plus 1pt \setleading{15pt} \advance\topskip by 1.2cm % Prevent underfull vbox error messages. \vbadness=10000 % Following George Bush, just get rid of widows and orphans. \widowpenalty=10000 \clubpenalty=10000 % Use TeX 3.0's \emergencystretch to help line breaking, but if we're % using an old version of TeX, don't do anything. We want the amount of % stretch added to depend on the line length, hence the dependence on % \hsize. This makes it come to about 9pt for the 8.5x11 format. % \ifx\emergencystretch\thisisundefined % Allow us to assign to \emergencystretch anyway. \def\emergencystretch{\dimen0}% \else \emergencystretch = \hsize \divide\emergencystretch by 45 \fi % Use @smallbook to reset parameters for 7x9.5 format (or else 7x9.25) \def\smallbook{ % These values for secheadingskip and subsecheadingskip are % experiments. RJC 7 Aug 1992 \global\secheadingskip = 17pt plus 6pt minus 3pt \global\subsecheadingskip = 14pt plus 6pt minus 3pt \global\lispnarrowing = 0.3in \setleading{12pt} \advance\topskip by -1cm \global\parskip 3pt plus 1pt \global\hsize = 5in \global\vsize=7.5in \global\tolerance=700 \global\hfuzz=1pt \global\contentsrightmargin=0pt \global\deftypemargin=0pt \global\defbodyindent=.5cm \global\pagewidth=\hsize \global\pageheight=\vsize \global\let\smalllisp=\smalllispx \global\let\smallexample=\smalllispx \global\def\Esmallexample{\Esmalllisp} } % Use @afourpaper to print on European A4 paper. \def\afourpaper{ \global\tolerance=700 \global\hfuzz=1pt \setleading{12pt} \global\parskip 15pt plus 1pt \global\vsize= 53\baselineskip \advance\vsize by \topskip %\global\hsize= 5.85in % A4 wide 10pt \global\hsize= 6.5in \global\outerhsize=\hsize \global\advance\outerhsize by 0.5in \global\outervsize=\vsize \global\advance\outervsize by 0.6in \global\pagewidth=\hsize \global\pageheight=\vsize } % Allow control of the text dimensions. Parameters in order: textheight; % textwidth; \voffset; \hoffset (!); binding offset. All require a dimension; % header is additional; added length extends the bottom of the page. \def\changepagesizes#1#2#3#4#5{ \global\vsize= #1 \advance\vsize by \topskip \global\voffset= #3 \global\hsize= #2 \global\outerhsize=\hsize \global\advance\outerhsize by 0.5in \global\outervsize=\vsize \global\advance\outervsize by 0.6in \global\pagewidth=\hsize \global\pageheight=\vsize \global\normaloffset= #4 \global\bindingoffset= #5} % This layout is compatible with Latex on A4 paper. \def\afourlatex{\changepagesizes{22cm}{15cm}{7mm}{4.6mm}{5mm}} % Use @afourwide to print on European A4 paper in wide format. \def\afourwide{\afourpaper \changepagesizes{9.5in}{6.5in}{\hoffset}{\normaloffset}{\bindingoffset}} % Define macros to output various characters with catcode for normal text. \catcode`\"=\other \catcode`\~=\other \catcode`\^=\other \catcode`\_=\other \catcode`\|=\other \catcode`\<=\other \catcode`\>=\other \catcode`\+=\other \def\normaldoublequote{"} \def\normaltilde{~} \def\normalcaret{^} \def\normalunderscore{_} \def\normalverticalbar{|} \def\normalless{<} \def\normalgreater{>} \def\normalplus{+} % This macro is used to make a character print one way in ttfont % where it can probably just be output, and another way in other fonts, % where something hairier probably needs to be done. % % #1 is what to print if we are indeed using \tt; #2 is what to print % otherwise. Since all the Computer Modern typewriter fonts have zero % interword stretch (and shrink), and it is reasonable to expect all % typewriter fonts to have this, we can check that font parameter. % \def\ifusingtt#1#2{\ifdim \fontdimen3\the\font=0pt #1\else #2\fi} % Turn off all special characters except @ % (and those which the user can use as if they were ordinary). % Most of these we simply print from the \tt font, but for some, we can % use math or other variants that look better in normal text. \catcode`\"=\active \def\activedoublequote{{\tt \char '042}} \let"=\activedoublequote \catcode`\~=\active \def~{{\tt \char '176}} \chardef\hat=`\^ \catcode`\^=\active \def\auxhat{\def^{'hat}} \def^{{\tt \hat}} \catcode`\_=\active \def_{\ifusingtt\normalunderscore\_} % Subroutine for the previous macro. \def\_{\lvvmode \kern.06em \vbox{\hrule width.3em height.1ex}} % \lvvmode is equivalent in function to \leavevmode. % Using \leavevmode runs into trouble when written out to % an index file due to the expansion of \leavevmode into ``\unhbox % \voidb@x'' ---which looks to TeX like ``\unhbox \voidb\x'' due to our % magic tricks with @. \def\lvvmode{\vbox to 0pt{}} \catcode`\|=\active \def|{{\tt \char '174}} \chardef \less=`\< \catcode`\<=\active \def<{{\tt \less}} \chardef \gtr=`\> \catcode`\>=\active \def>{{\tt \gtr}} \catcode`\+=\active \def+{{\tt \char 43}} %\catcode 27=\active %\def^^[{$\diamondsuit$} % Set up an active definition for =, but don't enable it most of the time. {\catcode`\==\active \global\def={{\tt \char 61}}} \catcode`+=\active \catcode`\_=\active % If a .fmt file is being used, characters that might appear in a file % name cannot be active until we have parsed the command line. % So turn them off again, and have \everyjob (or @setfilename) turn them on. % \otherifyactive is called near the end of this file. \def\otherifyactive{\catcode`+=\other \catcode`\_=\other} \catcode`\@=0 % \rawbackslashxx output one backslash character in current font \global\chardef\rawbackslashxx=`\\ %{\catcode`\\=\other %@gdef@rawbackslashxx{\}} % \rawbackslash redefines \ as input to do \rawbackslashxx. {\catcode`\\=\active @gdef@rawbackslash{@let\=@rawbackslashxx }} % \normalbackslash outputs one backslash in fixed width font. \def\normalbackslash{{\tt\rawbackslashxx}} % Say @foo, not \foo, in error messages. \escapechar=`\@ % \catcode 17=0 % Define control-q \catcode`\\=\active % Used sometimes to turn off (effectively) the active characters % even after parsing them. @def@turnoffactive{@let"=@normaldoublequote @let\=@realbackslash @let~=@normaltilde @let^=@normalcaret @let_=@normalunderscore @let|=@normalverticalbar @let<=@normalless @let>=@normalgreater @let+=@normalplus} @def@normalturnoffactive{@let"=@normaldoublequote @let\=@normalbackslash @let~=@normaltilde @let^=@normalcaret @let_=@normalunderscore @let|=@normalverticalbar @let<=@normalless @let>=@normalgreater @let+=@normalplus} % Make _ and + \other characters, temporarily. % This is canceled by @fixbackslash. @otherifyactive % If a .fmt file is being used, we don't want the `\input texinfo' to show up. % That is what \eatinput is for; after that, the `\' should revert to printing % a backslash. % @gdef@eatinput input texinfo{@fixbackslash} @global@let\ = @eatinput % On the other hand, perhaps the file did not have a `\input texinfo'. Then % the first `\{ in the file would cause an error. This macro tries to fix % that, assuming it is called before the first `\' could plausibly occur. % Also back turn on active characters that might appear in the input % file name, in case not using a pre-dumped format. % @gdef@fixbackslash{@ifx\@eatinput @let\ = @normalbackslash @fi @catcode`+=@active @catcode`@_=@active} %% These look ok in all fonts, so just make them not special. The @rm below %% makes sure that the current font starts out as the newly loaded cmr10 @catcode`@$=@other @catcode`@%=@other @catcode`@&=@other @catcode`@#=@other @textfonts @rm @c Local variables: @c page-delimiter: "^\\\\message" @c End: dist/doc/rx.texi.~1~100660 765 144 53314 6070123654 13241 0ustar lordusers\input texinfo @c -*-texinfo-*- @c %**start of header @setfilename rx.info @settitle Rx @setchapternewpage odd @c %**end of header @c This title page illustrates only one of the @c two methods of forming a title page. @titlepage @title Rx @subtitle A Dynamic Automata Library @author Tom Lord @c The following two commands @c start the copyright page. @page @vskip 0pt plus 1filll Copyright @copyright{} 1995 Cygnus Support Permission is granted to make and distribute verbatim copies of this manual provided the copyright notice and this permission notice are preserved on all copies. Permission is granted to copy and distribute modified versions of this manual under the conditions for verbatim copying, provided that the entire resulting derived work is distributed under the terms of a permission notice identical to this one. Permission is granted to copy and distribute translations of this manual into another language, under the above conditions for modified versions, except that this permission notice may be stated in a translation approved by the author. @end titlepage @node Top, Posix Functions, (dir), (dir) @top Rx This document describes Rx. This document applies to version "Rx-diSpencer" of the library named librx.a @ifinfo @menu * Posix Functions:: Special considerations when using Rx. * Semantic Tests:: A test-suite for Posix matcher semantics. * General Functions:: Rx's native entry points. * Rx Theory:: A quick tour of the implementation. * Improvements to Make:: The Rx TODO list. @end menu @end ifinfo @node Posix Functions, Semantic Tests, Top, Top @section Posix Functions @cindex Posix functions In order to use the Rx implementation of Posix regexp functions, you should include @code{} in your program. In order to use the Rx implementation of Posix regexp functions most effectively, it may help to know a bit about performance tuning in Rx. @menu * Avoiding Redundant Compilations :: Compiling Regexps is costly. * Controlling Memory Use:: You can limit Rx's memory use. @end menu @node Avoiding Redundant Compilations , Controlling Memory Use, Posix Functions, Posix Functions @subsection Avoiding Redundant Compilations Rx implements the Posix regexp entry points @code{regcomp}, @code{regerror}, @code{regexec}, and @code{regfree}. Some special considerations apply when using the Rx versions. First, @code{regcomp} is a comparatively expensive operation. Addiitonally, @code{regexec} tends to perform better when a compiled expression is compiled once and used twice than when the pattern is compiled twice and used twice. Internally, Rx caches that information about a pattern which is constructed dynamicly by regexec. You can save on compilation costs and maximize the effectiveness of the Rx cache by re-using compiled expressions when it is convenient. For example, for reasons unrelated to Rx, it has long been the practice in GNU emacs to always save one compiled regular expression (the most recent). Before compiling a new expression, Emacs checks to see if the new pattern is the same as the one that is already compiled. This is the sort of optimization that Rx likes. (It may even win, in some cases, to cache more than a single compiled regexp.) In the long-run, Rx should be modified to cache compilations on its own. (@xref{Improvements to Make}.) Sometimes programmers write programs with many regexps, all known staticly at compile time. This can be a problem with Rx because when the program starts up, or as it runs, all of those staticly known regexps have to be compiled, which may be too time consuming. One easy fix for programs with many static regexps is to use @code{flex} or another lexical-analyzer generator program instead. Lexer-generators are optimized for the case of compiling staticly known regexps. There are still reasons why staticly known regexps may not be convertable to @code{flex} input -- for example, the Posix pattern language is more general than @code{flex}'s. In the long-run, it may be a good idea to extend Rx to support static compilation of regexps. (@xref{Improvements to Make}.) @node Controlling Memory Use, , Avoiding Redundant Compilations , Posix Functions @subsection Controlling Memory Use The size of the cache used by regexec is subject to control by programmers. Additionally, by using lower level entry points, programmers can create multiple, distinct Rx caches. This part of the code is subject to change and so is not yet documented. See "default_cache" in rxsuper.c and the parameters at the top of "rxbasic.c" to find some parameters you can tune or hints about creating alternative regexec caches. @node Semantic Tests, General Functions, Posix Functions, Top @section Semantic Tests @cindex Semantic tests The distribution of @emph{Rx} includes a test-suite, aimed at testing the conformance of an implementation of the Posix regexp functions to the standard. Currently, the tests come from two sources: a test-suite started by Henry Spencer, and the "Khadafy" test suite by Scott Anderson. The tests come in three files: @itemize @bullet @item runtests.c -- The driver program. @item TESTS -- The list of test cases @item TESTS2C.sed -- A script to convert test cases into C. @end itemize @menu * Running the Tests:: Invoking the test suite. * Adding New Tests:: Extending the test suite. @end menu @node Running the Tests, Adding New Tests, Semantic Tests, Semantic Tests @subsection Running the Tests To build the test program, use @code{make runtests}. (Detailed configuration and build instructions can be found in INSTALL). Invoked with no arguments, @code{runtests} runs all test-cases. For each test case, a sequence number is printed. If there is a problem with that case, more information is printed. Output like this: @example #0 #1 #2 ... @end example indicates a successful run. (The output format is likely to change in the future to make it easier to automate testing of Rx.) With a single numeric argument, runtests executes just that test and no others. This is handy when debugging. Sometimes a bug will occur when all tests are run, but disappear when just the problematic test is run. Usually this has to do with memory or cache corruption. @node Adding New Tests, , Running the Tests, Semantic Tests @subsection Adding New Tests This list of tests used by @code{runtests} is found in the file @code{TESTS}. Each line of that file is a list of colon separated fields similar to: @example 1:^(ab|cd)e:abcde @end example The first field is the expected return value of regexec, or '2' meaning that the pattern is not valid. The second field is the regular expression being tested. The third field is the string to which the pattern is to be compared. The file @code{TESTS2C.sed} is used to convert the test cases into initializers for an array of structures (which is automated in the Makefile). To add a new test case, add lines to @code{TESTS} and recompile @code{runtests}. @node General Functions, Rx Theory, Semantic Tests, Top @section General Functions @cindex Generalized entry points Here is a whirlwind tour of the lower-level Rx entry points and data structures. This is not intended to be complete, but rather to be a help to anyone reading the code itself. These are presented in the same order in which they might typically be used. @menu * Compiling Expressions:: Strings->Syntax Trees * Posixly Correct Matches:: Checking for a Posixly correct match. * Hairy Matches:: Matching with multi-part strings. * NFA Functions:: Forget Posix, eat NFAs raw. @end menu @node Compiling Expressions, Posixly Correct Matches, General Functions, General Functions @subsection Compiling Expressions @deftypefn {Internal Function} reg_errcode_t rx_parse (struct rexp_node ** rexp_p, __const__ char *pattern, int size, unsigned long syntax, int cset_size, unsigned char *translate); Compile a regular expression. @var{rexp_p} will return a syntax tree for the pattern. See @code{"rxnode.h"} to learn about syntax trees. Minimally, you should know that syntax trees are reference counted using @code{rx_save_rexp} and @code{rx_free_rexp}. @var{pattern} and @var{size} are the expression and its length. The expression need not be 0-terminated. The bits set in @var{syntax} control details of the language understood by the parser. The meaning of each bit is described in @code{"rxgnucomp.h"}. @var{cset_size} is usually 256. It should not be much larger than that or space-performance will suffer badly. @var{translate} is a translation table -- an array of characters. Characters in the pattern are looked up in @var{translate} as they are read. Additionally, the pattern is compiled presuming that the same translation table will be applied to characters in the string being searched (meaning that the pattern is compiled in such a way that the effect will be as if every character in the target string is looked up in the translation table, even though the expense of that lookup is usually avoided). A return of 0 indicates success, otherwise, the error can be looked up in the @code{rx_error_msg} array. @end deftypefn @deftypefn {Internal Function} int rx_posix_analyze_rexp (struct rexp_node *** subexps, int * n_subexps, struct rexp_node * node, int id) Staticly analyze a regular expression in preparation for matching. This function fills in some fields of the nodes of a syntax tree. @code{subexps} and @code{n_subexps} return a malloced array of pointers into the syntax tree, one per parenthesized subexpression. The caller must eventually free this array. @code{node} is the pattern to be analyzed. The tree returned by @code{rx_parse} is suitable for this. @code{id} should be 1. Ignore the return value. This function is not robust in the case that @code{malloc} or @code{realloc} returns 0 (which should be fixed). @end deftypefn You might expect that @code{rx_posix_analyze_rexp} and @code{rx_parse} should be combined. They are not because the actions of @code{rx_posix_analyze_rexp} make sense for any syntax tree, regardless of how it is constructed. @code{rx_parse} is just one way to build a syntax tree -- it is subject to replacement and trees can be built without using a parser at all (c.f. @code{"rxnode.h"}). @node Posixly Correct Matches, Hairy Matches, Compiling Expressions, General Functions @subsection Checking for Posixly Correct Matches with One-Piece Strings Looking for a match that is Posixly correct is conceptually tricky. Here is one way to think of it: The Posix expression languages, absent of any consideration of the meaning of "leftmost-longest", underspecify the return value of @code{regexec}. Consider the example: @example Pattern: foo\(.*\)\(.*\)bar Target string: fooxxxbar @end example Initially it is ambiguous how this should be matched. The ambiguity is exposed by considering the possible bindings for the subexpressions @code{\1} and @code{\2}: @example Possible outcomes: \1 == "xxx" \2 == "" \1 == "xx" \2 == "x" \1 == "x" \2 == "xx" \1 == "" \2 == "xxx" @end example Posix tells us that the first outcome is the prefered one if the illustrated pattern is used alone (this is a consequence of the "leftmost longest" rule). But Posix also says the answer may be different if the pattern is a sub-pattern of a larger pattern. For example: @example Pattern: foo\(.*\)\(.*\)bar\2 Target string: fooxxxbarxx @end example Now the only acceptable outcome is: @example \1 == "x" \2 == "xx" @end example Three entry points provide a relatively simple interface to this moderately complicated situation: @deftypefn {Internal Function} {struct rx_solutions *} rx_basic_make_solutions (struct rx_registers * regs, struct rexp_node * expression, struct rexp_node ** subexps, int start, int end, struct rx_context_rules * rules, char * str) @deftypefnx {Internal Function} void rx_basic_free_solutions (struct rx_solutions * solns); @deftypefnx {Internal Function} {enum rx_answers} rx_next_solution (struct rx_solutions * solns) @deftypefnx {Internal Function} void rx_basic_free_solutions (struct rx_solutions * solns) @code{make_solutions} returns a lazilly constructed stream of possible solutions for a given regexp and target string. @code{next_solution} constructs and returns the next solution from a list returned by @code{make_solutions}. Solutions are returned in order of preferability according to the Posix spec. So, the first solution is the leftmost-longest, and the last, the rightmost-shortest. @var{regs} is where subexpression extents are tracked. @var{expression} is the expression to solve. @var{subexps} is as returned by @code{rx_posix_analyze_rexp}. @var{start} and @var{end} are the integer addresses of the data to be compared to the pattern. A solution must match all characters starting at @var{start} and ending at @code{@var{end} - 1}. @var{rules} are some bits that control the precise meaning of "^" and "$". See @code{"rxcontext.h"}. @var{str} is the data to be matched. Note that anchoring operators can cause the matcher to look beyond @var{start} and @var{end} when testing a match. It may make sense to pass @var{start} > 0, for example. @code{make_solutions} returns a pointer to a solution list, or 0 if an allocation fails. @code{next_solution} returns a status which can be: @example @code{rx_yes} -- The pattern matches exactly. @code{rx_maybe} -- The pattern does not match, but might if more characters were added to the string. @code{rx_no} -- The pattern does not match and could not be made to match even by adding more characters. @end example In addition, @var{regs} is filled in by @code{next_solution}. Finally, @code{free_solutions} is used to release resources consumed by a solution list. @end deftypefn @node Hairy Matches, NFA Functions, Posixly Correct Matches, General Functions @subsection Hairy Matches and Suspendable Searches The functions @code{rx_basic_make_solutions} and @code{rx_basic_free_solutions} are used to find matches in a one-piece string. More general entry points are provided for multi-piece strings, which may or may not be entirely available in memory at one time: @deftypefn {Internal Function} {struct rx_solutions *} rx_make_solutions (struct rx_registers * regs, struct rx_unfaniverse * verse, struct rexp_node * expression, struct rexp_node ** subexps, int cset_size, int start, int end, rx_vmfn vmfn, rx_contextfn contextfn, void * closure) @deftypefnx {Internal Function} void rx_free_solutions (struct rx_solutions * solns) These functions are similar to their @code{_basic_} analogues, but substitute some new parameters for the one piece string: @var{vmfn} is a function provided by the caller that returns at least part of the string being matched. @var{contextfn} is a function provided by the caller that does the work of the backreference operators, "^", and "$". @var{closure} is data provided by the caller, passed through to vmfn and contextfn. @example typedef enum rx_answers (*rx_vmfn) P((void * closure, unsigned char ** burst, int * len, int * offset, int start, int end, int need)); typedef enum rx_answers (*rx_contextfn) P((void * closure, int type, int start, int end, struct rx_registers * regs)); @end example @code{vmfn} is asked for a range of characters from @var{start} to @var{end}. @var{need} is in that range and is the only position that must be returned by vmfn. @var{burst}, @var{len}, and @var{offset} are output parameters. @var{offset} will get the integer address of burst, and @var{len} the number of valid characters in the burst. @var{vmfn} may be asked for the empty string from @var{end} to @var{end} and should be able to handle that case. If @code{vmfn} provides the needed character, it should return @code{rx_yes}. If @code{vmfn} can't provide the character now, but might be able to later, it should return @code{rx_maybe}. This will cause @code{rx_next_solution} to quickly return @code{rx_maybe}. If @code{rx_next_solution} is called again, it will retry the call to vmfn and possibly resume the search. If @code{vmfn} can not provide the needed character, it should return @code{rx_no}. @code{rx_next_solution} will not normally ask for out-of-range characters, so usually there is no good reason to return @code{rx_no}, but the code is supposed to be robust in case you do. The @code{contextfn} also returns a @code{yes/no/maybe} status. It tests whether the characters in @var{start} to @var{end} satisfy the operator @var{type}. @code{type} is an ascii value; one of the characters in "^$123456789". @end deftypefn @node NFA Functions, , Hairy Matches, General Functions @subsection NFA Functions and Super-NFA Functions Not really documented yet. @example rx.h, rxnfa.h -- translating an expression into an NFA. rxsuper.h -- optimizing the NFA into a DFA, or at least an NFA with fewer branch points. rxanal.h -- how to use DFAs to find longest matches, matches of a specific length, the shortest match, etc. rxsimp.h -- Translating a Posix "regular expression" into a true regular expression that describes a superset language. @end example @node Rx Theory, Improvements to Make, General Functions, Top @section Rx Theory @cindex Rx Theory There are two match algorithms. One is for truly regular regexps (those that can be reduced to a dfa). The other is for non-regular regexps. The dfa algorithm implements the idea suggested in @cite{Compilers} by Aho, Sethi and Ullman: @quotation [One] approach [to pattern matching regular expressions] is to use a DFA, but avoid constructing all of the transition table by using a technique called "lazy transition evaluation". Here, transitions are computed at run time [when] actually needed. [T]ransitions are stored in a cache. [....] If the cache becomes full, we can erase some previously computed transition to make room for the new transition. @end quotation The implementation in Rx is generalized from that, but the above description covers what is used for Posix patterns. The non-dfa algorithm implements a "recursive decomposition" technique described in email by Henry Spencer. For a given pattern, this algorithm first checks to see if a simpler, superset language, DFA-pattern matches. If it does, then this algorithm does the detail-work to see if the non-DFA pattern matches. The detail work usually involves recursing on subpatterns. For example, a concatentation of two subexpressions matches a string if the string can be divided into two parts, each matching one subexpression, in the right order. More than one solution is often possible for a given pattern. This ambiguity is the subject of the "leftmost longest" rules in the spec, and the back-tracking oriented stream-of-solution functions @code{rx_make_solutions}, @code{rx_next_solution} and @code{rx_free_solutions}. @example rxspencer.[ch] -- The non-DFA algorithm rxanal.[ch] rxsuper.[ch] rxnfa.[ch] -- The DFA algorithm @end example @node Improvements to Make, , Rx Theory, Top @section Improvements to Make @cindex Improvements to make An approximation of the GNU entry points is needed. This can not easily be an exact approximation because the GNU entry points have no function for freeing a compiled expression -- they assume that compiled expressions can be freed using a single call to "free(3)". For that reason, do not use the names of the old GNU entry points ...use "rx_" instead. This version of Rx hasn't been profiled and tuned. It might be nice to have some C++ types to hide reference counting and provide handy constructors for rexp_nodes and nfa parts. Right now, the regexp->regular regexp mapping in rxsimp.[ch] is cached for a particular input expression. I think it would probably be a big win in many situations if it were cached, like unique nfas (c.f. rxunfa.[ch]), for equivalent expressions. It would speed up matching to lift the non-regular constructs to the top of the syntax tree. e.g., if you have: @example pattern: "abc$" syntax tree: (concat (cset "a") (concat (cset "b") (concat (cset "c") (operator "$")))) @end example it involves a lot more work (in "next_solution) than if you have the equivalent: @example same pattern: "abc$" syntax tree: (concat (concat (cset "a") (concat (cset "b") (cset "c"))) (operator "$")) @end example in fact, while doing that, notice that subtrees of nothing-but-concat can conceivably be optimized so that next_solution just uses "strcmp". For that matter, a tree which is a disjunction of conjunction (a tree of "|" whose leaves are trees of concat nodes) could be optimized to use a table method (Boyer/Moore?). Again, this would be for next_solution but might involve adding stuff to rx_posix_analyze_rexp. As more and more optimizations are added, programmers should be given simple control over which optimizations are important for a particular pattern. For example, packages in emacs often include a few regexps that are used heavily -- it would be worth the trouble to arrange for those regexps to be compiled specially. Static compilation of regexps is mostly a job for lex, but posix patterns are more general and have a nicer interface. What's more, it'd be nice to have the benefits of static compilation even for dynamicly loaded regexps (e.g., imagine if when an emacs lisp file is byte-compiled, the regexps it contains are also compiled). A lot of optimizations apply to just syntax trees. One approach to static compilation is to make an ugly but fast-reading syntax for trees, and a printer for that syntax. Then, after optimiztaions, a pattern could be stored as a string and later read back in, already optimized. A similar trick could be done for the nfa built in rxnfa.c. next_solution already considers the length of a fixed-length subexpression, but many more lenght optimizations are possible. I'd guess these will pay off well. @cindex TODO dist/install-sh100770 765 144 11243 6071377322 12417 0ustar lordusers#!/bin/sh # # install - install a program, script, or datafile # This comes from X11R5. # # Calling this script install-sh is preferred over install.sh, to prevent # `make' implicit rules from creating a file called install from it # when there is no Makefile. # # This script is compatible with the BSD install script, but was written # from scratch. # # set DOITPROG to echo to test this script # Don't use :- since 4.3BSD and earlier shells don't like it. doit="${DOITPROG-}" # put in absolute paths if you don't have them in your path; or use env. vars. mvprog="${MVPROG-mv}" cpprog="${CPPROG-cp}" chmodprog="${CHMODPROG-chmod}" chownprog="${CHOWNPROG-chown}" chgrpprog="${CHGRPPROG-chgrp}" stripprog="${STRIPPROG-strip}" rmprog="${RMPROG-rm}" mkdirprog="${MKDIRPROG-mkdir}" tranformbasename="" transform_arg="" instcmd="$mvprog" chmodcmd="$chmodprog 0755" chowncmd="" chgrpcmd="" stripcmd="" rmcmd="$rmprog -f" mvcmd="$mvprog" src="" dst="" dir_arg="" while [ x"$1" != x ]; do case $1 in -c) instcmd="$cpprog" shift continue;; -d) dir_arg=true shift continue;; -m) chmodcmd="$chmodprog $2" shift shift continue;; -o) chowncmd="$chownprog $2" shift shift continue;; -g) chgrpcmd="$chgrpprog $2" shift shift continue;; -s) stripcmd="$stripprog" shift continue;; -t=*) transformarg=`echo $1 | sed 's/-t=//'` shift continue;; -b=*) transformbasename=`echo $1 | sed 's/-b=//'` shift continue;; *) if [ x"$src" = x ] then src=$1 else # this colon is to work around a 386BSD /bin/sh bug : dst=$1 fi shift continue;; esac done if [ x"$src" = x ] then echo "install: no input file specified" exit 1 else true fi if [ x"$dir_arg" != x ]; then dst=$src src="" if [ -d $dst ]; then instcmd=: else instcmd=mkdir fi else # Waiting for this to be detected by the "$instcmd $src $dsttmp" command # might cause directories to be created, which would be especially bad # if $src (and thus $dsttmp) contains '*'. if [ -f $src -o -d $src ] then true else echo "install: $src does not exist" exit 1 fi if [ x"$dst" = x ] then echo "install: no destination specified" exit 1 else true fi # If destination is a directory, append the input filename; if your system # does not like double slashes in filenames, you may need to add some logic if [ -d $dst ] then dst="$dst"/`basename $src` else true fi fi ## this sed command emulates the dirname command dstdir=`echo $dst | sed -e 's,[^/]*$,,;s,/$,,;s,^$,.,'` # Make sure that the destination directory exists. # this part is taken from Noah Friedman's mkinstalldirs script # Skip lots of stat calls in the usual case. if [ ! -d "$dstdir" ]; then defaultIFS=' ' IFS="${IFS-${defaultIFS}}" oIFS="${IFS}" # Some sh's can't handle IFS=/ for some reason. IFS='%' set - `echo ${dstdir} | sed -e 's@/@%@g' -e 's@^%@/@'` IFS="${oIFS}" pathcomp='' while [ $# -ne 0 ] ; do pathcomp="${pathcomp}${1}" shift if [ ! -d "${pathcomp}" ] ; then $mkdirprog "${pathcomp}" else true fi pathcomp="${pathcomp}/" done fi if [ x"$dir_arg" != x ] then $doit $instcmd $dst && if [ x"$chowncmd" != x ]; then $doit $chowncmd $dst; else true ; fi && if [ x"$chgrpcmd" != x ]; then $doit $chgrpcmd $dst; else true ; fi && if [ x"$stripcmd" != x ]; then $doit $stripcmd $dst; else true ; fi && if [ x"$chmodcmd" != x ]; then $doit $chmodcmd $dst; else true ; fi else # If we're going to rename the final executable, determine the name now. if [ x"$transformarg" = x ] then dstfile=`basename $dst` else dstfile=`basename $dst $transformbasename | sed $transformarg`$transformbasename fi # don't allow the sed command to completely eliminate the filename if [ x"$dstfile" = x ] then dstfile=`basename $dst` else true fi # Make a temp file name in the proper directory. dsttmp=$dstdir/#inst.$$# # Move or copy the file name to the temp name $doit $instcmd $src $dsttmp && trap "rm -f ${dsttmp}" 0 && # and set any options; do chmod last to preserve setuid bits # If any of these fail, we abort the whole thing. If we want to # ignore errors from any of these, just make sure not to ignore # errors from the above "$doit $instcmd $src $dsttmp" command. if [ x"$chowncmd" != x ]; then $doit $chowncmd $dsttmp; else true;fi && if [ x"$chgrpcmd" != x ]; then $doit $chgrpcmd $dsttmp; else true;fi && if [ x"$stripcmd" != x ]; then $doit $stripcmd $dsttmp; else true;fi && if [ x"$chmodcmd" != x ]; then $doit $chmodcmd $dsttmp; else true;fi && # Now rename the file to the real destination. $doit $rmcmd -f $dstdir/$dstfile && $doit $mvcmd $dsttmp $dstdir/$dstfile fi && exit 0 dist/RX100660 765 144 5757 6071536157 10670 0ustar lordusersRx, version 1.0, is now available. Copies can be obtained by ftp from prep.ai.mit.edu:pub/gnu/rx-1.0.tar.gz and from other sites (see the list of mirror sites at the end of this message). Rx is an implementation of the standard regular expression matching functions specified by POSIX.2 (and then some). The distribution now includes a texinfo manual. The largest change in this version of Rx is an algorithmic improvement based on advice form Henry Spencer. The resulting matcher is often a little slower than older versions of Rx, but more importantly, where older versions of Rx used unreasonable amounts of time or space, the new version gives reasonable performance. Older versions of Rx included implementations of the standard GNU regexp entry points, as well as of the traditional BSD entry points. In the interest of saving time, these have not been ported to Rx 1.0, though contributions of implementations are welcome. Please report bugs to bug-gnu-utils@prep.ai.mit.edu. [ Most GNU software is packed using the GNU `gzip' compression program. Source code is available on most sites distributing GNU software. For information on how to order GNU software on tape, floppy or cd-rom, or printed GNU manuals, check the file etc/ORDERS in the GNU Emacs distribution, ftp the file /pub/gnu/GNUinfo/ORDERS on prep, or e-mail a request to: gnu@prep.ai.mit.edu By ordering your GNU software from the FSF, you help us continue to develop more free software. Media revenues are our primary source of support. Donations to FSF are deductible on US tax returns. The above software will soon be at these ftp sites as well. Please try them before prep.ai.mit.edu as prep is very busy! thanx -gnu@prep.ai.mit.edu ASIA: ftp.cs.titech.ac.jp, utsun.s.u-tokyo.ac.jp:/ftpsync/prep, cair-archive.kaist.ac.kr:/pub/gnu, ftp.nectec.or.th:/pub/mirrors/gnu AUSTRALIA: archie.au:/gnu (archie.oz or archie.oz.au for ACSnet) AFRICA: ftp.sun.ac.za:/pub/gnu MIDDLE-EAST: ftp.technion.ac.il:/pub/unsupported/gnu EUROPE: irisa.irisa.fr:/pub/gnu, ftp.univ-lyon1.fr:pub/gnu, ftp.mcc.ac.uk, unix.hensa.ac.uk:/mirrors/uunet/systems/gnu, src.doc.ic.ac.uk:/gnu, ftp.ieunet.ie:pub/gnu, ftp.eunet.ch, nic.switch.ch:/mirror/gnu, ftp.informatik.rwth-aachen.de:/pub/gnu, ftp.informatik.tu-muenchen.de, ftp.win.tue.nl:/pub/gnu, ftp.nl.net, ftp.etsimo.uniovi.es:/pub/gnu ftp.funet.fi:/pub/gnu, ftp.denet.dk, ftp.stacken.kth.se, isy.liu.se, ftp.luth.se:/pub/unix/gnu, ftp.sunet.se:/pub/gnu, archive.eu.net SOUTH AMERICA: ftp.inf.utfsm.cl:/pub/gnu, ftp.unicamp.br:/pub/gnu WESTERN CANADA: ftp.cs.ubc.ca:/mirror2/gnu USA: wuarchive.wustl.edu:/systems/gnu, labrea.stanford.edu, ftp.digex.net:/pub/gnu, ftp.kpc.com:/pub/mirror/gnu, f.ms.uky.edu:/pub3/gnu, jaguar.utah.edu:/gnustuff, ftp.hawaii.edu:/mirrors/gnu, uiarchive.cso.uiuc.edu:/pub/gnu, ftp.cs.columbia.edu:/archives/gnu/prep, col.hp.com:/mirrors/gnu, gatekeeper.dec.com:/pub/GNU, ftp.uu.net:/systems/gnu ] dist/rx/hashrexp.c100660 765 144 3455 6071416222 13012 0ustar lordusers/*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "rxall.h" #include "rxnode.h" #include "rxhash.h" static int rexp_node_equal (va, vb) void * va; void * vb; { struct rexp_node * a; struct rexp_node * b; a = (struct rexp_node *)va; b = (struct rexp_node *)vb; return ( (va == vb) || ( (a->type == b->type) && (a->params.intval == b->params.intval) && (a->params.intval2 == b->params.intval2) && rx_bitset_is_equal (a->params.cset_size, a->params.cset, b->params.cset) && rexp_node_equal (a->params.pair.left, b->params.pair.left) && rexp_node_equal (a->params.pair.right, b->params.pair.right))); } static struct rx_hash_rules rexp_node_hash_rules = { rexp_node_equal, 0, 0, 0, 0 }; dist/rx/runtests.c100660 765 144 5336 6071416222 13057 0ustar lordusers/*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "rxall.h" #include #include "rxposix.h" #include "malloc.h" struct a_test { int expected; char * pattern; unsigned char * data; }; struct a_test the_tests[] = { #include "testcases.h" {-1, 0, 0} }; void run_a_test (id, t) int id; struct a_test * t; { static char * last_pattern = 0; static regex_t r; int err; char errmsg[100]; int x; regmatch_t regs[10]; if (!last_pattern || strcmp (last_pattern, t->pattern)) { if (last_pattern) regfree (&r); last_pattern = t->pattern; err = regcomp (&r, t->pattern, REG_EXTENDED); if (err) { if (t->expected) return; regerror (err, &r, errmsg, 100); printf ("test %d\n", id); puts (errmsg); return; } } err = regexec (&r, t->data, 10, regs, 0); if (err != t->expected) { printf ("test %d\n", id); printf ("pattern \"%s\" data \"%s\" wanted %d got %d\n", t->pattern, t->data, t->expected, err); for (x = 0; x < 10; ++x) printf ("reg %d == (%d, %d) %.*s\n", x, regs[x].rm_so, regs[x].rm_eo, regs[x].rm_eo - regs[x].rm_so, t->data + regs[x].rm_so); } } main(argc, argv) int argc; char * argv[]; { int x; int lo; int hi; #if 0 { union dbmalloptarg val; val.i = 1; dbmallopt (MALLOC_CKCHAIN, &val); } #endif lo = 0; hi = (sizeof (the_tests) / sizeof (the_tests[0])) - 1; if (argc > 1) { lo = atoi (argv[1]); hi = lo + 1; if (argc > 2) hi = atoi (argv[2]); } for (x = lo; x < hi; ++x) { printf ("#%d\n", x); run_a_test (x, &the_tests[x]); } { exit (0); } } dist/rx/rx.c100660 765 144 3457 6071417412 11625 0ustar lordusers/*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "rxall.h" #include "rxhash.h" #include "rxnfa.h" const char rx_version_string[] = "GNU Rx version 1.0"; #ifdef __STDC__ struct rx * rx_make_rx (int cset_size) #else struct rx * rx_make_rx (cset_size) int cset_size; #endif { static int rx_id = 0; struct rx * new_rx; new_rx = (struct rx *)malloc (sizeof (*new_rx)); bzero (new_rx, sizeof (*new_rx)); new_rx->rx_id = rx_id++; new_rx->cache = rx_default_cache; new_rx->local_cset_size = cset_size; new_rx->instruction_table = rx_id_instruction_table; new_rx->next_nfa_id = 0; return new_rx; } #ifdef __STDC__ void rx_free_rx (struct rx * rx) #else void rx_free_rx (rx) struct rx * rx; #endif { rx_free_nfa (rx); free (rx); } dist/rx/rxanal.c100660 765 144 42603 6071416222 12473 0ustar lordusers/*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "rxall.h" #include "rxanal.h" #include "rxbitset.h" #ifdef __STDC__ int rx_posix_analyze_rexp (struct rexp_node *** subexps, int * n_subexps, struct rexp_node * node, int id) #else int rx_posix_analyze_rexp (subexps, n_subexps, node, id) struct rexp_node *** subexps; int * n_subexps; struct rexp_node * node; int id; #endif { if (node) { int this_subexp; if (node->type == r_parens) { this_subexp = *n_subexps; ++*n_subexps; if (!*subexps) *subexps = (struct rexp_node **)malloc (sizeof (struct rexp_node *) * *n_subexps); else *subexps = (struct rexp_node **)realloc (*subexps, sizeof (struct rexp_node *) * *n_subexps); } if (node->params.pair.left) id = rx_posix_analyze_rexp (subexps, n_subexps, node->params.pair.left, id); if (node->params.pair.right) id = rx_posix_analyze_rexp (subexps, n_subexps, node->params.pair.right, id); switch (node->type) { case r_cset: node->len = 1; node->observed = 0; break; case r_concat: case r_alternate: { int lob, rob; int llen, rlen; lob = (!node->params.pair.left ? 0 : node->params.pair.left->observed); rob = (!node->params.pair.right ? 0 : node->params.pair.right->observed); llen = (!node->params.pair.left ? 0 : node->params.pair.left->len); rlen = (!node->params.pair.right ? 0 : node->params.pair.right->len); node->len = ((llen >= 0) && (rlen >= 0) ? ((node->type == r_concat) ? llen + rlen : ((llen == rlen) ? llen : -1)) : -1); node->observed = lob || rob; break; } case r_opt: case r_star: case r_plus: case r_interval: node->len = -1; node->observed = (node->params.pair.left ? node->params.pair.left->observed : 0); break; case r_parens: node->observed = 1; node->len = (node->params.pair.left ? node->params.pair.left->len : 0); (*subexps)[this_subexp] = node; break; case r_context: switch (node->params.intval) { default: node->observed = 1; node->len = -1; break; case '$': case '=': case '<': case '>': case 'b': case 'B': case '`': case '\'': node->observed = 1; node->len = 0; break; } break; } if (node->observed) node->id = id++; return id; } } /* Returns 0 unless the pattern can match the empty string. */ #ifdef __STDC__ int rx_fill_in_fastmap (int cset_size, unsigned char * map, struct rexp_node * exp) #else int rx_fill_in_fastmap (cset_size, map, exp) int cset_size; unsigned char * map; struct rexp_node * exp; #endif { if (!exp) { can_match_empty: { int x; for (x = 0; x < cset_size; ++x) map[x] = 1; } return 1; } switch (exp->type) { case r_cset: { int x; int most; most = exp->params.cset_size; for (x = 0; x < most; ++x) if (RX_bitset_member (exp->params.cset, x)) map[x] = 1; } return 0; case r_concat: return rx_fill_in_fastmap (cset_size, map, exp->params.pair.left); /* Why not the right branch? If the left branch * can't be empty it doesn't matter. If it can, then * the fastmap is already saturated, and again, the * right branch doesn't matter. */ case r_alternate: return ( rx_fill_in_fastmap (cset_size, map, exp->params.pair.left) | rx_fill_in_fastmap (cset_size, map, exp->params.pair.right)); case r_parens: case r_plus: return rx_fill_in_fastmap (cset_size, map, exp->params.pair.left); case r_opt: case r_star: goto can_match_empty; /* Why not the subtree? These operators already saturate * the fastmap. */ case r_interval: if (exp->params.intval == 0) goto can_match_empty; else return rx_fill_in_fastmap (cset_size, map, exp->params.pair.left); case r_context: goto can_match_empty; } /* this should never happen but gcc seems to like it */ return 0; } #ifdef __STDC__ int rx_is_anchored_p (struct rexp_node * exp) #else int rx_is_anchored_p (exp) struct rexp_node * exp; #endif { if (!exp) return 0; switch (exp->type) { case r_opt: case r_star: case r_cset: return 0; case r_parens: case r_plus: case r_concat: return rx_is_anchored_p (exp->params.pair.left); case r_alternate: return ( rx_is_anchored_p (exp->params.pair.left) && rx_is_anchored_p (exp->params.pair.right)); case r_interval: if (exp->params.intval == 0) return 0; else return rx_is_anchored_p (exp->params.pair.left); case r_context: return (exp->params.intval == '^'); } /* this should never happen but gcc seems to like it */ return 0; } #ifdef __STDC__ enum rx_answers rx_start_superstate (struct rx_classical_system * frame) #else enum rx_answers rx_start_superstate (frame) struct rx_classical_system * frame; #endif { struct rx_superset * start_contents; struct rx_nfa_state_set * start_nfa_set; if (frame->state) { rx_unlock_superstate (frame->rx, frame->state); frame->state = 0; } if ( frame->rx->start_set && (frame->rx->start_set->starts_for == frame->rx) && (frame->rx->start_set->id == frame->rx->rx_id)) start_contents = frame->rx->start_set; else { { struct rx_possible_future * futures; futures = rx_state_possible_futures (frame->rx, frame->rx->start_nfa_states); if (!futures) return rx_bogus; if (futures->next) return rx_start_state_with_too_many_futures; start_nfa_set = futures->destset; } start_contents = rx_superstate_eclosure_union (frame->rx, rx_superset_cons (frame->rx, 0, 0), start_nfa_set); if (!start_contents) return rx_bogus; start_contents->starts_for = frame->rx; frame->rx->start_set = start_contents; } if ( start_contents->superstate && (start_contents->superstate->rx_id == frame->rx->rx_id)) { frame->state = start_contents->superstate; rx_lock_superstate (frame->rx, frame->state); return rx_yes; } else { rx_protect_superset (frame->rx, start_contents); frame->state = rx_superstate (frame->rx, start_contents); rx_release_superset (frame->rx, start_contents); if (!frame->state) return rx_bogus; rx_lock_superstate (frame->rx, frame->state); return rx_yes; } } #ifdef __STDC__ enum rx_answers rx_match_here_p (struct rx_classical_system * frame, unsigned char * burst, int len) #else enum rx_answers rx_match_here_p (frame, burst, len) struct rx_classical_system * frame; unsigned char * burst; int len; #endif { struct rx_inx * inx_table; if (!frame->state) return rx_bogus; if (!len) { return (frame->state->contents->is_final ? rx_yes : rx_maybe); } inx_table = frame->state->transitions; rx_unlock_superstate (frame->rx, frame->state); while (len--) { struct rx_inx * inx; struct rx_inx * next_table; inx = inx_table + *burst; next_table = (struct rx_inx *)inx->data; while (!next_table) { struct rx_superstate * state; state = ((struct rx_superstate *) ((char *)inx_table - ((unsigned long) ((struct rx_superstate *)0)->transitions))); switch ((int)inx->inx) { case rx_backtrack: /* RX_BACKTRACK means that we've reached the empty * superstate, indicating that match can't succeed * from this point. */ frame->state = 0; return rx_no; case rx_cache_miss: /* Because the superstate NFA is lazily constructed, * and in fact may erode from underneath us, we sometimes * have to construct the next instruction from the hard way. * This invokes one step in the lazy-conversion. */ inx = rx_handle_cache_miss (frame->rx, state, *burst, inx->data_2); if (!inx) { frame->state = 0; return rx_bogus; } next_table = (struct rx_inx *)inx->data; continue; /* No other instructions are legal here. * (In particular, this function does not handle backtracking * or the related instructions.) */ default: frame->state = 0; return rx_bogus; } } if (inx->data_2) /* indicates a final superstate */ { frame->state = ((struct rx_superstate *) ((char *)inx_table - ((unsigned long) ((struct rx_superstate *)0)->transitions))); rx_lock_superstate (frame->rx, frame->state); return rx_yes; } inx_table = next_table; ++burst; } frame->state = ((struct rx_superstate *) ((char *)inx_table - ((unsigned long) ((struct rx_superstate *)0)->transitions))); rx_lock_superstate (frame->rx, frame->state); return rx_maybe; } #ifdef __STDC__ enum rx_answers rx_fit_p (struct rx_classical_system * frame, unsigned char * burst, int len) #else enum rx_answers rx_fit_p (frame, burst, len) struct rx_classical_system * frame; unsigned char * burst; int len; #endif { struct rx_inx * inx_table; struct rx_inx * inx; if (!frame->state) return rx_bogus; if (!len) { return (frame->state->contents->is_final ? rx_yes : rx_maybe); } inx_table = frame->state->transitions; rx_unlock_superstate (frame->rx, frame->state); while (len--) { struct rx_inx * next_table; inx = inx_table + *burst; next_table = (struct rx_inx *)inx->data; while (!next_table) { struct rx_superstate * state; state = ((struct rx_superstate *) ((char *)inx_table - ((unsigned long) ((struct rx_superstate *)0)->transitions))); switch ((int)inx->inx) { case rx_backtrack: /* RX_BACKTRACK means that we've reached the empty * superstate, indicating that match can't succeed * from this point. */ frame->state = 0; return rx_no; case rx_cache_miss: /* Because the superstate NFA is lazily constructed, * and in fact may erode from underneath us, we sometimes * have to construct the next instruction from the hard way. * This invokes one step in the lazy-conversion. */ inx = rx_handle_cache_miss (frame->rx, state, *burst, inx->data_2); if (!inx) { frame->state = 0; return rx_bogus; } next_table = (struct rx_inx *)inx->data; continue; /* No other instructions are legal here. * (In particular, this function does not handle backtracking * or the related instructions.) */ default: frame->state = 0; return rx_bogus; } } inx_table = next_table; ++burst; } if (inx->data_2) /* indicates a final superstate */ { frame->state = ((struct rx_superstate *) ((char *)inx_table - ((unsigned long) ((struct rx_superstate *)0)->transitions))); rx_lock_superstate (frame->rx, frame->state); return rx_yes; } frame->state = ((struct rx_superstate *) ((char *)inx_table - ((unsigned long) ((struct rx_superstate *)0)->transitions))); rx_lock_superstate (frame->rx, frame->state); return rx_maybe; } #ifdef __STDC__ enum rx_answers rx_longest (int * last_matched, int base_addr, struct rx_classical_system * frame, unsigned char * burst, int len) #else enum rx_answers rx_longest (last_matched, base_addr, frame, burst, len) int * last_matched; int base_addr; struct rx_classical_system * frame; unsigned char * burst; int len; #endif { struct rx_inx * inx_table; int len_in; if (!frame->state) return rx_bogus; if (!len) { if (frame->state->contents->is_final) { *last_matched = base_addr; return rx_yes; } else { return rx_maybe; } } len_in = len; inx_table = frame->state->transitions; rx_unlock_superstate (frame->rx, frame->state); while (len--) { struct rx_inx * inx; struct rx_inx * next_table; inx = inx_table + *burst; next_table = (struct rx_inx *)inx->data; while (!next_table) { struct rx_superstate * state; state = ((struct rx_superstate *) ((char *)inx_table - ((unsigned long) ((struct rx_superstate *)0)->transitions))); switch ((int)inx->inx) { case rx_backtrack: /* RX_BACKTRACK means that we've reached the empty * superstate, indicating that match can't succeed * from this point. */ frame->state = 0; return rx_no; case rx_cache_miss: /* Because the superstate NFA is lazily constructed, * and in fact may erode from underneath us, we sometimes * have to construct the next instruction from the hard way. * This invokes one step in the lazy-conversion. */ inx = rx_handle_cache_miss (frame->rx, state, *burst, inx->data_2); if (!inx) { frame->state = 0; return rx_bogus; } next_table = (struct rx_inx *)inx->data; continue; /* No other instructions are legal here. * (In particular, this function does not handle backtracking * or the related instructions.) */ default: frame->state = 0; return rx_bogus; } } if (inx->data_2) /* indicates a final superstate */ *last_matched = base_addr + len_in - len; inx_table = next_table; ++burst; } frame->state = ((struct rx_superstate *) ((char *)inx_table - ((unsigned long) ((struct rx_superstate *)0)->transitions))); rx_lock_superstate (frame->rx, frame->state); return rx_maybe; } #ifdef __STDC__ enum rx_answers rx_advance (struct rx_classical_system * frame, unsigned char * burst, int len) #else enum rx_answers rx_advance (frame, burst, len) struct rx_classical_system * frame; unsigned char * burst; int len; #endif { struct rx_inx * inx_table; int len_in; if (!frame->state) return rx_bogus; if (!len) return rx_yes; inx_table = frame->state->transitions; rx_unlock_superstate (frame->rx, frame->state); while (len--) { struct rx_inx * inx; struct rx_inx * next_table; inx = inx_table + *burst; next_table = (struct rx_inx *)inx->data; while (!next_table) { struct rx_superstate * state; state = ((struct rx_superstate *) ((char *)inx_table - ((unsigned long) ((struct rx_superstate *)0)->transitions))); switch ((int)inx->inx) { case rx_backtrack: /* RX_BACKTRACK means that we've reached the empty * superstate, indicating that match can't succeed * from this point. */ frame->state = 0; return rx_no; case rx_cache_miss: /* Because the superstate NFA is lazily constructed, * and in fact may erode from underneath us, we sometimes * have to construct the next instruction from the hard way. * This invokes one step in the lazy-conversion. */ inx = rx_handle_cache_miss (frame->rx, state, *burst, inx->data_2); if (!inx) { frame->state = 0; return rx_bogus; } next_table = (struct rx_inx *)inx->data; continue; /* No other instructions are legal here. * (In particular, this function does not handle backtracking * or the related instructions.) */ default: frame->state = 0; return rx_bogus; } } inx_table = next_table; ++burst; } frame->state = ((struct rx_superstate *) ((char *)inx_table - ((unsigned long) ((struct rx_superstate *)0)->transitions))); rx_lock_superstate (frame->rx, frame->state); return rx_yes; } #ifdef __STDC__ void rx_terminate_system (struct rx_classical_system * frame) #else void rx_terminate_system (frame) struct rx_classical_system * frame; #endif { if (frame->state) { rx_unlock_superstate (frame->rx, frame->state); frame->state = 0; } } #ifdef __STDC__ void rx_init_system (struct rx_classical_system * frame, struct rx * rx) #else void rx_init_system (frame, rx) struct rx_classical_system * frame; struct rx * rx; #endif { frame->rx = rx; frame->state = 0; } dist/rx/rxbasic.c100660 765 144 5426 6071416222 12623 0ustar lordusers/*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "rxall.h" #include "rxbasic.h" #include "rxunfa.h" #include "rxstr.h" int rx_basic_unfaniverse_delay = 64; struct rx_unfaniverse * rx_basic_uv = 0; static int init_basic_once () { if (rx_basic_uv) return 0; rx_basic_uv = rx_make_unfaniverse (rx_basic_unfaniverse_delay); return (rx_basic_uv ? 0 : -1); } static char * silly_hack = 0; #ifdef __STDC__ struct rx_solutions * rx_basic_make_solutions (struct rx_registers * regs, struct rexp_node * expression, struct rexp_node ** subexps, int start, int end, struct rx_context_rules * rules, char * str) #else struct rx_solutions * rx_basic_make_solutions (regs, rexp, subexps, start, end, rules, str) struct rx_registers * regs; struct rexp_node * expression; struct rexp_node ** subexps; int start; int end; struct rx_context_rules * rules; char * str; #endif { struct rx_str_closure * closure; if (init_basic_once ()) return 0; if (silly_hack) { closure = (struct rx_str_closure *)silly_hack; silly_hack = 0; } else closure = (struct rx_str_closure *)malloc (sizeof (*closure)); if (!closure) return 0; closure->str = str; closure->len = end; closure->rules = *rules; return rx_make_solutions (regs, rx_basic_uv, expression, subexps, 256, start, end, rx_str_vmfn, rx_str_contextfn, (void *)closure); } #ifdef __STDC__ void rx_basic_free_solutions (struct rx_solutions * solns) #else void rx_basic_free_solutions (solns) struct rx_solutions * solns; #endif { if (!silly_hack) silly_hack = (char *)solns->closure; else free (solns->closure); solns->closure = 0; rx_free_solutions (solns); } dist/rx/rxbitset.c100660 765 144 15253 6071416222 13053 0ustar lordusers/*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ /* * Tom Lord (lord@cygnus.com, lord@gnu.ai.mit.edu) */ #include "rxall.h" #include "rxbitset.h" #ifdef __STDC__ int rx_bitset_is_equal (int size, rx_Bitset a, rx_Bitset b) #else int rx_bitset_is_equal (size, a, b) int size; rx_Bitset a; rx_Bitset b; #endif { int x; RX_subset s; if (size == 0) return 1; s = b[0]; b[0] = ~a[0]; for (x = rx_bitset_numb_subsets(size) - 1; a[x] == b[x]; --x) ; b[0] = s; return !x && (s == a[0]); } #ifdef __STDC__ int rx_bitset_is_subset (int size, rx_Bitset a, rx_Bitset b) #else int rx_bitset_is_subset (size, a, b) int size; rx_Bitset a; rx_Bitset b; #endif { int x; x = rx_bitset_numb_subsets(size) - 1; while (x-- && ((a[x] & b[x]) == a[x])); return x == -1; } #ifdef __STDC__ int rx_bitset_empty (int size, rx_Bitset set) #else int rx_bitset_empty (size, set) int size; rx_Bitset set; #endif { int x; RX_subset s; s = set[0]; set[0] = 1; for (x = rx_bitset_numb_subsets(size) - 1; !set[x]; --x) ; set[0] = s; return !s; } #ifdef __STDC__ void rx_bitset_null (int size, rx_Bitset b) #else void rx_bitset_null (size, b) int size; rx_Bitset b; #endif { bzero (b, rx_sizeof_bitset(size)); } #ifdef __STDC__ void rx_bitset_universe (int size, rx_Bitset b) #else void rx_bitset_universe (size, b) int size; rx_Bitset b; #endif { int x = rx_bitset_numb_subsets (size); while (x--) *b++ = ~(RX_subset)0; } #ifdef __STDC__ void rx_bitset_complement (int size, rx_Bitset b) #else void rx_bitset_complement (size, b) int size; rx_Bitset b; #endif { int x = rx_bitset_numb_subsets (size); while (x--) { *b = ~*b; ++b; } } #ifdef __STDC__ void rx_bitset_assign (int size, rx_Bitset a, rx_Bitset b) #else void rx_bitset_assign (size, a, b) int size; rx_Bitset a; rx_Bitset b; #endif { int x; for (x = rx_bitset_numb_subsets(size) - 1; x >=0; --x) a[x] = b[x]; } #ifdef __STDC__ void rx_bitset_union (int size, rx_Bitset a, rx_Bitset b) #else void rx_bitset_union (size, a, b) int size; rx_Bitset a; rx_Bitset b; #endif { int x; for (x = rx_bitset_numb_subsets(size) - 1; x >=0; --x) a[x] |= b[x]; } #ifdef __STDC__ void rx_bitset_intersection (int size, rx_Bitset a, rx_Bitset b) #else void rx_bitset_intersection (size, a, b) int size; rx_Bitset a; rx_Bitset b; #endif { int x; for (x = rx_bitset_numb_subsets(size) - 1; x >=0; --x) a[x] &= b[x]; } #ifdef __STDC__ void rx_bitset_difference (int size, rx_Bitset a, rx_Bitset b) #else void rx_bitset_difference (size, a, b) int size; rx_Bitset a; rx_Bitset b; #endif { int x; for (x = rx_bitset_numb_subsets(size) - 1; x >=0; --x) a[x] &= ~ b[x]; } #ifdef __STDC__ void rx_bitset_revdifference (int size, rx_Bitset a, rx_Bitset b) #else void rx_bitset_revdifference (size, a, b) int size; rx_Bitset a; rx_Bitset b; #endif { int x; for (x = rx_bitset_numb_subsets(size) - 1; x >=0; --x) a[x] = ~a[x] & b[x]; } #ifdef __STDC__ void rx_bitset_xor (int size, rx_Bitset a, rx_Bitset b) #else void rx_bitset_xor (size, a, b) int size; rx_Bitset a; rx_Bitset b; #endif { int x; for (x = rx_bitset_numb_subsets(size) - 1; x >=0; --x) a[x] ^= b[x]; } #ifdef __STDC__ unsigned long rx_bitset_hash (int size, rx_Bitset b) #else unsigned long rx_bitset_hash (size, b) int size; rx_Bitset b; #endif { int x; unsigned long answer; answer = 0; for (x = 0; x < size; ++x) { if (RX_bitset_member (b, x)) answer += (answer << 3) + x; } return answer; } RX_subset rx_subset_singletons [RX_subset_bits] = { 0x1, 0x2, 0x4, 0x8, 0x10, 0x20, 0x40, 0x80, 0x100, 0x200, 0x400, 0x800, 0x1000, 0x2000, 0x4000, 0x8000, 0x10000, 0x20000, 0x40000, 0x80000, 0x100000, 0x200000, 0x400000, 0x800000, 0x1000000, 0x2000000, 0x4000000, 0x8000000, 0x10000000, 0x20000000, 0x40000000, 0x80000000 }; /* * (define l (let loop ((x 0) (l '())) (if (eq? x 256) l (loop (+ x 1) (cons x l))))) * (define lb (map (lambda (n) (number->string n 2)) l)) * (define lc (map string->list lb)) * (define ln (map (lambda (l) (map (lambda (c) (if (eq? c #\1) 1 0)) l)) lc)) * (define lt (map (lambda (l) (apply + l)) ln)) */ static int char_pops[256] = { 0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4, 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, 4, 5, 5, 6, 5, 6, 6, 7, 5, 6, 6, 7, 6, 7, 7, 8 }; #define RX_char_population(C) (char_pops[C]) #ifdef __STDC__ int rx_bitset_population (int size, rx_Bitset a) #else int rx_bitset_population (size, a) int size; rx_Bitset a; #endif { int x; int total; unsigned char s; if (size == 0) return 0; total = 0; x = sizeof (RX_subset) * rx_bitset_numb_subsets(size) - 1; while (x >= 0) { s = ((unsigned char *)a)[x]; --x; total = total + RX_char_population (s); } return total; } dist/rx/rxcset.c100660 765 144 3552 6071416221 12475 0ustar lordusers/*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ /* * Tom Lord (lord@cygnus.com, lord@gnu.ai.mit.edu) */ #include "rxall.h" #include "rxcset.h" /* Utilities for manipulating bitset represntations of characters sets. */ #ifdef __STDC__ rx_Bitset rx_cset (int size) #else rx_Bitset rx_cset (size) int size; #endif { rx_Bitset b; b = (rx_Bitset) malloc (rx_sizeof_bitset (size)); if (b) rx_bitset_null (size, b); return b; } #ifdef __STDC__ rx_Bitset rx_copy_cset (int size, rx_Bitset a) #else rx_Bitset rx_copy_cset (size, a) int size; rx_Bitset a; #endif { rx_Bitset cs; cs = rx_cset (size); if (cs) rx_bitset_union (size, cs, a); return cs; } #ifdef __STDC__ void rx_free_cset (rx_Bitset c) #else void rx_free_cset (c) rx_Bitset c; #endif { if (c) free ((char *)c); } dist/rx/rxdbug.c100660 765 144 11546 6071416221 12502 0ustar lordusers/*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "rxall.h" #include "rxgnucomp.h" #include "rxnfa.h" #ifdef __GNUC__ #define AT(X) [X] = #else #define AT(X) #endif char *node_type_names[] = { AT(r_cset) "r_cset", AT(r_concat) "r_concat", AT(r_alternate) "r_alternate", AT(r_opt) "r_opt", AT(r_star) "r_star", AT(r_plus) "r_plus", AT(r_interval) "r_interval", AT(r_parens) "r_parens", AT(r_context) "r_conext" }; void print_cset (cset_size, cs) int cset_size; rx_Bitset cs; { int x; if (!cs) printf ("nil"); else { putchar ('['); for (x = 0; x < cset_size; ++x) if (RX_bitset_member (cs, x)) { if (isprint(x)) putchar (x); else printf ("\\0%o ", x); } putchar (']'); } } void spaces (n) int n; { while (n--) putchar (' '); } void print_rexp (cset_size, indent, rexp) int cset_size; int indent; struct rexp_node * rexp; { spaces (indent); if (!rexp) printf ("nil\n"); else { printf ("Node %d type %d (%s), iv=%d(%c), iv2=%d, len=%d obs=%d cs=", rexp->id, rexp->type, node_type_names[rexp->type], rexp->params.intval, (isprint (rexp->params.intval) ? rexp->params.intval : ' '), rexp->params.intval2, rexp->len, rexp->observed); print_cset (cset_size, rexp->params.cset); putchar ('\n'); if (rexp->params.pair.left || rexp->params.pair.right) { print_rexp (cset_size, indent + 2, rexp->params.pair.left); print_rexp (cset_size, indent + 2, rexp->params.pair.right); } } } void unparse_print_rexp (cset_size, rexp) int cset_size; struct rexp_node * rexp; { if (!rexp) return; else switch (rexp->type) { case r_cset: if (1 != rx_bitset_population (cset_size, rexp->params.cset)) print_cset (cset_size, rexp->params.cset); else { int x; rx_Bitset cs; cs = rexp->params.cset; for (x = 0; x < cset_size; ++x) if (RX_bitset_member (cs, x)) { if (isprint(x)) putchar (x); else printf ("\\0%o ", x); } } break; case r_parens: putchar ('('); unparse_print_rexp (cset_size, rexp->params.pair.left); putchar (')'); break; case r_context: putchar ('\\'); putchar (rexp->params.intval); break; case r_concat: unparse_print_rexp (cset_size, rexp->params.pair.left); unparse_print_rexp (cset_size, rexp->params.pair.right); break; case r_alternate: unparse_print_rexp (cset_size, rexp->params.pair.left); putchar ('|'); unparse_print_rexp (cset_size, rexp->params.pair.right); break; case r_opt: unparse_print_rexp (cset_size, rexp->params.pair.left); putchar ('?'); break; case r_star: unparse_print_rexp (cset_size, rexp->params.pair.left); putchar ('*'); break; case r_plus: unparse_print_rexp (cset_size, rexp->params.pair.left); putchar ('+'); break; case r_interval: unparse_print_rexp (cset_size, rexp->params.pair.left); printf ("{%d,%d}", rexp->params.intval, rexp->params.intval2); break; } } void print_nfa_state (rx, state) struct rx * rx; struct rx_nfa_state * state; { struct rx_nfa_edge * e; printf ("state %d, is_final %d, is_start %d\n", state->id, state->is_final, state->is_start); for (e = state->edges; e; e = e->next) { printf ("\tEdge %s to %d ", (e->type == ne_cset ? "cset" : (e->type == ne_epsilon ? "epsilon" : "side effect")), e->dest->id); if (e->type == ne_cset) print_cset (rx->local_cset_size, e->params.cset); else printf ("%d", (int)e->params.side_effect); putchar ('\n'); } } void print_nfa (rx) struct rx * rx; { struct rx_nfa_state * state; for (state = rx->nfa_states; state; state = state->next) print_nfa_state (rx, state); } dist/rx/rxgnucomp.c100660 765 144 115464 6071532113 13254 0ustar lordusers/* Copyright (C) 1992, 1993, 1994, 1995 Free Software Foundation, Inc. * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU Library General Public License as published by * the Free Software Foundation; either version 2, or (at your option) * any later version. * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU Library General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this software; see the file COPYING. If not, write to * the Free Software Foundation, 59 Temple Place - Suite 330, * Boston, MA 02111-1307, USA. */ #include "rxall.h" #include "rxgnucomp.h" /* {A Syntax Table} */ /* Define the syntax basics for \<, \>, etc. */ #ifndef emacs #define CHARBITS 8 #define CHAR_SET_SIZE (1 << CHARBITS) #define Sword 1 #define SYNTAX(c) re_syntax_table[c] char re_syntax_table[CHAR_SET_SIZE]; #ifdef __STDC__ static void init_syntax_once (void) #else static void init_syntax_once () #endif { register int c; static int done = 0; if (done) return; bzero (re_syntax_table, sizeof re_syntax_table); for (c = 'a'; c <= 'z'; c++) re_syntax_table[c] = Sword; for (c = 'A'; c <= 'Z'; c++) re_syntax_table[c] = Sword; for (c = '0'; c <= '9'; c++) re_syntax_table[c] = Sword; re_syntax_table['_'] = Sword; done = 1; } #endif /* not emacs */ const char *rx_error_msg[] = { 0, /* REG_NOUT */ "No match", /* REG_NOMATCH */ "Invalid regular expression", /* REG_BADPAT */ "Invalid collation character", /* REG_ECOLLATE */ "Invalid character class name", /* REG_ECTYPE */ "Trailing backslash", /* REG_EESCAPE */ "Invalid back reference", /* REG_ESUBREG */ "Unmatched [ or [^", /* REG_EBRACK */ "Unmatched ( or \\(", /* REG_EPAREN */ "Unmatched \\{", /* REG_EBRACE */ "Invalid content of \\{\\}", /* REG_BADBR */ "Invalid range end", /* REG_ERANGE */ "Memory exhausted", /* REG_ESPACE */ "Invalid preceding regular expression", /* REG_BADRPT */ "Premature end of regular expression", /* REG_EEND */ "Regular expression too big", /* REG_ESIZE */ "Unmatched ) or \\)", /* REG_ERPAREN */ }; /* * Macros used while compiling patterns. * * By convention, PEND points just past the end of the uncompiled pattern, * P points to the read position in the pattern. `translate' is the name * of the translation table (`TRANSLATE' is the name of a macro that looks * things up in `translate'). */ /* * Fetch the next character in the uncompiled pattern---translating it * if necessary. *Also cast from a signed character in the constant * string passed to us by the user to an unsigned char that we can use * as an array index (in, e.g., `translate'). */ #define PATFETCH(c) \ do {if (p == pend) return REG_EEND; \ c = (unsigned char) *p++; \ c = translate[c]; \ } while (0) /* * Fetch the next character in the uncompiled pattern, with no * translation. */ #define PATFETCH_RAW(c) \ do {if (p == pend) return REG_EEND; \ c = (unsigned char) *p++; \ } while (0) /* Go backwards one character in the pattern. */ #define PATUNFETCH p-- #define TRANSLATE(d) translate[(unsigned char) (d)] typedef unsigned regnum_t; /* Since offsets can go either forwards or backwards, this type needs to * be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */ typedef int pattern_offset_t; typedef struct { struct rexp_node ** top_expression; /* was begalt */ struct rexp_node ** last_expression; /* was laststart */ pattern_offset_t inner_group_offset; regnum_t regnum; } compile_stack_elt_t; typedef struct { compile_stack_elt_t *stack; unsigned size; unsigned avail; /* Offset of next open position. */ } compile_stack_type; #define INIT_COMPILE_STACK_SIZE 32 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0) #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size) /* The next available element. */ #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail]) /* Set the bit for character C in a list. */ #define SET_LIST_BIT(c) \ (b[((unsigned char) (c)) / CHARBITS] \ |= 1 << (((unsigned char) c) % CHARBITS)) /* Get the next unsigned number in the uncompiled pattern. */ #define GET_UNSIGNED_NUMBER(num) \ { if (p != pend) \ { \ PATFETCH (c); \ while (isdigit (c)) \ { \ if (num < 0) \ num = 0; \ num = num * 10 + c - '0'; \ if (p == pend) \ break; \ PATFETCH (c); \ } \ } \ } #define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */ #define IS_CHAR_CLASS(string) \ (!strcmp (string, "alpha") || !strcmp (string, "upper") \ || !strcmp (string, "lower") || !strcmp (string, "digit") \ || !strcmp (string, "alnum") || !strcmp (string, "xdigit") \ || !strcmp (string, "space") || !strcmp (string, "print") \ || !strcmp (string, "punct") || !strcmp (string, "graph") \ || !strcmp (string, "cntrl") || !strcmp (string, "blank")) /* These predicates are used in regex_compile. */ /* P points to just after a ^ in PATTERN. Return true if that ^ comes * after an alternative or a begin-subexpression. We assume there is at * least one character before the ^. */ #ifdef __STDC__ static int at_begline_loc_p (const char *pattern, const char * p, unsigned long syntax) #else static int at_begline_loc_p (pattern, p, syntax) const char *pattern; const char * p; unsigned long syntax; #endif { const char *prev = p - 2; int prev_prev_backslash = ((prev > pattern) && (prev[-1] == '\\')); return (/* After a subexpression? */ ((*prev == '(') && ((syntax & RE_NO_BK_PARENS) || prev_prev_backslash)) || /* After an alternative? */ ((*prev == '|') && ((syntax & RE_NO_BK_VBAR) || prev_prev_backslash)) ); } /* The dual of at_begline_loc_p. This one is for $. We assume there is * at least one character after the $, i.e., `P < PEND'. */ #ifdef __STDC__ static int at_endline_loc_p (const char *p, const char *pend, int syntax) #else static int at_endline_loc_p (p, pend, syntax) const char *p; const char *pend; int syntax; #endif { const char *next = p; int next_backslash = (*next == '\\'); const char *next_next = (p + 1 < pend) ? (p + 1) : 0; return ( /* Before a subexpression? */ ((syntax & RE_NO_BK_PARENS) ? (*next == ')') : (next_backslash && next_next && (*next_next == ')'))) || /* Before an alternative? */ ((syntax & RE_NO_BK_VBAR) ? (*next == '|') : (next_backslash && next_next && (*next_next == '|'))) ); } unsigned char rx_id_translation[256] = { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255 }; /* The compiler keeps an inverted translation table. * This looks up/inititalize elements. * VALID is an array of booleans that validate CACHE. */ #ifdef __STDC__ static rx_Bitset inverse_translation (int cset_size, char * valid, rx_Bitset cache, unsigned char * translate, int c) #else static rx_Bitset inverse_translation (cset_size, valid, cache, translate, c) int cset_size; char * valid; rx_Bitset cache; unsigned char * translate; int c; #endif { rx_Bitset cs; cs = cache + c * rx_bitset_numb_subsets (cset_size); if (!valid[c]) { int x; int c_tr = TRANSLATE(c); rx_bitset_null (cset_size, cs); for (x = 0; x < 256; ++x) /* &&&& 13.37 */ if (TRANSLATE(x) == c_tr) RX_bitset_enjoin (cs, x); valid[c] = 1; } return cs; } /* More subroutine declarations and macros for regex_compile. */ /* Returns true if REGNUM is in one of COMPILE_STACK's elements and * false if it's not. */ #ifdef __STDC__ static int group_in_compile_stack (compile_stack_type compile_stack, regnum_t regnum) #else static int group_in_compile_stack (compile_stack, regnum) compile_stack_type compile_stack; regnum_t regnum; #endif { int this_element; for (this_element = compile_stack.avail - 1; this_element >= 0; this_element--) if (compile_stack.stack[this_element].regnum == regnum) return 1; return 0; } /* * Read the ending character of a range (in a bracket expression) from the * uncompiled pattern *P_PTR (which ends at PEND). We assume the * starting character is in `P[-2]'. (`P[-1]' is the character `-'.) * Then we set the translation of all bits between the starting and * ending characters (inclusive) in the compiled pattern B. * * Return an error code. * * We use these short variable names so we can use the same macros as * `regex_compile' itself. */ #ifdef __STDC__ static reg_errcode_t compile_range (int cset_size, rx_Bitset cs, const char ** p_ptr, const char * pend, unsigned char * translate, unsigned long syntax, rx_Bitset inv_tr, char * valid_inv_tr) #else static reg_errcode_t compile_range (cset_size, cs, p_ptr, pend, translate, syntax, inv_tr, valid_inv_tr) int cset_size; rx_Bitset cs; const char ** p_ptr; const char * pend; unsigned char * translate; unsigned long syntax; rx_Bitset inv_tr; char * valid_inv_tr; #endif { unsigned this_char; const char *p = *p_ptr; unsigned char range_end; unsigned char range_start = TRANSLATE(p[-2]); if (p == pend) return REG_ERANGE; PATFETCH (range_end); (*p_ptr)++; if (range_start > range_end) return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR; for (this_char = range_start; this_char <= range_end; this_char++) { rx_Bitset it = inverse_translation (cset_size, valid_inv_tr, inv_tr, translate, this_char); rx_bitset_union (cset_size, cs, it); } return REG_NOERROR; } #ifdef __STDC__ static int pointless_if_repeated (struct rexp_node * node) #else static int pointless_if_repeated (node) struct rexp_node * node; #endif { if (!node) return 1; switch (node->type) { case r_cset: return 0; case r_concat: case r_alternate: return (pointless_if_repeated (node->params.pair.left) && pointless_if_repeated (node->params.pair.right)); case r_opt: case r_star: case r_interval: case r_parens: return pointless_if_repeated (node->params.pair.left); case r_context: switch (node->params.intval) { case '=': case '<': case '^': case 'b': case 'B': case '`': case '\'': return 1; default: return 0; } default: return 0; } } #define isa_blank(C) (((C) == ' ') || ((C) != '\t')) #ifdef __STDC__ reg_errcode_t rx_parse (struct rexp_node ** rexp_p, const char *pattern, int size, unsigned long syntax, int cset_size, unsigned char *translate) #else reg_errcode_t rx_parse (rexp_p, pattern, size, syntax, cset_size, translate) struct rexp_node ** rexp_p; const char *pattern; int size; unsigned long syntax; int cset_size; unsigned char *translate; #endif { reg_errcode_t compile_error; RX_subset inverse_translate [CHAR_SET_SIZE * rx_bitset_numb_subsets(CHAR_SET_SIZE)]; char validate_inv_tr [CHAR_SET_SIZE * rx_bitset_numb_subsets(CHAR_SET_SIZE)]; /* We fetch characters from PATTERN here. Even though PATTERN is `char *' (i.e., signed), we declare these variables as unsigned, so they can be reliably used as array indices. */ register unsigned char c; register unsigned char c1; /* A random tempory spot in PATTERN. */ const char *p1; /* Keeps track of unclosed groups. */ compile_stack_type compile_stack; /* Points to the current (ending) position in the pattern. */ const char *p; const char *pend; /* When parsing is done, this will hold the expression tree. */ struct rexp_node * rexp; /* This and top_expression are saved on the compile stack. */ struct rexp_node ** top_expression; struct rexp_node ** last_expression; /* Parameter to `goto append_node' */ struct rexp_node * append; /* Counts open-groups as they are encountered. This is the index of the * innermost group being compiled. */ regnum_t regnum; /* Place in the uncompiled pattern (i.e., the {) to * which to go back if the interval is invalid. */ const char *beg_interval; int side; if (!translate) translate = rx_id_translation; /* Points to the current (ending) position in the pattern. */ p = pattern; pend = pattern + size; /* When parsing is done, this will hold the expression tree. */ rexp = 0; /* In the midst of compilation, this holds onto the regexp * first parst while rexp goes on to aquire additional constructs. */ top_expression = &rexp; last_expression = top_expression; /* Counts open-groups as they are encountered. This is the index of the * innermost group being compiled. */ regnum = 0; bzero (validate_inv_tr, sizeof (validate_inv_tr)); /* Initialize the compile stack. */ compile_stack.stack = (( compile_stack_elt_t *) malloc ((INIT_COMPILE_STACK_SIZE) * sizeof ( compile_stack_elt_t))); if (compile_stack.stack == 0) return REG_ESPACE; compile_stack.size = INIT_COMPILE_STACK_SIZE; compile_stack.avail = 0; #if !defined (emacs) && !defined (SYNTAX_TABLE) /* Initialize the syntax table. */ init_syntax_once (); #endif /* Loop through the uncompiled pattern until we're at the end. */ while (p != pend) { PATFETCH (c); switch (c) { case '^': { if ( /* If at start of pattern, it's an operator. */ p == pattern + 1 /* If context independent, it's an operator. */ || syntax & RE_CONTEXT_INDEP_ANCHORS /* Otherwise, depends on what's come before. */ || at_begline_loc_p (pattern, p, syntax)) { struct rexp_node * n = rx_mk_r_int (r_context, '^'); if (!n) goto space_error; append = n; goto append_node; } else goto normal_char; } break; case '$': { if ( /* If at end of pattern, it's an operator. */ p == pend /* If context independent, it's an operator. */ || syntax & RE_CONTEXT_INDEP_ANCHORS /* Otherwise, depends on what's next. */ || at_endline_loc_p (p, pend, syntax)) { struct rexp_node * n = rx_mk_r_int (r_context, '$'); if (!n) goto space_error; append = n; goto append_node; } else goto normal_char; } break; case '+': case '?': if ((syntax & RE_BK_PLUS_QM) || (syntax & RE_LIMITED_OPS)) goto normal_char; handle_plus: case '*': /* If there is no previous pattern... */ if (pointless_if_repeated (*last_expression)) { if (syntax & RE_CONTEXT_INVALID_OPS) { compile_error = REG_BADRPT; goto error_return; } else if (!(syntax & RE_CONTEXT_INDEP_OPS)) goto normal_char; } { /* 1 means zero (many) matches is allowed. */ char zero_times_ok = 0, many_times_ok = 0; /* If there is a sequence of repetition chars, collapse it down to just one (the right one). We can't combine interval operators with these because of, e.g., `a{2}*', which should only match an even number of `a's. */ for (;;) { zero_times_ok |= c != '+'; many_times_ok |= c != '?'; if (p == pend) break; PATFETCH (c); if (c == '*' || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?'))) ; else if (syntax & RE_BK_PLUS_QM && c == '\\') { if (p == pend) { compile_error = REG_EESCAPE; goto error_return; } PATFETCH (c1); if (!(c1 == '+' || c1 == '?')) { PATUNFETCH; PATUNFETCH; break; } c = c1; } else { PATUNFETCH; break; } /* If we get here, we found another repeat character. */ } /* Now we know whether or not zero matches is allowed * and also whether or not two or more matches is allowed. */ { struct rexp_node * inner_exp; struct rexp_node * star; inner_exp = *last_expression; star = rx_mk_r_monop ((many_times_ok ? (zero_times_ok ? r_star : r_plus) : r_opt), inner_exp); if (!star) goto space_error; *last_expression = star; } } break; case '.': { rx_Bitset cs; struct rexp_node * n; cs = rx_cset (cset_size); if (!cs) goto space_error; n = rx_mk_r_cset (r_cset, cset_size, cs); if (!n) { rx_free_cset (cs); goto space_error; } rx_bitset_universe (cset_size, cs); if (!(syntax & RE_DOT_NEWLINE)) RX_bitset_remove (cs, '\n'); if (!(syntax & RE_DOT_NOT_NULL)) RX_bitset_remove (cs, 0); append = n; goto append_node; break; } case '[': if (p == pend) { compile_error = REG_EBRACK; goto error_return; } { int had_char_class; rx_Bitset cs; struct rexp_node * node; int is_inverted; had_char_class = 0; is_inverted = *p == '^'; cs = rx_cset (cset_size); if (!cs) goto space_error; node = rx_mk_r_cset (r_cset, cset_size ,cs); if (!node) { rx_free_cset (cs); goto space_error; } /* This branch of the switch is normally exited with *`goto append_node' */ append = node; if (is_inverted) p++; /* Remember the first position in the bracket expression. */ p1 = p; /* Read in characters and ranges, setting map bits. */ for (;;) { if (p == pend) { compile_error = REG_EBRACK; goto error_return; } PATFETCH (c); /* \ might escape characters inside [...] and [^...]. */ if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\') { if (p == pend) { compile_error = REG_EESCAPE; goto error_return; } PATFETCH (c1); { rx_Bitset it = inverse_translation (cset_size, validate_inv_tr, inverse_translate, translate, c1); rx_bitset_union (cset_size, cs, it); } continue; } /* Could be the end of the bracket expression. If it's not (i.e., when the bracket expression is `[]' so far), the ']' character bit gets set way below. */ if (c == ']' && p != p1 + 1) goto finalize_class_and_append; /* Look ahead to see if it's a range when the last thing was a character class. */ if (had_char_class && c == '-' && *p != ']') { compile_error = REG_ERANGE; goto error_return; } /* Look ahead to see if it's a range when the last thing was a character: if this is a hyphen not at the beginning or the end of a list, then it's the range operator. */ if (c == '-' && !(p - 2 >= pattern && p[-2] == '[') && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^') && *p != ']') { reg_errcode_t ret = compile_range (cset_size, cs, &p, pend, translate, syntax, inverse_translate, validate_inv_tr); if (ret != REG_NOERROR) { compile_error = ret; goto error_return; } } else if (p[0] == '-' && p[1] != ']') { /* This handles ranges made up of characters only. */ reg_errcode_t ret; /* Move past the `-'. */ PATFETCH (c1); ret = compile_range (cset_size, cs, &p, pend, translate, syntax, inverse_translate, validate_inv_tr); if (ret != REG_NOERROR) { compile_error = ret; goto error_return; } } /* See if we're at the beginning of a possible character class. */ else if ((syntax & RE_CHAR_CLASSES) && (c == '[') && (*p == ':')) { char str[CHAR_CLASS_MAX_LENGTH + 1]; PATFETCH (c); c1 = 0; /* If pattern is `[[:'. */ if (p == pend) { compile_error = REG_EBRACK; goto error_return; } for (;;) { PATFETCH (c); if (c == ':' || c == ']' || p == pend || c1 == CHAR_CLASS_MAX_LENGTH) break; str[c1++] = c; } str[c1] = '\0'; /* If isn't a word bracketed by `[:' and:`]': undo the ending character, the letters, and leave the leading `:' and `[' (but set bits for them). */ if (c == ':' && *p == ']') { int ch; int is_alnum = !strcmp (str, "alnum"); int is_alpha = !strcmp (str, "alpha"); int is_blank = !strcmp (str, "blank"); int is_cntrl = !strcmp (str, "cntrl"); int is_digit = !strcmp (str, "digit"); int is_graph = !strcmp (str, "graph"); int is_lower = !strcmp (str, "lower"); int is_print = !strcmp (str, "print"); int is_punct = !strcmp (str, "punct"); int is_space = !strcmp (str, "space"); int is_upper = !strcmp (str, "upper"); int is_xdigit = !strcmp (str, "xdigit"); if (!IS_CHAR_CLASS (str)) { compile_error = REG_ECTYPE; goto error_return; } /* Throw away the ] at the end of the character class. */ PATFETCH (c); if (p == pend) { compile_error = REG_EBRACK; goto error_return; } for (ch = 0; ch < 1 << CHARBITS; ch++) { if ( (is_alnum && isalnum (ch)) || (is_alpha && isalpha (ch)) || (is_blank && isa_blank (ch)) || (is_cntrl && iscntrl (ch)) || (is_digit && isdigit (ch)) || (is_graph && isgraph (ch)) || (is_lower && islower (ch)) || (is_print && isprint (ch)) || (is_punct && ispunct (ch)) || (is_space && isspace (ch)) || (is_upper && isupper (ch)) || (is_xdigit && isxdigit (ch))) { rx_Bitset it = inverse_translation (cset_size, validate_inv_tr, inverse_translate, translate, ch); rx_bitset_union (cset_size, cs, it); } } had_char_class = 1; } else { c1++; while (c1--) PATUNFETCH; { rx_Bitset it = inverse_translation (cset_size, validate_inv_tr, inverse_translate, translate, '['); rx_bitset_union (cset_size, cs, it); } { rx_Bitset it = inverse_translation (cset_size, validate_inv_tr, inverse_translate, translate, ':'); rx_bitset_union (cset_size, cs, it); } had_char_class = 0; } } else { had_char_class = 0; { rx_Bitset it = inverse_translation (cset_size, validate_inv_tr, inverse_translate, translate, c); rx_bitset_union (cset_size, cs, it); } } } finalize_class_and_append: if (is_inverted) { rx_bitset_complement (cset_size, cs); if (syntax & RE_HAT_LISTS_NOT_NEWLINE) RX_bitset_remove (cs, '\n'); } goto append_node; } break; case '(': if (syntax & RE_NO_BK_PARENS) goto handle_open; else goto normal_char; case ')': if (syntax & RE_NO_BK_PARENS) goto handle_close; else goto normal_char; case '\n': if (syntax & RE_NEWLINE_ALT) goto handle_alt; else goto normal_char; case '|': if (syntax & RE_NO_BK_VBAR) goto handle_alt; else goto normal_char; case '{': if ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) goto handle_interval; else goto normal_char; case '\\': if (p == pend) { compile_error = REG_EESCAPE; goto error_return; } /* Do not translate the character after the \, so that we can distinguish, e.g., \B from \b, even if we normally would translate, e.g., B to b. */ PATFETCH_RAW (c); switch (c) { case '(': if (syntax & RE_NO_BK_PARENS) goto normal_backslash; handle_open: regnum++; if (COMPILE_STACK_FULL) { compile_stack.stack = ((compile_stack_elt_t *) realloc (compile_stack.stack, (compile_stack.size << 1) * sizeof (compile_stack_elt_t))); if (compile_stack.stack == 0) goto space_error; compile_stack.size <<= 1; } if (*last_expression) { struct rexp_node * concat; concat = rx_mk_r_binop (r_concat, *last_expression, 0); if (!concat) goto space_error; *last_expression = concat; last_expression = &concat->params.pair.right; } /* * These are the values to restore when we hit end of this * group. */ COMPILE_STACK_TOP.top_expression = top_expression; COMPILE_STACK_TOP.last_expression = last_expression; COMPILE_STACK_TOP.regnum = regnum; compile_stack.avail++; top_expression = last_expression; break; case ')': if (syntax & RE_NO_BK_PARENS) goto normal_backslash; handle_close: /* See similar code for backslashed left paren above. */ if (COMPILE_STACK_EMPTY) if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD) goto normal_char; else { compile_error = REG_ERPAREN; goto error_return; } /* Since we just checked for an empty stack above, this * ``can't happen''. */ { /* We don't just want to restore into `regnum', because * later groups should continue to be numbered higher, * as in `(ab)c(de)' -- the second group is #2. */ regnum_t this_group_regnum; struct rexp_node ** inner; struct rexp_node * parens; inner = top_expression; compile_stack.avail--; top_expression = COMPILE_STACK_TOP.top_expression; last_expression = COMPILE_STACK_TOP.last_expression; this_group_regnum = COMPILE_STACK_TOP.regnum; parens = rx_mk_r_monop (r_parens, *inner); if (!parens) goto space_error; parens->params.intval = this_group_regnum; *inner = parens; break; } case '|': /* `\|'. */ if ((syntax & RE_LIMITED_OPS) || (syntax & RE_NO_BK_VBAR)) goto normal_backslash; handle_alt: if (syntax & RE_LIMITED_OPS) goto normal_char; { struct rexp_node * alt; alt = rx_mk_r_binop (r_alternate, *top_expression, 0); if (!alt) goto space_error; *top_expression = alt; last_expression = &alt->params.pair.right; } break; case '{': /* If \{ is a literal. */ if (!(syntax & RE_INTERVALS) /* If we're at `\{' and it's not the open-interval operator. */ || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) || (p - 2 == pattern && p == pend)) goto normal_backslash; handle_interval: { /* If got here, then the syntax allows intervals. */ /* At least (most) this many matches must be made. */ int lower_bound; int upper_bound; lower_bound = -1; upper_bound = -1; /* We're about to parse the bounds of the interval. * It may turn out that this isn't an interval after * all, in which case these same characters will have * to be reparsed as literals. This remembers * the backtrack point in the parse: */ beg_interval = p - 1; if (p == pend) { if (syntax & RE_NO_BK_BRACES) goto unfetch_interval; else { compile_error = REG_EBRACE; goto error_return; } } GET_UNSIGNED_NUMBER (lower_bound); if (c == ',') { GET_UNSIGNED_NUMBER (upper_bound); if (upper_bound < 0) upper_bound = RE_DUP_MAX; } else /* Interval such as `{n}' => match exactly n times. */ upper_bound = lower_bound; if (lower_bound < 0 || upper_bound > RE_DUP_MAX || lower_bound > upper_bound) { if (syntax & RE_NO_BK_BRACES) goto unfetch_interval; else { compile_error = REG_BADBR; goto error_return; } } if (!(syntax & RE_NO_BK_BRACES)) { if (c != '\\') { compile_error = REG_EBRACE; goto error_return; } PATFETCH (c); } if (c != '}') { if (syntax & RE_NO_BK_BRACES) goto unfetch_interval; else { compile_error = REG_BADBR; goto error_return; } } /* We just parsed a valid interval. * lower_bound and upper_bound are set. */ /* If it's invalid to have no preceding re. */ if (pointless_if_repeated (*last_expression)) { if (syntax & RE_CONTEXT_INVALID_OPS) { compile_error = REG_BADRPT; goto error_return; } else if (!(syntax & RE_CONTEXT_INDEP_OPS)) /* treat the interval spec as literal chars. */ goto unfetch_interval; } { struct rexp_node * interval; interval = rx_mk_r_monop (r_interval, *last_expression); if (!interval) goto space_error; interval->params.intval = lower_bound; interval->params.intval2 = upper_bound; *last_expression = interval; } beg_interval = 0; } break; unfetch_interval: /* If an invalid interval, match the characters as literals. */ p = beg_interval; beg_interval = 0; /* normal_char and normal_backslash need `c'. */ PATFETCH (c); if (!(syntax & RE_NO_BK_BRACES)) { if ((p > pattern) && (p[-1] == '\\')) goto normal_backslash; } goto normal_char; #ifdef emacs /* There is no way to specify the before_dot and after_dot * operators. rms says this is ok. --karl */ case '=': side = '='; goto add_side_effect; break; case 's': case 'S': { rx_Bitset cs; struct rexp_node * set; cs = rx_cset (&cset_size); if (!cs) goto space_error; set = rx_mk_r_cset (r_cset, cset_size, cs); if (!set) { rx_free_cset (cs); goto space_error; } if (c == 'S') rx_bitset_universe (cset_size, cs); PATFETCH (c); { int x; enum syntaxcode code = syntax_spec_code [c]; for (x = 0; x < 256; ++x) { if (SYNTAX (x) == code) { rx_Bitset it = inverse_translation (cset_size, validate_inv_tr, inverse_translate, translate, x); rx_bitset_xor (cset_size, cs, it); } } } append = set; goto append_node; } break; #endif /* emacs */ case 'w': case 'W': { rx_Bitset cs; struct rexp_node * n; cs = rx_cset (cset_size); n = (cs ? rx_mk_r_cset (r_cset, cset_size, cs) : 0); if (!(cs && n)) { if (cs) rx_free_cset (cs); goto space_error; } if (c == 'W') rx_bitset_universe (cset_size ,cs); { int x; for (x = cset_size - 1; x > 0; --x) if (SYNTAX(x) & Sword) RX_bitset_toggle (cs, x); } append = n; goto append_node; } break; case '<': side = '<'; goto add_side_effect; break; case '>': side = '>'; goto add_side_effect; break; case 'b': side = 'b'; goto add_side_effect; break; case 'B': side = 'B'; goto add_side_effect; break; case '`': side = '`'; goto add_side_effect; break; case '\'': side = '\''; goto add_side_effect; break; add_side_effect: { struct rexp_node * se; se = rx_mk_r_int (r_context, side); if (!se) goto space_error; append = se; goto append_node; } break; case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9': if (syntax & RE_NO_BK_REFS) goto normal_char; c1 = c - '0'; if (c1 > regnum) { compile_error = REG_ESUBREG; goto error_return; } side = c; goto add_side_effect; break; case '+': case '?': if (syntax & RE_BK_PLUS_QM) goto handle_plus; else goto normal_backslash; default: normal_backslash: /* You might think it would be useful for \ to mean * not to translate; but if we don't translate it * it will never match anything. */ c = TRANSLATE (c); goto normal_char; } break; default: /* Expects the character in `c'. */ normal_char: { rx_Bitset cs; struct rexp_node * match; rx_Bitset it; cs = rx_cset (cset_size); match = (cs ? rx_mk_r_cset (r_cset, cset_size, cs) : 0); if (!(cs && match)) { if (cs) rx_free_cset (cs); goto space_error; } it = inverse_translation (cset_size, validate_inv_tr, inverse_translate, translate, c); rx_bitset_union (CHAR_SET_SIZE, cs, it); append = match; append_node: /* This genericly appends the rexp APPEND to *LAST_EXPRESSION * and then parses the next character normally. */ if (*last_expression) { struct rexp_node * concat; concat = rx_mk_r_binop (r_concat, *last_expression, append); if (!concat) goto space_error; *last_expression = concat; last_expression = &concat->params.pair.right; } else *last_expression = append; } } /* switch (c) */ } /* while p != pend */ /* Through the pattern now. */ if (!COMPILE_STACK_EMPTY) { compile_error = REG_EPAREN; goto error_return; } free (compile_stack.stack); *rexp_p = rexp; return REG_NOERROR; space_error: compile_error = REG_ESPACE; error_return: free (compile_stack.stack); /* Free expressions pushed onto the compile stack! */ if (rexp) rx_free_rexp (rexp); return compile_error; } dist/rx/rxhash.c100660 765 144 22323 6071416221 12477 0ustar lordusers/*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ /* * Tom Lord (lord@cygnus.com, lord@gnu.ai.mit.edu) */ #include "rxall.h" #include "rxhash.h" #ifdef __STDC__ static struct rx_hash * default_hash_alloc (struct rx_hash_rules * rules) #else static struct rx_hash * default_hash_alloc (rules) struct rx_hash_rules * rules; #endif { return (struct rx_hash *)malloc (sizeof (struct rx_hash)); } #ifdef __STDC__ static struct rx_hash_item * default_hash_item_alloc (struct rx_hash_rules * rules, void * value) #else static struct rx_hash_item * default_hash_item_alloc (rules, value) struct rx_hash_rules * rules; void * value; #endif { struct rx_hash_item * it; it = (struct rx_hash_item *)malloc (sizeof (*it)); if (it) { it->data = value; it->binding = 0; } return it; } #ifdef __STDC__ static void default_free_hash (struct rx_hash * tab, struct rx_hash_rules * rules) #else static void default_free_hash (tab, rules) struct rx_hash * tab; struct rx_hash_rules * rules; #endif { free ((char *)tab); } #ifdef __STDC__ static void default_free_hash_item (struct rx_hash_item * item, struct rx_hash_rules * rules) #else static void default_free_hash_item (item, rules) struct rx_hash_item * item; struct rx_hash_rules * rules; #endif { free ((char *)item); } #ifdef __STDC__ static int default_eq (void * va, void * vb) #else static int default_eq (va, vb) void * va; void * vb; #endif { return va == vb; } #define EQ(rules) ((rules && rules->eq) ? rules->eq : default_eq) #define HASH_ALLOC(rules) ((rules && rules->hash_alloc) ? rules->hash_alloc : default_hash_alloc) #define FREE_HASH(rules) ((rules && rules->free_hash) ? rules->free_hash : default_free_hash) #define ITEM_ALLOC(rules) ((rules && rules->hash_item_alloc) ? rules->hash_item_alloc : default_hash_item_alloc) #define FREE_HASH_ITEM(rules) ((rules && rules->free_hash_item) ? rules->free_hash_item : default_free_hash_item) static unsigned long rx_hash_masks[4] = { 0x12488421, 0x96699669, 0xbe7dd7eb, 0xffffffff }; /* hash to bucket */ #define JOIN_BYTE(H, B) (((H) + ((H) << 3) + (B)) & 0xf) #define H2B(X) JOIN_BYTE (JOIN_BYTE (JOIN_BYTE ((X & 0xf), ((X>>8) & 0xf)), ((X>>16) & 0xf)), ((X>>24) & 0xf)) #define BKTS 16 /* Hash tables */ #ifdef __STDC__ struct rx_hash_item * rx_hash_find (struct rx_hash * table, unsigned long hash, void * value, struct rx_hash_rules * rules) #else struct rx_hash_item * rx_hash_find (table, hash, value, rules) struct rx_hash * table; unsigned long hash; void * value; struct rx_hash_rules * rules; #endif { rx_hash_eq eq = EQ (rules); int maskc = 0; long mask = rx_hash_masks [0]; int bucket = H2B(hash & mask); while (RX_bitset_member (&table->nested_p, bucket)) { table = (struct rx_hash *)(table->children [bucket]); ++maskc; mask = rx_hash_masks[maskc]; bucket = H2B (hash & mask); } { struct rx_hash_item * it; it = (struct rx_hash_item *)(table->children[bucket]); while (it) if (eq (it->data, value)) return it; else it = it->next_same_hash; } return 0; } #ifdef __STDC__ static int listlen (struct rx_hash_item * bucket) #else static int listlen (bucket) struct rx_hash_item * bucket; #endif { int i; for (i = 0; bucket; ++i, bucket = bucket->next_same_hash) ; return i; } #ifdef __STDC__ static int overflows (struct rx_hash_item * bucket) #else static int overflows (bucket) struct rx_hash_item * bucket; #endif { return !( bucket && bucket->next_same_hash && bucket->next_same_hash->next_same_hash && bucket->next_same_hash->next_same_hash->next_same_hash); } #ifdef __STDC__ struct rx_hash_item * rx_hash_store (struct rx_hash * table, unsigned long hash, void * value, struct rx_hash_rules * rules) #else struct rx_hash_item * rx_hash_store (table, hash, value, rules) struct rx_hash * table; unsigned long hash; void * value; struct rx_hash_rules * rules; #endif { rx_hash_eq eq = EQ (rules); int maskc = 0; long mask = rx_hash_masks [0]; int bucket = H2B(hash & mask); int depth = 0; while (RX_bitset_member (&table->nested_p, bucket)) { table = (struct rx_hash *)(table->children [bucket]); ++maskc; mask = rx_hash_masks[maskc]; bucket = H2B(hash & mask); ++depth; } { struct rx_hash_item * it; it = (struct rx_hash_item *)(table->children[bucket]); while (it) if (eq (it->data, value)) return it; else it = it->next_same_hash; } { if ( (depth < 3) && (overflows ((struct rx_hash_item *)table->children [bucket]))) { struct rx_hash * newtab; newtab = (struct rx_hash *) HASH_ALLOC(rules) (rules); if (!newtab) goto add_to_bucket; bzero (newtab, sizeof (*newtab)); newtab->parent = table; { int old_bucket_size; struct rx_hash_item * them; unsigned long newmask; them = (struct rx_hash_item *)table->children[bucket]; newmask = rx_hash_masks[maskc + 1]; while (them) { struct rx_hash_item * save = them->next_same_hash; int new_buck = H2B(them->hash & newmask); them->next_same_hash = ((struct rx_hash_item *) newtab->children[new_buck]); ((struct rx_hash_item **)newtab->children)[new_buck] = them; them->table = newtab; them = save; ++newtab->refs; --table->refs; } ((struct rx_hash **)table->children)[bucket] = newtab; RX_bitset_enjoin (&table->nested_p, bucket); ++table->refs; table = newtab; bucket = H2B(hash & newmask); } } } add_to_bucket: { struct rx_hash_item * it = ((struct rx_hash_item *) ITEM_ALLOC(rules) (rules, value)); if (!it) return 0; it->hash = hash; it->table = table; /* DATA and BINDING are to be set in hash_item_alloc */ it->next_same_hash = (struct rx_hash_item *)table->children [bucket]; ((struct rx_hash_item **)table->children)[bucket] = it; ++table->refs; return it; } } #ifdef __STDC__ void rx_hash_free (struct rx_hash_item * it, struct rx_hash_rules * rules) #else void rx_hash_free (it, rules) struct rx_hash_item * it; struct rx_hash_rules * rules; #endif { if (it) { struct rx_hash * table = it->table; unsigned long hash = it->hash; int depth = (table->parent ? (table->parent->parent ? (table->parent->parent->parent ? 3 : 2) : 1) : 0); int bucket = H2B (hash & rx_hash_masks [depth]); struct rx_hash_item ** pos = (struct rx_hash_item **)&table->children [bucket]; while (*pos != it) pos = &(*pos)->next_same_hash; *pos = it->next_same_hash; FREE_HASH_ITEM(rules) (it, rules); --table->refs; while (!table->refs && depth) { struct rx_hash * save = table; table = table->parent; --depth; bucket = H2B(hash & rx_hash_masks [depth]); --table->refs; table->children[bucket] = 0; RX_bitset_remove (&table->nested_p, bucket); FREE_HASH (rules) (save, rules); } } } #ifdef __STDC__ void rx_free_hash_table (struct rx_hash * tab, rx_hash_freefn freefn, struct rx_hash_rules * rules) #else void rx_free_hash_table (tab, freefn, rules) struct rx_hash * tab; rx_hash_freefn freefn; struct rx_hash_rules * rules; #endif { int x; for (x = 0; x < BKTS; ++x) if (RX_bitset_member (&tab->nested_p, x)) { rx_free_hash_table ((struct rx_hash *)tab->children[x], freefn, rules); FREE_HASH (rules) ((struct rx_hash *)tab->children[x], rules); } else { struct rx_hash_item * them = (struct rx_hash_item *)tab->children[x]; while (them) { struct rx_hash_item * that = them; them = that->next_same_hash; freefn (that); FREE_HASH_ITEM (rules) (that, rules); } } } #ifdef __STDC__ int rx_count_hash_nodes (struct rx_hash * st) #else int rx_count_hash_nodes (st) struct rx_hash * st; #endif { int x; int count = 0; for (x = 0; x < BKTS; ++x) count += ((RX_bitset_member (&st->nested_p, x)) ? rx_count_hash_nodes ((struct rx_hash *)st->children[x]) : listlen ((struct rx_hash_item *)(st->children[x]))); return count; } dist/rx/rxnfa.c100660 765 144 43266 6071416221 12331 0ustar lordusers/*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ /* * Tom Lord (lord@cygnus.com, lord@gnu.ai.mit.edu) */ #include "rxall.h" #include "rxnfa.h" #ifndef realloc extern char * realloc (); #endif #ifndef malloc extern char * malloc (); #endif #define remalloc(M, S) (M ? realloc (M, S) : malloc (S)) /* {Low Level Data Structure Code} */ /* Constructs a new nfa node. */ #ifdef __STDC__ struct rx_nfa_state * rx_nfa_state (struct rx *rx) #else struct rx_nfa_state * rx_nfa_state (rx) struct rx *rx; #endif { struct rx_nfa_state * n = (struct rx_nfa_state *)malloc (sizeof (*n)); if (!n) return 0; bzero (n, sizeof (*n)); n->next = rx->nfa_states; rx->nfa_states = n; return n; } #ifdef __STDC__ static void rx_free_nfa_state (struct rx_nfa_state * n) #else static void rx_free_nfa_state (n) struct rx_nfa_state * n; #endif { free ((char *)n); } /* This adds an edge between two nodes, but doesn't initialize the * edge label. */ #ifdef __STDC__ struct rx_nfa_edge * rx_nfa_edge (struct rx *rx, enum rx_nfa_etype type, struct rx_nfa_state *start, struct rx_nfa_state *dest) #else struct rx_nfa_edge * rx_nfa_edge (rx, type, start, dest) struct rx *rx; enum rx_nfa_etype type; struct rx_nfa_state *start; struct rx_nfa_state *dest; #endif { struct rx_nfa_edge *e; e = (struct rx_nfa_edge *)malloc (sizeof (*e)); if (!e) return 0; e->next = start->edges; start->edges = e; e->type = type; e->dest = dest; return e; } #ifdef __STDC__ static void rx_free_nfa_edge (struct rx_nfa_edge * e) #else static void rx_free_nfa_edge (e) struct rx_nfa_edge * e; #endif { free ((char *)e); } /* This constructs a POSSIBLE_FUTURE, which is a kind epsilon-closure * of an NFA. These are added to an nfa automaticly by eclose_nfa. */ #ifdef __STDC__ static struct rx_possible_future * rx_possible_future (struct rx * rx, struct rx_se_list * effects) #else static struct rx_possible_future * rx_possible_future (rx, effects) struct rx * rx; struct rx_se_list * effects; #endif { struct rx_possible_future *ec; ec = (struct rx_possible_future *) malloc (sizeof (*ec)); if (!ec) return 0; ec->destset = 0; ec->next = 0; ec->effects = effects; return ec; } #ifdef __STDC__ static void rx_free_possible_future (struct rx_possible_future * pf) #else static void rx_free_possible_future (pf) struct rx_possible_future * pf; #endif { free ((char *)pf); } #ifdef __STDC__ static void rx_free_nfa_graph (struct rx *rx) #else static void rx_free_nfa_graph (rx) struct rx *rx; #endif { while (rx->nfa_states) { while (rx->nfa_states->edges) { switch (rx->nfa_states->edges->type) { case ne_cset: rx_free_cset (rx->nfa_states->edges->params.cset); break; default: break; } { struct rx_nfa_edge * e; e = rx->nfa_states->edges; rx->nfa_states->edges = rx->nfa_states->edges->next; rx_free_nfa_edge (e); } } /* while (rx->nfa_states->edges) */ { /* Iterate over the partial epsilon closures of rx->nfa_states */ struct rx_possible_future * pf = rx->nfa_states->futures; while (pf) { struct rx_possible_future * pft = pf; pf = pf->next; rx_free_possible_future (pft); } } { struct rx_nfa_state *n; n = rx->nfa_states; rx->nfa_states = rx->nfa_states->next; rx_free_nfa_state (n); } } } /* {Translating a Syntax Tree into an NFA} * */ /* This is the Thompson regexp->nfa algorithm. * It is modified to allow for `side-effect epsilons.' Those are * edges that are taken whenever a similarly placed epsilon edge * would be, but which also imply that some side effect occurs * when the edge is taken. * * Side effects are used to model parts of the pattern langauge * that are not regular. */ #ifdef __STDC__ int rx_build_nfa (struct rx *rx, struct rexp_node *rexp, struct rx_nfa_state **start, struct rx_nfa_state **end) #else int rx_build_nfa (rx, rexp, start, end) struct rx *rx; struct rexp_node *rexp; struct rx_nfa_state **start; struct rx_nfa_state **end; #endif { struct rx_nfa_edge *edge; /* Start & end nodes may have been allocated by the caller. */ *start = *start ? *start : rx_nfa_state (rx); if (!*start) return 0; if (!rexp) { *end = *start; return 1; } *end = *end ? *end : rx_nfa_state (rx); if (!*end) { rx_free_nfa_state (*start); return 0; } switch (rexp->type) { case r_cset: edge = rx_nfa_edge (rx, ne_cset, *start, *end); if (!edge) return 0; edge->params.cset = rx_copy_cset (rx->local_cset_size, rexp->params.cset); if (!edge->params.cset) { rx_free_nfa_edge (edge); return 0; } return 1; case r_opt: return (rx_build_nfa (rx, rexp->params.pair.left, start, end) && rx_nfa_edge (rx, ne_epsilon, *start, *end)); case r_plus: { struct rx_nfa_state * star_start = 0; struct rx_nfa_state * star_end = 0; struct rx_nfa_state * shared; shared = 0; if (!rx_build_nfa (rx, rexp->params.pair.left, start, &shared)) return 0; return (rx_build_nfa (rx, rexp->params.pair.left, &star_start, &star_end) && star_start && star_end && rx_nfa_edge (rx, ne_epsilon, star_start, star_end) && rx_nfa_edge (rx, ne_epsilon, shared, star_start) && rx_nfa_edge (rx, ne_epsilon, star_end, *end) && rx_nfa_edge (rx, ne_epsilon, star_end, star_start)); } case r_interval: case r_star: { struct rx_nfa_state * star_start = 0; struct rx_nfa_state * star_end = 0; return (rx_build_nfa (rx, rexp->params.pair.left, &star_start, &star_end) && star_start && star_end && rx_nfa_edge (rx, ne_epsilon, star_start, star_end) && rx_nfa_edge (rx, ne_epsilon, *start, star_start) && rx_nfa_edge (rx, ne_epsilon, star_end, *end) && rx_nfa_edge (rx, ne_epsilon, star_end, star_start)); } case r_parens: return rx_build_nfa (rx, rexp->params.pair.left, start, end); case r_concat: { struct rx_nfa_state *shared = 0; return (rx_build_nfa (rx, rexp->params.pair.left, start, &shared) && rx_build_nfa (rx, rexp->params.pair.right, &shared, end)); } case r_alternate: { struct rx_nfa_state *ls = 0; struct rx_nfa_state *le = 0; struct rx_nfa_state *rs = 0; struct rx_nfa_state *re = 0; return (rx_build_nfa (rx, rexp->params.pair.left, &ls, &le) && rx_build_nfa (rx, rexp->params.pair.right, &rs, &re) && rx_nfa_edge (rx, ne_epsilon, *start, ls) && rx_nfa_edge (rx, ne_epsilon, *start, rs) && rx_nfa_edge (rx, ne_epsilon, le, *end) && rx_nfa_edge (rx, ne_epsilon, re, *end)); } case r_context: edge = rx_nfa_edge (rx, ne_side_effect, *start, *end); if (!edge) return 0; edge->params.side_effect = (void *)rexp->params.intval; return 1; } /* this should never happen */ return 0; } /* {Low Level Data Structures for the Static NFA->SuperNFA Analysis} * * There are side effect lists -- lists of side effects occuring * along an uninterrupted, acyclic path of side-effect epsilon edges. * Such paths are collapsed to single edges in the course of computing * epsilon closures. The resulting single edges are labled with a list * of all the side effects from the original multi-edge path. Equivalent * lists of side effects are made == by the constructors below. * * There are also nfa state sets. These are used to hold a list of all * states reachable from a starting state for a given type of transition * and side effect list. These are also hash-consed. */ /* The next several functions compare, construct, etc. lists of side * effects. See ECLOSE_NFA (below) for details. */ /* Ordering of rx_se_list * (-1, 0, 1 return value convention). */ #ifdef __STDC__ static int se_list_cmp (void * va, void * vb) #else static int se_list_cmp (va, vb) void * va; void * vb; #endif { struct rx_se_list * a = (struct rx_se_list *)va; struct rx_se_list * b = (struct rx_se_list *)vb; return ((va == vb) ? 0 : (!va ? -1 : (!vb ? 1 : ((long)a->car < (long)b->car ? 1 : ((long)a->car > (long)b->car ? -1 : se_list_cmp ((void *)a->cdr, (void *)b->cdr)))))); } #ifdef __STDC__ static int se_list_equal (void * va, void * vb) #else static int se_list_equal (va, vb) void * va; void * vb; #endif { return !(se_list_cmp (va, vb)); } static struct rx_hash_rules se_list_hash_rules = { se_list_equal, 0, 0, 0, 0 }; #ifdef __STDC__ static struct rx_se_list * side_effect_cons (struct rx * rx, void * se, struct rx_se_list * list) #else static struct rx_se_list * side_effect_cons (rx, se, list) struct rx * rx; void * se; struct rx_se_list * list; #endif { struct rx_se_list * l; l = ((struct rx_se_list *) malloc (sizeof (*l))); if (!l) return 0; l->car = se; l->cdr = list; return l; } #ifdef __STDC__ static struct rx_se_list * hash_cons_se_prog (struct rx * rx, struct rx_hash * memo, void * car, struct rx_se_list * cdr) #else static struct rx_se_list * hash_cons_se_prog (rx, memo, car, cdr) struct rx * rx; struct rx_hash * memo; void * car; struct rx_se_list * cdr; #endif { long hash = (long)car ^ (long)cdr; struct rx_se_list template; template.car = car; template.cdr = cdr; { struct rx_hash_item * it = rx_hash_store (memo, hash, (void *)&template, &se_list_hash_rules); if (!it) return 0; if (it->data == (void *)&template) { struct rx_se_list * consed; consed = (struct rx_se_list *) malloc (sizeof (*consed)); *consed = template; it->data = (void *)consed; } return (struct rx_se_list *)it->data; } } #ifdef __STDC__ static struct rx_se_list * hash_se_prog (struct rx * rx, struct rx_hash * memo, struct rx_se_list * prog) #else static struct rx_se_list * hash_se_prog (rx, memo, prog) struct rx * rx; struct rx_hash * memo; struct rx_se_list * prog; #endif { struct rx_se_list * answer = 0; while (prog) { answer = hash_cons_se_prog (rx, memo, prog->car, answer); if (!answer) return 0; prog = prog->cdr; } return answer; } /* {Constructors, etc. for NFA State Sets} */ #ifdef __STDC__ static int nfa_set_cmp (void * va, void * vb) #else static int nfa_set_cmp (va, vb) void * va; void * vb; #endif { struct rx_nfa_state_set * a = (struct rx_nfa_state_set *)va; struct rx_nfa_state_set * b = (struct rx_nfa_state_set *)vb; return ((va == vb) ? 0 : (!va ? -1 : (!vb ? 1 : (a->car->id < b->car->id ? 1 : (a->car->id > b->car->id ? -1 : nfa_set_cmp ((void *)a->cdr, (void *)b->cdr)))))); } #ifdef __STDC__ static int nfa_set_equal (void * va, void * vb) #else static int nfa_set_equal (va, vb) void * va; void * vb; #endif { return !nfa_set_cmp (va, vb); } static struct rx_hash_rules nfa_set_hash_rules = { nfa_set_equal, 0, 0, 0, 0 }; #ifdef __STDC__ static struct rx_nfa_state_set * nfa_set_cons (struct rx * rx, struct rx_hash * memo, struct rx_nfa_state * state, struct rx_nfa_state_set * set) #else static struct rx_nfa_state_set * nfa_set_cons (rx, memo, state, set) struct rx * rx; struct rx_hash * memo; struct rx_nfa_state * state; struct rx_nfa_state_set * set; #endif { struct rx_nfa_state_set template; struct rx_hash_item * node; template.car = state; template.cdr = set; node = rx_hash_store (memo, (((long)state) >> 8) ^ (long)set, &template, &nfa_set_hash_rules); if (!node) return 0; if (node->data == &template) { struct rx_nfa_state_set * l; l = (struct rx_nfa_state_set *) malloc (sizeof (*l)); node->data = (void *) l; if (!l) return 0; *l = template; } return (struct rx_nfa_state_set *)node->data; } #ifdef __STDC__ static struct rx_nfa_state_set * nfa_set_enjoin (struct rx * rx, struct rx_hash * memo, struct rx_nfa_state * state, struct rx_nfa_state_set * set) #else static struct rx_nfa_state_set * nfa_set_enjoin (rx, memo, state, set) struct rx * rx; struct rx_hash * memo; struct rx_nfa_state * state; struct rx_nfa_state_set * set; #endif { if (!set || state->id < set->car->id) return nfa_set_cons (rx, memo, state, set); if (state->id == set->car->id) return set; else { struct rx_nfa_state_set * newcdr = nfa_set_enjoin (rx, memo, state, set->cdr); if (newcdr != set->cdr) set = nfa_set_cons (rx, memo, set->car, newcdr); return set; } } /* {Computing Epsilon/Side-effect Closures.} */ struct eclose_frame { struct rx_se_list *prog_backwards; }; /* This is called while computing closures for "outnode". * The current node in the traversal is "node". * "frame" contains a list of a all side effects between * "outnode" and "node" from most to least recent. * * Explores forward from "node", adding new possible * futures to outnode. * * Returns 0 on allocation failure. */ #ifdef __STDC__ static int eclose_node (struct rx *rx, struct rx_nfa_state *outnode, struct rx_nfa_state *node, struct eclose_frame *frame) #else static int eclose_node (rx, outnode, node, frame) struct rx *rx; struct rx_nfa_state *outnode; struct rx_nfa_state *node; struct eclose_frame *frame; #endif { struct rx_nfa_edge *e = node->edges; /* For each node, we follow all epsilon paths to build the closure. * The closure omits nodes that have only epsilon edges. * The closure is split into partial closures -- all the states in * a partial closure are reached by crossing the same list of * of side effects (though not necessarily the same path). */ if (node->mark) return 1; node->mark = 1; /* If "node" has more than just epsilon and * and side-effect transitions (id >= 0), or is final, * then it has to be added to the possible futures * of "outnode". */ if (node->id >= 0 || node->is_final) { struct rx_possible_future **ec; struct rx_se_list * prog_in_order; int cmp; prog_in_order = ((struct rx_se_list *)hash_se_prog (rx, &rx->se_list_memo, frame->prog_backwards)); ec = &outnode->futures; while (*ec) { cmp = se_list_cmp ((void *)(*ec)->effects, (void *)prog_in_order); if (cmp <= 0) break; ec = &(*ec)->next; } if (!*ec || (cmp < 0)) { struct rx_possible_future * pf; pf = rx_possible_future (rx, prog_in_order); if (!pf) return 0; pf->next = *ec; *ec = pf; } if (node->id >= 0) { (*ec)->destset = nfa_set_enjoin (rx, &rx->set_list_memo, node, (*ec)->destset); if (!(*ec)->destset) return 0; } } /* Recurse on outgoing epsilon and side effect nodes. */ while (e) { switch (e->type) { case ne_epsilon: if (!eclose_node (rx, outnode, e->dest, frame)) return 0; break; case ne_side_effect: { frame->prog_backwards = side_effect_cons (rx, e->params.side_effect, frame->prog_backwards); if (!frame->prog_backwards) return 0; if (!eclose_node (rx, outnode, e->dest, frame)) return 0; { struct rx_se_list * dying = frame->prog_backwards; frame->prog_backwards = frame->prog_backwards->cdr; free ((char *)dying); } break; } default: break; } e = e->next; } node->mark = 0; return 1; } #ifdef __STDC__ struct rx_possible_future * rx_state_possible_futures (struct rx * rx, struct rx_nfa_state * n) #else struct rx_possible_future * rx_state_possible_futures (rx, n) struct rx * rx; struct rx_nfa_state * n; #endif { if (n->futures_computed) return n->futures; else { struct eclose_frame frame; frame.prog_backwards = 0; if (!eclose_node (rx, n, n, &frame)) return 0; else { n->futures_computed = 1; return n->futures; } } } /* {Storing the NFA in a Contiguous Region of Memory} */ #ifdef __STDC__ static void se_memo_freer (struct rx_hash_item * node) #else static void se_memo_freer (node) struct rx_hash_item * node; #endif { free ((char *)node->data); } #ifdef __STDC__ static void nfa_set_freer (struct rx_hash_item * node) #else static void nfa_set_freer (node) struct rx_hash_item * node; #endif { free ((char *)node->data); } #ifdef __STDC__ void rx_free_nfa (struct rx *rx) #else void rx_free_nfa (rx) struct rx *rx; #endif { rx_free_hash_table (&rx->se_list_memo, se_memo_freer, &se_list_hash_rules); bzero (&rx->se_list_memo, sizeof (rx->se_list_memo)); rx_free_hash_table (&rx->set_list_memo, nfa_set_freer, &nfa_set_hash_rules); bzero (&rx->set_list_memo, sizeof (rx->set_list_memo)); rx_free_nfa_graph (rx); } dist/rx/rxnode.c100660 765 144 14117 6071416221 12503 0ustar lordusers/* classes: src_files */ /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "rxall.h" #include "rxnode.h" #ifdef __STDC__ struct rexp_node * rexp_node (int type) #else struct rexp_node * rexp_node (type) int type; #endif { struct rexp_node *n; n = (struct rexp_node *)malloc (sizeof (*n)); bzero (n, sizeof (*n)); if (n) { n->type = type; n->id = -1; n->refs = 1; } return n; } /* free_rexp_node assumes that the bitset passed to rx_mk_r_cset * can be freed using rx_free_cset. */ #ifdef __STDC__ struct rexp_node * rx_mk_r_cset (int type, int size, rx_Bitset b) #else struct rexp_node * rx_mk_r_cset (type, int size, b) int type; int size; rx_Bitset b; #endif { struct rexp_node * n; n = rexp_node (type); if (n) { n->params.cset = b; n->params.cset_size = size; } return n; } #ifdef __STDC__ struct rexp_node * rx_mk_r_int (int type, int intval) #else struct rexp_node * rx_mk_r_int (type, intval) int type; int intval; #endif { struct rexp_node * n; n = rexp_node (type); if (n) n->params.intval = intval; return n; } #ifdef __STDC__ struct rexp_node * rx_mk_r_binop (int type, struct rexp_node * a, struct rexp_node * b) #else struct rexp_node * rx_mk_r_binop (type, a, b) int type; struct rexp_node * a; struct rexp_node * b; #endif { struct rexp_node * n = rexp_node (type); if (n) { n->params.pair.left = a; n->params.pair.right = b; } return n; } #ifdef __STDC__ struct rexp_node * rx_mk_r_monop (int type, struct rexp_node * a) #else struct rexp_node * rx_mk_r_monop (type, a) int type; struct rexp_node * a; #endif { return rx_mk_r_binop (type, a, 0); } #ifdef __STDC__ void rx_free_rexp (struct rexp_node * node) #else void rx_free_rexp (node) struct rexp_node * node; #endif { if (node && !--node->refs) { if (node->params.cset) rx_free_cset (node->params.cset); rx_free_rexp (node->params.pair.left); rx_free_rexp (node->params.pair.right); rx_free_rexp (node->simplified); free ((char *)node); } } #ifdef __STDC__ void rx_save_rexp (struct rexp_node * node) #else void rx_save_rexp (node) struct rexp_node * node; #endif { if (node) ++node->refs; } #ifdef __STDC__ struct rexp_node * rx_copy_rexp (int cset_size, struct rexp_node *node) #else struct rexp_node * rx_copy_rexp (cset_size, node) int cset_size; struct rexp_node *node; #endif { if (!node) return 0; else { struct rexp_node *n; n = rexp_node (node->type); if (!n) return 0; if (node->params.cset) { n->params.cset = rx_copy_cset (cset_size, node->params.cset); if (!n->params.cset) { rx_free_rexp (n); return 0; } } n->params.intval = node->params.intval; n->params.pair.left = rx_copy_rexp (cset_size, node->params.pair.left); n->params.pair.right = rx_copy_rexp (cset_size, node->params.pair.right); if ( (node->params.pair.left && !n->params.pair.left) || (node->params.pair.right && !n->params.pair.right)) { rx_free_rexp (n); return 0; } return n; } } int re_abs = 0; int re_ans = 0; #ifdef __STDC__ int rx_rexp_equal (struct rexp_node * a, struct rexp_node * b) #else int rx_rexp_equal (a, b) struct rexp_node * a; struct rexp_node * b; #endif { int ret; if (a == b) return 1; if ((a == 0) || (b == 0)) return 0; if (a->type != b->type) return 0; switch (a->type) { case r_cset: ret = ( (a->params.cset_size == b->params.cset_size) && rx_bitset_is_equal (a->params.cset_size, a->params.cset, b->params.cset)); break; case r_concat: case r_alternate: ret = ( rx_rexp_equal (a->params.pair.left, b->params.pair.left) && rx_rexp_equal (a->params.pair.right, b->params.pair.right)); break; case r_opt: case r_star: case r_plus: ret = rx_rexp_equal (a->params.pair.left, b->params.pair.left); break; case r_interval: ret = ( (a->params.intval == b->params.intval) && (a->params.intval2 == b->params.intval2)); break; case r_parens: case r_context: ret = (a->params.intval == b->params.intval); break; default: return 0; } /**/ ++re_abs; re_ans += (ret*2 - 1); return ret; } #ifdef __STDC__ unsigned long rx_rexp_hash (struct rexp_node * node, unsigned long seed) #else unsigned long rx_rexp_hash (node, seed) struct rexp_node * node; unsigned long seed; #endif { if (!node) return 0; seed += (seed << 3) + rx_rexp_hash (node->params.pair.left, seed); seed += (seed << 3) + rx_rexp_hash (node->params.pair.right, seed); seed += (seed << 3) + rx_bitset_hash (node->params.cset_size, node->params.cset); seed += (seed << 3) + node->params.intval; seed += (seed << 3) + node->params.intval2; seed += (seed << 3) + node->type; seed += (seed << 3) + node->id; seed += (seed << 3) + node->len; seed += (seed << 3) + node->observed; return seed; } dist/rx/rxposix.c100660 765 144 22442 6071416221 12720 0ustar lordusers/*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "rxall.h" #include "rxposix.h" #include "rxgnucomp.h" #include "rxbasic.h" #include "rxsimp.h" /* regcomp takes a regular expression as a string and compiles it. * * PATTERN is the address of the pattern string. * * CFLAGS is a series of bits which affect compilation. * * If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we * use POSIX basic syntax. * * If REG_NEWLINE is set, then . and [^...] don't match newline. * Also, regexec will try a match beginning after every newline. * * If REG_ICASE is set, then we considers upper- and lowercase * versions of letters to be equivalent when matching. * * If REG_NOSUB is set, then when PREG is passed to regexec, that * routine will report only success or failure, and nothing about the * registers. * * It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for * the return codes and their meanings.) */ #ifdef __STDC__ int regcomp (regex_t * preg, const char * pattern, int cflags) #else int regcomp (preg, pattern, cflags) regex_t * preg; const char * pattern; int cflags; #endif { reg_errcode_t ret; unsigned int syntax; unsigned char xlate[256]; bzero (preg, sizeof (*preg)); syntax = ((cflags & REG_EXTENDED) ? RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC); if (!(cflags & REG_ICASE)) preg->translate = 0; else { unsigned i; preg->translate = (unsigned char *) malloc (256); if (!preg->translate) return (int) REG_ESPACE; /* Map uppercase characters to corresponding lowercase ones. */ for (i = 0; i < CHAR_SET_SIZE; i++) preg->translate[i] = isupper (i) ? tolower (i) : i; } /* If REG_NEWLINE is set, newlines are treated differently. */ if (!(cflags & REG_NEWLINE)) preg->newline_anchor = 0; else { /* REG_NEWLINE implies neither . nor [^...] match newline. */ syntax &= ~RE_DOT_NEWLINE; syntax |= RE_HAT_LISTS_NOT_NEWLINE; /* It also changes the matching behavior. */ preg->newline_anchor = 1; } preg->no_sub = !!(cflags & REG_NOSUB); /* POSIX says a null character in the pattern terminates it, so we * can use strlen here in compiling the pattern. */ ret = rx_parse (&preg->pattern, pattern, strlen (pattern), syntax, 256, preg->translate); /* POSIX doesn't distinguish between an unmatched open-group and an * unmatched close-group: both are REG_EPAREN. */ if (ret == REG_ERPAREN) ret = REG_EPAREN; if (!ret) { preg->n_subexps = 1; preg->subexps = 0; rx_posix_analyze_rexp (&preg->subexps, &preg->n_subexps, preg->pattern, 0); rx_fill_in_fastmap (256, preg->fastmap, preg->pattern); preg->is_anchored = rx_is_anchored_p (preg->pattern); } return (int) ret; } /* Returns a message corresponding to an error code, ERRCODE, returned from either regcomp or regexec. */ #ifdef __STDC__ size_t regerror (int errcode, const regex_t *preg, char *errbuf, size_t errbuf_size) #else size_t regerror (errcode, preg, errbuf, errbuf_size) int errcode; const regex_t *preg; char *errbuf; size_t errbuf_size; #endif { const char *msg; size_t msg_size; msg = rx_error_msg[errcode] == 0 ? "Success" : rx_error_msg[errcode]; msg_size = strlen (msg) + 1; /* Includes the 0. */ if (errbuf_size != 0) { if (msg_size > errbuf_size) { strncpy (errbuf, msg, errbuf_size - 1); errbuf[errbuf_size - 1] = 0; } else strcpy (errbuf, msg); } return msg_size; } #ifdef __STDC__ int rx_regmatch (regmatch_t pmatch[], regex_t *preg, struct rx_context_rules * rules, int start, int end, char *string) #else int rx_regmatch (pmatch, preg, rules, start, end, string) regmatch_t pmatch[]; regex_t *preg; struct rx_context_rules * rules; int start; int end; char *string; #endif { struct rx_solutions * solutions; enum rx_answers answer; struct rx_context_rules local_rules; int orig_end; local_rules = *rules; orig_end = end; while (end >= start) { local_rules.not_eol = (rules->not_eol ? ( (end == orig_end) || !local_rules.newline_anchor || (string[end] != '\n')) : ( (end != orig_end) && (!local_rules.newline_anchor || (string[end] != '\n')))); solutions = rx_basic_make_solutions (pmatch, preg->pattern, preg->subexps, start, end, &local_rules, string); if (!solutions) return REG_ESPACE; do { answer = rx_next_solution (solutions); } while (answer == rx_maybe); rx_basic_free_solutions (solutions); if (answer == rx_yes) { if (pmatch) { pmatch[0].rm_so = start; pmatch[0].rm_eo = end; } return 0; } --end; } switch (answer) { default: case rx_bogus: return REG_ESPACE; case rx_no: case rx_maybe: return REG_NOMATCH; } } #ifdef __STDC__ int rx_regexec (regmatch_t pmatch[], regex_t *preg, struct rx_context_rules * rules, int start, int end, char *string) #else int rx_regexec (pmatch, preg, rules, start, end, string) regmatch_t pmatch[]; regex_t *preg; struct rx_context_rules * rules; int start; int end; char *string; #endif { int x; int stat; int anchored; anchored = preg->is_anchored; for (x = start; x <= end; ++x) { if (preg->fastmap[((unsigned char *)string)[x]]) { stat = rx_regmatch (pmatch, preg, rules, x, end, string); if (!stat || (stat != REG_NOMATCH)) return stat; } if (anchored) return REG_NOMATCH; } return REG_NOMATCH; } /* regexec searches for a given pattern, specified by PREG, in the * string STRING. * * If NMATCH is zero or REG_NOSUB was set in the cflags argument to * `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at * least NMATCH elements, and we set them to the offsets of the * corresponding matched substrings. * * EFLAGS specifies `execution flags' which affect matching: if * REG_NOTBOL is set, then ^ does not match at the beginning of the * string; if REG_NOTEOL is set, then $ does not match at the end. * * We return 0 if we find a match and REG_NOMATCH if not. */ #ifdef __STDC__ int regnexec (regex_t *preg, int len, char *string, size_t nmatch, regmatch_t pmatch[], int eflags) #else int regnexec (preg, len, string, nmatch, pmatch, eflags) regex_t *preg; int len; char *string; size_t nmatch; regmatch_t pmatch[]; int eflags; #endif { int want_reg_info; struct rx_context_rules rules; regmatch_t * regs; int nregs; int stat; want_reg_info = (!preg->no_sub && (nmatch > 0)); rules.newline_anchor = preg->newline_anchor; rules.not_bol = !!(eflags & REG_NOTBOL); rules.not_eol = !!(eflags & REG_NOTEOL); if (nmatch >= preg->n_subexps) { regs = pmatch; nregs = nmatch; } else { regs = (regmatch_t *)malloc (preg->n_subexps * sizeof (*regs)); nregs = preg->n_subexps; } { int x; for (x = 0; x < nregs; ++x) regs[x].rm_so = regs[x].rm_eo = -1; } stat = rx_regexec (pmatch, preg, &rules, 0, len, string); if (!stat && want_reg_info && (regs != pmatch)) { int x; for (x = 0; x < nmatch; ++x) pmatch[x] = regs[x]; free (regs); } return ((stat == rx_yes) ? 0 : REG_NOMATCH); } #ifdef __STDC__ int regexec (regex_t *preg, char *string, size_t nmatch, regmatch_t pmatch[], int eflags) #else int regexec (preg, string, nmatch, pmatch, eflags) regex_t *preg; char *string; size_t nmatch; regmatch_t pmatch[]; int eflags; #endif { return regnexec (preg, strlen (string), string, nmatch, pmatch, eflags); } /* Free dynamically allocated space used by PREG. */ #ifdef __STDC__ void regfree (regex_t *preg) #else void regfree (preg) regex_t *preg; #endif { if (preg->pattern) { rx_free_rexp (preg->pattern); preg->pattern = 0; } if (preg->subexps) { free (preg->subexps); preg->subexps = 0; } if (preg->translate != 0) { free (preg->translate); preg->translate = 0; } } dist/rx/rxsimp.c100660 765 144 6310 6071416221 12502 0ustar lordusers/*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "rxall.h" #include "rxsimp.h" /* Could reasonably hashcons instead of in rxunfa.c */ #ifdef __STDC__ int rx_simple_rexp (struct rexp_node ** answer, int cset_size, struct rexp_node *node, struct rexp_node ** subexps) #else int rx_simple_rexp (answer, cset_size, node, subexps) struct rexp_node ** answer; int cset_size; struct rexp_node *node; struct rexp_node ** subexps; #endif { int stat; if (!node) { *answer = 0; return 0; } if (!node->observed) { rx_save_rexp (node); *answer = node; return 0; } if (node->simplified) { rx_save_rexp (node->simplified); *answer = node->simplified; return 0; } switch (node->type) { default: case r_cset: return -2; /* an internal error, really */ case r_parens: stat = rx_simple_rexp (answer, cset_size, node->params.pair.left, subexps); break; case r_context: if (isdigit (node->params.intval)) stat = rx_simple_rexp (answer, cset_size, subexps [node->params.intval - '0'], subexps); else { *answer = 0; stat = 0; } break; case r_concat: case r_alternate: case r_opt: case r_star: case r_plus: case r_interval: { struct rexp_node *n; n = rexp_node (node->type); if (!n) return -1; if (node->params.cset) { n->params.cset = rx_copy_cset (cset_size, node->params.cset); if (!n->params.cset) { rx_free_rexp (n); return -1; } } n->params.intval = node->params.intval; n->params.intval2 = node->params.intval2; { int s; if (!(s = rx_simple_rexp (&n->params.pair.left, cset_size, node->params.pair.left, subexps)) && !( s= rx_simple_rexp (&n->params.pair.right, cset_size, node->params.pair.right, subexps))) { *answer = n; stat = 0; } else { rx_free_rexp (n); stat = s; } } } break; } if (!stat) { node->simplified = *answer; rx_save_rexp (node->simplified); } return stat; } dist/rx/rxspencer.c100660 765 144 44536 6071416221 13225 0ustar lordusers/*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "rxall.h" #include "rxspencer.h" static int rx_ms = 0; static int rx_fs = 0; static char * silly_hack_2 = 0; #ifdef __STDC__ struct rx_solutions * rx_make_solutions (struct rx_registers * regs, struct rx_unfaniverse * verse, struct rexp_node * expression, struct rexp_node ** subexps, int cset_size, int start, int end, rx_vmfn vmfn, rx_contextfn contextfn, void * closure) #else struct rx_solutions * rx_make_solutions (regs, verse, expression, subexps, cset_size, start, end, vmfn, contextfn, closure) struct rx_registers * regs; struct rx_unfaniverse * verse; struct rexp_node * expression; struct rexp_node ** subexps; int cset_size; int start; int end; rx_vmfn vmfn; rx_contextfn contextfn; void * closure; #endif { struct rx_solutions * solns; unsigned char * burst; int len; int offset; ++rx_ms; if (silly_hack_2) { solns = (struct rx_solutions *)silly_hack_2; silly_hack_2 = 0; } else solns = (struct rx_solutions *)malloc (sizeof (*solns)); if (!solns) return 0; bzero (solns, sizeof (*solns)); solns->step = 0; solns->cset_size = cset_size; solns->subexps = subexps; solns->exp = expression; rx_save_rexp (expression); solns->verse = verse; solns->regs = regs; solns->start = start; solns->end = end; solns->vmfn = vmfn; solns->contextfn = contextfn; solns->closure = closure; solns->current_pos = start; solns->interval_x = 0; if (!solns->exp->observed) { solns->dfa = rx_unfa (verse, expression, cset_size); if (!solns->dfa) goto err_return; rx_init_system (&solns->match_engine, solns->dfa->nfa); { struct rx * rx; rx = solns->match_engine.rx; #if 0 if ( rx->start_set && (rx->start_set->starts_for == rx) && (rx->start_set->id == rx->rx_id) && rx->start_set->superstate && (rx->start_set->superstate->rx_id == rx->rx_id)) { solns->match_engine.state = solns->match_engine.rx->start_set->superstate; rx_lock_superstate (solns->match_engine.rx, solns->match_engine.state); } else ; #endif if (rx_yes != rx_start_superstate (&solns->match_engine)) goto err_return; } } else { struct rexp_node * simplified; int status; status = rx_simple_rexp (&simplified, cset_size, solns->exp, subexps); if (status) goto err_return; solns->dfa = rx_unfa (verse, simplified, cset_size); if (!solns->dfa) { rx_free_rexp (simplified); goto err_return; } rx_init_system (&solns->match_engine, solns->dfa->nfa); if (rx_yes != rx_start_superstate (&solns->match_engine)) goto err_return; rx_free_rexp (simplified); } return solns; err_return: rx_free_rexp (solns->exp); free (solns); return 0; } #ifdef __STDC__ void rx_free_solutions (struct rx_solutions * solns) #else void rx_free_solutions (solns) struct rx_solutions * solns; #endif { if (!solns) return; ++rx_fs; if (solns->left) { rx_free_solutions (solns->left); solns->left = 0; } if (solns->right) { rx_free_solutions (solns->right); solns->right = 0; } if (solns->dfa) { rx_free_unfa (solns->dfa); solns->dfa = 0; } rx_terminate_system (&solns->match_engine); if (solns->exp) { rx_free_rexp (solns->exp); solns->exp = 0; } if (!silly_hack_2) silly_hack_2 = (char *)solns; else free (solns); } /* This doesn't deal with resumable rx_maybe right. */ enum rx_answers rx_soluntion_fit_p (solns) struct rx_solutions * solns; { unsigned char * burst; int burst_addr; int burst_len; int burst_end_addr; int rel_pos_in_burst; enum rx_answers vmstat; next_burst: vmstat = solns->vmfn (solns->closure, &burst, &burst_len, &burst_addr, solns->current_pos, solns->end, solns->current_pos); if (vmstat != rx_yes) return vmstat; rel_pos_in_burst = solns->current_pos - burst_addr; burst_end_addr = burst_addr + burst_len; if (burst_end_addr >= solns->end) { enum rx_answers fit_status; fit_status = rx_fit_p (&solns->match_engine, burst + rel_pos_in_burst, solns->end - solns->current_pos); return fit_status; } else { enum rx_answers fit_status; fit_status = rx_advance (&solns->match_engine, burst + rel_pos_in_burst, burst_len - rel_pos_in_burst); if (fit_status != rx_yes) { return fit_status; } else { solns->current_pos += burst_len - rel_pos_in_burst; goto next_burst; } } } #ifdef __STDC__ enum rx_answers rx_next_solution (struct rx_solutions * solns) #else enum rx_answers rx_next_solution (solns) struct rx_solutions * solns; #endif { if (!solns) return rx_bogus; if (!solns->exp) { if (solns->step != 0) return rx_no; else { solns->step = 1; return (solns->start == solns->end ? rx_yes : rx_no); } } else if ( (solns->exp->len >= 0) && (solns->exp->len != (solns->end - solns->start))) { return rx_no; } else if (!solns->exp->observed) { if (solns->step != 0) return rx_no; else { enum rx_answers ans; ans = rx_soluntion_fit_p (solns); solns->step = -1; return ans; } } else if (solns->exp->observed) { enum rx_answers fit_p; switch (solns->step) { case -1: return rx_no; case 0: fit_p = rx_soluntion_fit_p (solns); switch (fit_p) { case rx_no: case rx_maybe: solns->step = -1; return rx_no; case rx_yes: solns->step = 1; goto resolve_fit; case rx_bogus: default: solns->step = -1; return fit_p; } default: resolve_fit: switch (solns->exp->type) { case r_cset: solns->step = -1; return rx_bogus; case r_parens: { enum rx_answers paren_stat; switch (solns->step) { case 1: if ( !solns->exp->params.pair.left || !solns->exp->params.pair.left->observed) { solns->regs[solns->exp->params.intval].rm_so = solns->start; solns->regs[solns->exp->params.intval].rm_eo = solns->end; solns->step = -1; return rx_yes; } else { solns->left = rx_make_solutions (solns->regs, solns->verse, solns->exp->params.pair.left, solns->subexps, solns->cset_size, solns->start, solns->end, solns->vmfn, solns->contextfn, solns->closure); if (!solns->left) { solns->step = -1; return rx_bogus; } } solns->step = 2; /* fall through */ case 2: solns->regs[solns->exp->params.intval].rm_so = -1; solns->regs[solns->exp->params.intval].rm_eo = -1; paren_stat = rx_next_solution (solns->left); if (paren_stat == rx_yes) { solns->regs[solns->exp->params.intval].rm_so = solns->start; solns->regs[solns->exp->params.intval].rm_eo = solns->end; return rx_yes; } else if (paren_stat != rx_maybe) { solns->step = -1; rx_free_solutions (solns->left); solns->left = 0; return paren_stat; } else /* paren_stat == rx_maybe */ return paren_stat; } } case r_opt: { enum rx_answers opt_stat; switch (solns->step) { case 1: solns->left = rx_make_solutions (solns->regs, solns->verse, solns->exp->params.pair.left, solns->subexps, solns->cset_size, solns->start, solns->end, solns->vmfn, solns->contextfn, solns->closure); if (!solns->left) { solns->step = -1; return rx_bogus; } solns->step = 2; /* fall through */ case 2: opt_stat = rx_next_solution (solns->left); if (opt_stat == rx_yes) return rx_yes; else if (opt_stat != rx_maybe) { solns->step = -1; rx_free_solutions (solns->left); solns->left = 0; return ((solns->start == solns->end) ? rx_yes : rx_no); } else /* opt_stat == rx_maybe */ return opt_stat; } } case r_alternate: { enum rx_answers alt_stat; switch (solns->step) { case 1: solns->left = rx_make_solutions (solns->regs, solns->verse, solns->exp->params.pair.left, solns->subexps, solns->cset_size, solns->start, solns->end, solns->vmfn, solns->contextfn, solns->closure); if (!solns->left) { solns->step = -1; return rx_bogus; } solns->step = 2; /* fall through */ case 2: alt_stat = rx_next_solution (solns->left); if ( (alt_stat == rx_yes) || (alt_stat == rx_maybe)) return alt_stat; else { solns->step = 3; rx_free_solutions (solns->left); solns->left = 0; /* fall through */ } case 3: solns->right = rx_make_solutions (solns->regs, solns->verse, solns->exp->params.pair.right, solns->subexps, solns->cset_size, solns->start, solns->end, solns->vmfn, solns->contextfn, solns->closure); if (!solns->right) { solns->step = -1; return rx_bogus; } solns->step = 4; /* fall through */ case 4: alt_stat = rx_next_solution (solns->right); if ( (alt_stat == rx_yes) || (alt_stat == rx_maybe)) return alt_stat; else { solns->step = -1; rx_free_solutions (solns->right); solns->right = 0; return alt_stat; } } } case r_concat: { switch (solns->step) { enum rx_answers concat_stat; case 1: solns->split_guess = solns->end; concat_split_guess_loop: solns->left = rx_make_solutions (solns->regs, solns->verse, solns->exp->params.pair.left, solns->subexps, solns->cset_size, solns->start, solns->split_guess, solns->vmfn, solns->contextfn, solns->closure); if (!solns->left) { solns->step = -1; return rx_bogus; } solns->step = 2; case 2: concat_try_next_left_match: concat_stat = rx_next_solution (solns->left); if (concat_stat == rx_maybe) return concat_stat; else if (concat_stat != rx_yes) { rx_free_solutions (solns->left); rx_free_solutions (solns->right); solns->left = solns->right = 0; solns->split_guess -= 1; if (solns->split_guess >= solns->start) goto concat_split_guess_loop; else { solns->step = -1; return concat_stat; } } else { solns->step = 3; /* fall through */ } case 3: solns->right = rx_make_solutions (solns->regs, solns->verse, solns->exp->params.pair.right, solns->subexps, solns->cset_size, solns->split_guess, solns->end, solns->vmfn, solns->contextfn, solns->closure); if (!solns->right) { rx_free_solutions (solns->left); solns->left = 0; solns->step = -1; return rx_bogus; } solns->step = 4; /* fall through */ case 4: concat_try_next_right_match: concat_stat = rx_next_solution (solns->right); if ((concat_stat == rx_maybe) || (concat_stat == rx_yes)) return concat_stat; else if (concat_stat == rx_no) { rx_free_solutions (solns->right); solns->right = 0; solns->step = 2; goto concat_try_next_left_match; } else /* concat_stat == rx_bogus */ { rx_free_solutions (solns->left); solns->left = 0; rx_free_solutions (solns->right); solns->right = 0; solns->step = -1; return concat_stat; } } } case r_plus: case r_star: { switch (solns->step) { enum rx_answers star_stat; case 1: solns->split_guess = solns->end; star_split_guess_loop: solns->left = rx_make_solutions (solns->regs, solns->verse, solns->exp->params.pair.left, solns->subexps, solns->cset_size, solns->start, solns->split_guess, solns->vmfn, solns->contextfn, solns->closure); if (!solns->left) { solns->step = -1; return rx_bogus; } solns->step = 2; case 2: star_try_next_left_match: star_stat = rx_next_solution (solns->left); if (star_stat == rx_maybe) return star_stat; else if (star_stat != rx_yes) { rx_free_solutions (solns->left); rx_free_solutions (solns->right); solns->left = solns->right = 0; solns->split_guess -= 1; if (solns->split_guess >= solns->start) goto star_split_guess_loop; else { solns->step = -1; if ( (solns->exp->type == r_star) && (solns->split_guess == solns->end)) return rx_yes; else return star_stat; } } else { solns->step = 3; /* fall through */ } if (solns->split_guess == solns->end) return rx_yes; case 3: solns->right = rx_make_solutions (solns->regs, solns->verse, solns->exp, solns->subexps, solns->cset_size, solns->split_guess, solns->end, solns->vmfn, solns->contextfn, solns->closure); if (!solns->right) { rx_free_solutions (solns->left); solns->left = 0; solns->step = -1; return rx_bogus; } solns->step = 4; /* fall through */ case 4: star_try_next_right_match: star_stat = rx_next_solution (solns->right); if ((star_stat == rx_maybe) || (star_stat == rx_yes)) return star_stat; else if (star_stat == rx_no) { rx_free_solutions (solns->right); solns->right = 0; solns->step = 2; goto star_try_next_left_match; } else /* star_stat == rx_bogus */ { rx_free_solutions (solns->left); solns->left = 0; rx_free_solutions (solns->right); solns->right = 0; solns->step = -1; return star_stat; } } } case r_context: { solns->step = -1; return solns->contextfn (solns->closure, solns->exp->params.intval, solns->start, solns->end, solns->regs); } case r_interval: { switch (solns->step) { enum rx_answers interval_stat; case 1: if (solns->interval_x >= solns->exp->params.intval2) { solns->step = -1; return rx_no; } solns->split_guess = solns->end; interval_split_guess_loop: solns->left = rx_make_solutions (solns->regs, solns->verse, solns->exp->params.pair.left, solns->subexps, solns->cset_size, solns->start, solns->split_guess, solns->vmfn, solns->contextfn, solns->closure); if (!solns->left) { solns->step = -1; return rx_bogus; } solns->step = 2; case 2: interval_try_next_left_match: interval_stat = rx_next_solution (solns->left); if (interval_stat == rx_maybe) return interval_stat; else if (interval_stat != rx_yes) { rx_free_solutions (solns->left); rx_free_solutions (solns->right); solns->left = solns->right = 0; solns->split_guess -= 1; if (solns->split_guess >= solns->start) goto interval_split_guess_loop; else { solns->step = -1; return (( (interval_stat == rx_no) && ( solns->interval_x == (solns->exp->params.intval2))) ? rx_yes : rx_no); } } else { solns->step = 3; /* fall through */ } if ( (solns->split_guess == solns->end) && (solns->interval_x >= (solns->exp->params.intval - 1))) return rx_yes; case 3: solns->right = rx_make_solutions (solns->regs, solns->verse, solns->exp, solns->subexps, solns->cset_size, solns->split_guess, solns->end, solns->vmfn, solns->contextfn, solns->closure); solns->right->interval_x = solns->interval_x + 1; if (!solns->right) { rx_free_solutions (solns->left); solns->left = 0; solns->step = -1; return rx_bogus; } solns->step = 4; /* fall through */ case 4: interval_try_next_right_match: interval_stat = rx_next_solution (solns->right); if ((interval_stat == rx_maybe) || (interval_stat == rx_yes)) return interval_stat; else if (interval_stat == rx_no) { rx_free_solutions (solns->right); solns->right = 0; solns->step = 2; goto interval_try_next_left_match; } else /* interval_stat == rx_bogus */ { rx_free_solutions (solns->left); solns->left = 0; rx_free_solutions (solns->right); solns->right = 0; solns->step = -1; return interval_stat; } } } } } return rx_bogus; } } dist/rx/rxstr.c100660 765 144 6176 6071416221 12354 0ustar lordusers/*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "rxall.h" #include "rxstr.h" #ifdef __STDC__ enum rx_answers rx_str_vmfn (void * closure, unsigned char ** burstp, int * lenp, int * offsetp, int start, int end, int need) #else enum rx_answers rx_str_vmfn (closure, burstp, lenp, offsetp, start, end, need) void * closure; unsigned char ** burstp; int * lenp; int * offsetp; int start; int end; int need; #endif { struct rx_str_closure * strc; strc = (struct rx_str_closure *)closure; if ( (need < 0) || (need > strc->len)) return rx_no; *burstp = strc->str; *lenp = strc->len; *offsetp = 0; return rx_yes; } #ifdef __STDC__ enum rx_answers rx_str_contextfn (void * closure, int type, int start, int end, struct rx_registers * regs) #else enum rx_answers rx_str_contextfn (closure, type, start, end, regs) void * closure; int type; int start; int end; struct rx_registers * regs; #endif { struct rx_str_closure * strc; strc = (struct rx_str_closure *)closure; switch (type) { case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9': { int cmp; int regn; regn = type - '0'; if ( (regs[regn].rm_so == -1) || ((end - start) != (regs[regn].rm_eo - regs[regn].rm_so))) return rx_no; else { cmp = strncmp (strc->str + start, strc->str + regs[regn].rm_so, end - start); return (!cmp ? rx_yes : rx_no); } } case '^': { return (( (start == end) && ( ((start == 0) && !strc->rules.not_bol) || ( (start > 0) && strc->rules.newline_anchor && (strc->str[start - 1] == '\n')))) ? rx_yes : rx_no); } case '$': { return (( (start == end) && ( ((start == strc->len) && !strc->rules.not_eol) || ( (start < strc->len) && strc->rules.newline_anchor && (strc->str[start] == '\n')))) ? rx_yes : rx_no); } case '<': case '>': case 'B': case 'b': default: return rx_bogus; } } dist/rx/rxsuper.c100660 765 144 111303 6071416221 12727 0ustar lordusers/*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ /* * Tom Lord (lord@cygnus.com, lord@gnu.ai.mit.edu) */ #include "rxall.h" #include "rxsuper.h" /* The functions in the next several pages define the lazy-NFA-conversion used * by matchers. The input to this construction is an NFA such as * is built by compactify_nfa (rx.c). The output is the superNFA. */ /* Match engines can use arbitrary values for opcodes. So, the parse tree * is built using instructions names (enum rx_opcode), but the superstate * nfa is populated with mystery opcodes (void *). * * For convenience, here is an id table. The opcodes are == to their inxs * * The lables in re_search_2 would make good values for instructions. */ void * rx_id_instruction_table[rx_num_instructions] = { (void *) rx_backtrack_point, (void *) rx_do_side_effects, (void *) rx_cache_miss, (void *) rx_next_char, (void *) rx_backtrack, (void *) rx_error_inx }; /* The partially instantiated superstate graph has a transition * table at every node. There is one entry for every character. * This fills in the transition for a set. */ #ifdef __STDC__ static void install_transition (struct rx_superstate *super, struct rx_inx *answer, rx_Bitset trcset) #else static void install_transition (super, answer, trcset) struct rx_superstate *super; struct rx_inx *answer; rx_Bitset trcset; #endif { struct rx_inx * transitions = super->transitions; int chr; for (chr = 0; chr < 256; ) if (!*trcset) { ++trcset; chr += 32; } else { RX_subset sub = *trcset; RX_subset mask = 1; int bound = chr + 32; while (chr < bound) { if (sub & mask) transitions [chr] = *answer; ++chr; mask <<= 1; } ++trcset; } } #ifdef __STDC__ static int qlen (struct rx_superstate * q) #else static int qlen (q) struct rx_superstate * q; #endif { int count = 1; struct rx_superstate * it; if (!q) return 0; for (it = q->next_recyclable; it != q; it = it->next_recyclable) ++count; return count; } #ifdef __STDC__ static void check_cache (struct rx_cache * cache) #else static void check_cache (cache) struct rx_cache * cache; #endif { struct rx_cache * you_fucked_up = 0; int total = cache->superstates; int semi = cache->semifree_superstates; if (semi != qlen (cache->semifree_superstate)) check_cache (you_fucked_up); if ((total - semi) != qlen (cache->lru_superstate)) check_cache (you_fucked_up); } #ifdef __STDC__ char * rx_cache_malloc (struct rx_cache * cache, int size) #else char * rx_cache_malloc (cache, size) struct rx_cache * cache; int size; #endif { char * answer; answer = (char *)malloc (size); if (answer) cache->bytes_used += size; return answer; } #ifdef __STDC__ void rx_cache_free (struct rx_cache * cache, int size, char * mem) #else void rx_cache_free (cache, size, mem) struct rx_cache * cache; int size; char * mem; #endif { free (mem); cache->bytes_used -= size; } /* When a superstate is old and neglected, it can enter a * semi-free state. A semi-free state is slated to die. * Incoming transitions to a semi-free state are re-written * to cause an (interpreted) fault when they are taken. * The fault handler revives the semi-free state, patches * incoming transitions back to normal, and continues. * * The idea is basicly to free in two stages, aborting * between the two if the state turns out to be useful again. * When a free is aborted, the rescued superstate is placed * in the most-favored slot to maximize the time until it * is next semi-freed. */ #ifdef __STDC__ static void semifree_superstate (struct rx_cache * cache) #else static void semifree_superstate (cache) struct rx_cache * cache; #endif { int disqualified = cache->semifree_superstates; if (disqualified == cache->superstates) return; while (cache->lru_superstate->locks) { cache->lru_superstate = cache->lru_superstate->next_recyclable; ++disqualified; if (disqualified == cache->superstates) return; } { struct rx_superstate * it = cache->lru_superstate; it->next_recyclable->prev_recyclable = it->prev_recyclable; it->prev_recyclable->next_recyclable = it->next_recyclable; cache->lru_superstate = (it == it->next_recyclable ? 0 : it->next_recyclable); if (!cache->semifree_superstate) { cache->semifree_superstate = it; it->next_recyclable = it; it->prev_recyclable = it; } else { it->prev_recyclable = cache->semifree_superstate->prev_recyclable; it->next_recyclable = cache->semifree_superstate; it->prev_recyclable->next_recyclable = it; it->next_recyclable->prev_recyclable = it; } { struct rx_distinct_future *df; it->is_semifree = 1; ++cache->semifree_superstates; df = it->transition_refs; if (df) { df->prev_same_dest->next_same_dest = 0; for (df = it->transition_refs; df; df = df->next_same_dest) { df->future_frame.inx = cache->instruction_table[rx_cache_miss]; df->future_frame.data = 0; df->future_frame.data_2 = (void *) df; /* If there are any NEXT-CHAR instruction frames that * refer to this state, we convert them to CACHE-MISS frames. */ if (!df->effects && (df->edge->options->next_same_super_edge[0] == df->edge->options)) install_transition (df->present, &df->future_frame, df->edge->cset); } df = it->transition_refs; df->prev_same_dest->next_same_dest = df; } } } } #ifdef __STDC__ static void refresh_semifree_superstate (struct rx_cache * cache, struct rx_superstate * super) #else static void refresh_semifree_superstate (cache, super) struct rx_cache * cache; struct rx_superstate * super; #endif { struct rx_distinct_future *df; if (super->transition_refs) { super->transition_refs->prev_same_dest->next_same_dest = 0; for (df = super->transition_refs; df; df = df->next_same_dest) { df->future_frame.inx = cache->instruction_table[rx_next_char]; df->future_frame.data = (void *) super->transitions; df->future_frame.data_2 = (void *)(super->contents->is_final); /* CACHE-MISS instruction frames that refer to this state, * must be converted to NEXT-CHAR frames. */ if (!df->effects && (df->edge->options->next_same_super_edge[0] == df->edge->options)) install_transition (df->present, &df->future_frame, df->edge->cset); } super->transition_refs->prev_same_dest->next_same_dest = super->transition_refs; } if (cache->semifree_superstate == super) cache->semifree_superstate = (super->prev_recyclable == super ? 0 : super->prev_recyclable); super->next_recyclable->prev_recyclable = super->prev_recyclable; super->prev_recyclable->next_recyclable = super->next_recyclable; if (!cache->lru_superstate) (cache->lru_superstate = super->next_recyclable = super->prev_recyclable = super); else { super->next_recyclable = cache->lru_superstate; super->prev_recyclable = cache->lru_superstate->prev_recyclable; super->next_recyclable->prev_recyclable = super; super->prev_recyclable->next_recyclable = super; } super->is_semifree = 0; --cache->semifree_superstates; } #ifdef __STDC__ static void rx_refresh_this_superstate (struct rx_cache * cache, struct rx_superstate * superstate) #else static void rx_refresh_this_superstate (cache, superstate) struct rx_cache * cache; struct rx_superstate * superstate; #endif { if (superstate->is_semifree) refresh_semifree_superstate (cache, superstate); else if (cache->lru_superstate == superstate) cache->lru_superstate = superstate->next_recyclable; else if (superstate != cache->lru_superstate->prev_recyclable) { superstate->next_recyclable->prev_recyclable = superstate->prev_recyclable; superstate->prev_recyclable->next_recyclable = superstate->next_recyclable; superstate->next_recyclable = cache->lru_superstate; superstate->prev_recyclable = cache->lru_superstate->prev_recyclable; superstate->next_recyclable->prev_recyclable = superstate; superstate->prev_recyclable->next_recyclable = superstate; } } #ifdef __STDC__ static void release_superset_low (struct rx_cache * cache, struct rx_superset *set) #else static void release_superset_low (cache, set) struct rx_cache * cache; struct rx_superset *set; #endif { if (!--set->refs) { if (set->cdr) release_superset_low (cache, set->cdr); set->starts_for = 0; rx_hash_free (&set->hash_item, &cache->superset_hash_rules); rx_cache_free (cache, sizeof (struct rx_superset), (char *)set); } } #ifdef __STDC__ void rx_release_superset (struct rx *rx, struct rx_superset *set) #else void rx_release_superset (rx, set) struct rx *rx; struct rx_superset *set; #endif { release_superset_low (rx->cache, set); } /* This tries to add a new superstate to the superstate freelist. * It might, as a result, free some edge pieces or hash tables. * If nothing can be freed because too many locks are being held, fail. */ #ifdef __STDC__ static int rx_really_free_superstate (struct rx_cache * cache) #else static int rx_really_free_superstate (cache) struct rx_cache * cache; #endif { int locked_superstates = 0; struct rx_superstate * it; if (!cache->superstates) return 0; /* scale these */ while ((cache->hits + cache->misses) > cache->superstates) { cache->hits >>= 1; cache->misses >>= 1; } /* semifree superstates twice as fast as they are freed * so popular states can be rescued. */ semifree_superstate (cache); semifree_superstate (cache); semifree_superstate (cache); #if 0 Redundant because semifree_superstate already makes this check; while (cache->semifree_superstate && cache->semifree_superstate->locks) { refresh_semifree_superstate (cache, cache->semifree_superstate); ++locked_superstates; if (locked_superstates == cache->superstates) return 0; } if (cache->semifree_superstate) insert the "it =" block which follows this "if 0" code; else { while (cache->lru_superstate->locks) { cache->lru_superstate = cache->lru_superstate->next_recyclable; ++locked_superstates; if (locked_superstates == cache->superstates) return 0; } it = cache->lru_superstate; it->next_recyclable->prev_recyclable = it->prev_recyclable; it->prev_recyclable->next_recyclable = it->next_recyclable; cache->lru_superstate = ((it == it->next_recyclable) ? 0 : it->next_recyclable); } #endif if (!cache->semifree_superstate) return 0; { it = cache->semifree_superstate; it->next_recyclable->prev_recyclable = it->prev_recyclable; it->prev_recyclable->next_recyclable = it->next_recyclable; cache->semifree_superstate = ((it == it->next_recyclable) ? 0 : it->next_recyclable); --cache->semifree_superstates; } if (it->transition_refs) { struct rx_distinct_future *df; for (df = it->transition_refs, df->prev_same_dest->next_same_dest = 0; df; df = df->next_same_dest) { df->future_frame.inx = cache->instruction_table[rx_cache_miss]; df->future_frame.data = 0; df->future_frame.data_2 = (void *) df; df->future = 0; } it->transition_refs->prev_same_dest->next_same_dest = it->transition_refs; } { struct rx_super_edge *tc = it->edges; while (tc) { struct rx_distinct_future * df; struct rx_super_edge *tct = tc->next; df = tc->options; df->next_same_super_edge[1]->next_same_super_edge[0] = 0; while (df) { struct rx_distinct_future *dft = df; df = df->next_same_super_edge[0]; if (dft->future && dft->future->transition_refs == dft) { dft->future->transition_refs = dft->next_same_dest; if (dft->future->transition_refs == dft) dft->future->transition_refs = 0; } dft->next_same_dest->prev_same_dest = dft->prev_same_dest; dft->prev_same_dest->next_same_dest = dft->next_same_dest; rx_cache_free (cache, sizeof (struct rx_distinct_future), (char *)dft); } rx_cache_free (cache, sizeof (struct rx_super_edge), (char *)tc); tc = tct; } } if (it->contents->superstate == it) it->contents->superstate = 0; release_superset_low (cache, it->contents); rx_cache_free (cache, (sizeof (struct rx_superstate) + cache->local_cset_size * sizeof (struct rx_inx)), (char *)it); --cache->superstates; return 1; } #ifdef __STDC__ static char * rx_cache_malloc_or_get (struct rx_cache * cache, int size) #else static char * rx_cache_malloc_or_get (cache, size) struct rx_cache * cache; int size; #endif { while ( (cache->bytes_used + size > cache->bytes_allowed) && rx_really_free_superstate (cache)) ; return rx_cache_malloc (cache, size); } #ifdef __STDC__ static int supersetcmp (void * va, void * vb) #else static int supersetcmp (va, vb) void * va; void * vb; #endif { struct rx_superset * a = (struct rx_superset *)va; struct rx_superset * b = (struct rx_superset *)vb; return ( (a == b) || (a && b && (a->id == b->id) && (a->car == b->car) && (a->cdr == b->cdr))); } #ifdef __STDC__ static struct rx_hash_item * superset_allocator (struct rx_hash_rules * rules, void * val) #else static struct rx_hash_item * superset_allocator (rules, val) struct rx_hash_rules * rules; void * val; #endif { struct rx_cache * cache; struct rx_superset * template; struct rx_superset * newset; cache = ((struct rx_cache *) ((char *)rules - (unsigned long)(&((struct rx_cache *)0)->superset_hash_rules))); template = (struct rx_superset *)val; newset = ((struct rx_superset *) rx_cache_malloc (cache, sizeof (*template))); if (!newset) return 0; newset->refs = 0; newset->car = template->car; newset->id = template->id; newset->cdr = template->cdr; newset->is_final = ( template->car->is_final || ( template->cdr && template->cdr->is_final)); newset->superstate = 0; rx_protect_superset (rx, template->cdr); newset->hash_item.data = (void *)newset; newset->hash_item.binding = 0; return &newset->hash_item; } #ifdef __STDC__ static struct rx_hash * super_hash_allocator (struct rx_hash_rules * rules) #else static struct rx_hash * super_hash_allocator (rules) struct rx_hash_rules * rules; #endif { struct rx_cache * cache; cache = ((struct rx_cache *) ((char *)rules - (unsigned long)(&((struct rx_cache *)0)->superset_hash_rules))); return ((struct rx_hash *) rx_cache_malloc (cache, sizeof (struct rx_hash))); } #ifdef __STDC__ static void super_hash_liberator (struct rx_hash * hash, struct rx_hash_rules * rules) #else static void super_hash_liberator (hash, rules) struct rx_hash * hash; struct rx_hash_rules * rules; #endif { struct rx_cache * cache = ((struct rx_cache *) (char *)rules - (long)(&((struct rx_cache *)0)->superset_hash_rules)); rx_cache_free (cache, sizeof (struct rx_hash), (char *)hash); } #ifdef __STDC__ static void superset_hash_item_liberator (struct rx_hash_item * it, struct rx_hash_rules * rules) #else static void superset_hash_item_liberator (it, rules) struct rx_hash_item * it; struct rx_hash_rules * rules; #endif { } int rx_cache_bound = 3; static int rx_default_cache_got = 0; static struct rx_cache default_cache = { { supersetcmp, super_hash_allocator, super_hash_liberator, superset_allocator, superset_hash_item_liberator, }, 0, 0, 0, 0, 0, 0, 0, 524288, 0, 256, rx_id_instruction_table, { 0, 0, 0, 0, 0 } }; struct rx_cache * rx_default_cache = &default_cache; /* This adds an element to a superstate set. These sets are lists, such * that lists with == elements are ==. The empty set is returned by * superset_cons (rx, 0, 0) and is NOT equivelent to * (struct rx_superset)0. */ #ifdef __STDC__ struct rx_superset * rx_superset_cons (struct rx * rx, struct rx_nfa_state *car, struct rx_superset *cdr) #else struct rx_superset * rx_superset_cons (rx, car, cdr) struct rx * rx; struct rx_nfa_state *car; struct rx_superset *cdr; #endif { struct rx_cache * cache = rx->cache; if (!car && !cdr) { if (!cache->empty_superset) { cache->empty_superset = ((struct rx_superset *) rx_cache_malloc (cache, sizeof (struct rx_superset))); if (!cache->empty_superset) return 0; bzero (cache->empty_superset, sizeof (struct rx_superset)); cache->empty_superset->refs = 1000; } return cache->empty_superset; } { struct rx_superset template; struct rx_hash_item * hit; template.car = car; template.cdr = cdr; template.id = rx->rx_id; hit = rx_hash_store (&cache->superset_table, (unsigned long)car ^ car->id ^ (unsigned long)cdr, (void *)&template, &cache->superset_hash_rules); return (hit ? (struct rx_superset *)hit->data : 0); } } /* This computes a union of two NFA state sets. The sets do not have the * same representation though. One is a RX_SUPERSET structure (part * of the superstate NFA) and the other is an NFA_STATE_SET (part of the NFA). */ #ifdef __STDC__ struct rx_superset * rx_superstate_eclosure_union (struct rx * rx, struct rx_superset *set, struct rx_nfa_state_set *ecl) #else struct rx_superset * rx_superstate_eclosure_union (rx, set, ecl) struct rx * rx; struct rx_superset *set; struct rx_nfa_state_set *ecl; #endif { if (!ecl) return set; if (!set->car) return rx_superset_cons (rx, ecl->car, rx_superstate_eclosure_union (rx, set, ecl->cdr)); if (set->car == ecl->car) return rx_superstate_eclosure_union (rx, set, ecl->cdr); { struct rx_superset * tail; struct rx_nfa_state * first; if (set->car > ecl->car) { tail = rx_superstate_eclosure_union (rx, set->cdr, ecl); first = set->car; } else { tail = rx_superstate_eclosure_union (rx, set, ecl->cdr); first = ecl->car; } if (!tail) return 0; else { struct rx_superset * answer; answer = rx_superset_cons (rx, first, tail); if (!answer) { rx_protect_superset (rx, tail); rx_release_superset (rx, tail); return 0; } else return answer; } } } /* * This makes sure that a list of rx_distinct_futures contains * a future for each possible set of side effects in the eclosure * of a given state. This is some of the work of filling in a * superstate transition. */ #ifdef __STDC__ static struct rx_distinct_future * include_futures (struct rx *rx, struct rx_distinct_future *df, struct rx_nfa_state *state, struct rx_superstate *superstate) #else static struct rx_distinct_future * include_futures (rx, df, state, superstate) struct rx *rx; struct rx_distinct_future *df; struct rx_nfa_state *state; struct rx_superstate *superstate; #endif { struct rx_possible_future *future; struct rx_cache * cache = rx->cache; for (future = rx_state_possible_futures (rx, state); future; future = future->next) { struct rx_distinct_future *dfp; struct rx_distinct_future *insert_before = 0; if (df) df->next_same_super_edge[1]->next_same_super_edge[0] = 0; for (dfp = df; dfp; dfp = dfp->next_same_super_edge[0]) if (dfp->effects == future->effects) break; else { int order = rx->se_list_cmp (rx, dfp->effects, future->effects); if (order > 0) { insert_before = dfp; dfp = 0; break; } } if (df) df->next_same_super_edge[1]->next_same_super_edge[0] = df; if (!dfp) { dfp = ((struct rx_distinct_future *) rx_cache_malloc (cache, sizeof (struct rx_distinct_future))); if (!dfp) return 0; if (!df) { df = insert_before = dfp; df->next_same_super_edge[0] = df->next_same_super_edge[1] = df; } else if (!insert_before) insert_before = df; else if (insert_before == df) df = dfp; dfp->next_same_super_edge[0] = insert_before; dfp->next_same_super_edge[1] = insert_before->next_same_super_edge[1]; dfp->next_same_super_edge[1]->next_same_super_edge[0] = dfp; dfp->next_same_super_edge[0]->next_same_super_edge[1] = dfp; dfp->next_same_dest = dfp->prev_same_dest = dfp; dfp->future = 0; dfp->present = superstate; dfp->future_frame.inx = rx->instruction_table[rx_cache_miss]; dfp->future_frame.data = 0; dfp->future_frame.data_2 = (void *) dfp; dfp->side_effects_frame.inx = rx->instruction_table[rx_do_side_effects]; dfp->side_effects_frame.data = 0; dfp->side_effects_frame.data_2 = (void *) dfp; dfp->effects = future->effects; } } return df; } /* This constructs a new superstate from its state set. The only * complexity here is memory management. */ #ifdef __STDC__ struct rx_superstate * rx_superstate (struct rx *rx, struct rx_superset *set) #else struct rx_superstate * rx_superstate (rx, set) struct rx *rx; struct rx_superset *set; #endif { struct rx_cache * cache = rx->cache; struct rx_superstate * superstate = 0; /* Does the superstate already exist in the cache? */ if (set->superstate) { if (set->superstate->rx_id != rx->rx_id) { /* Aha. It is in the cache, but belongs to a superstate * that refers to an NFA that no longer exists. * (We know it no longer exists because it was evidently * stored in the same region of memory as the current nfa * yet it has a different id.) */ superstate = set->superstate; if (!superstate->is_semifree) { if (cache->lru_superstate == superstate) { cache->lru_superstate = superstate->next_recyclable; if (cache->lru_superstate == superstate) cache->lru_superstate = 0; } { superstate->next_recyclable->prev_recyclable = superstate->prev_recyclable; superstate->prev_recyclable->next_recyclable = superstate->next_recyclable; if (!cache->semifree_superstate) { (cache->semifree_superstate = superstate->next_recyclable = superstate->prev_recyclable = superstate); } else { superstate->next_recyclable = cache->semifree_superstate; superstate->prev_recyclable = cache->semifree_superstate->prev_recyclable; superstate->next_recyclable->prev_recyclable = superstate; superstate->prev_recyclable->next_recyclable = superstate; cache->semifree_superstate = superstate; } ++cache->semifree_superstates; } } set->superstate = 0; goto handle_cache_miss; } ++cache->hits; superstate = set->superstate; rx_refresh_this_superstate (cache, superstate); return superstate; } handle_cache_miss: /* This point reached only for cache misses. */ ++cache->misses; #if RX_DEBUG if (rx_debug_trace > 1) { struct rx_superset * setp = set; fprintf (stderr, "Building a superstet %d(%d): ", rx->rx_id, set); while (setp) { fprintf (stderr, "%d ", setp->id); setp = setp->cdr; } fprintf (stderr, "(%d)\n", set); } #endif { int superstate_size; superstate_size = ( sizeof (*superstate) + ( sizeof (struct rx_inx) * rx->local_cset_size)); superstate = ((struct rx_superstate *) rx_cache_malloc_or_get (cache, superstate_size)); ++cache->superstates; } if (!superstate) return 0; if (!cache->lru_superstate) (cache->lru_superstate = superstate->next_recyclable = superstate->prev_recyclable = superstate); else { superstate->next_recyclable = cache->lru_superstate; superstate->prev_recyclable = cache->lru_superstate->prev_recyclable; ( superstate->prev_recyclable->next_recyclable = superstate->next_recyclable->prev_recyclable = superstate); } superstate->rx_id = rx->rx_id; superstate->transition_refs = 0; superstate->locks = 0; superstate->is_semifree = 0; set->superstate = superstate; superstate->contents = set; rx_protect_superset (rx, set); superstate->edges = 0; { int x; /* None of the transitions from this superstate are known yet. */ for (x = 0; x < rx->local_cset_size; ++x) /* &&&&& 3.8 % */ { struct rx_inx * ifr = &superstate->transitions[x]; ifr->inx = rx->instruction_table [rx_cache_miss]; ifr->data = ifr->data_2 = 0; } } return superstate; } /* This computes the destination set of one edge of the superstate NFA. * Note that a RX_DISTINCT_FUTURE is a superstate edge. * Returns 0 on an allocation failure. */ #ifdef __STDC__ static int solve_destination (struct rx *rx, struct rx_distinct_future *df) #else static int solve_destination (rx, df) struct rx *rx; struct rx_distinct_future *df; #endif { struct rx_super_edge *tc = df->edge; struct rx_superset *nfa_state; struct rx_superset *nil_set = rx_superset_cons (rx, 0, 0); struct rx_superset *solution = nil_set; struct rx_superstate *dest; rx_protect_superset (rx, solution); /* Iterate over all NFA states in the state set of this superstate. */ for (nfa_state = df->present->contents; nfa_state->car; nfa_state = nfa_state->cdr) { struct rx_nfa_edge *e; /* Iterate over all edges of each NFA state. */ for (e = nfa_state->car->edges; e; e = e->next) /* If we find an edge that is labeled with * the characters we are solving for..... */ if ((e->type == ne_cset) && rx_bitset_is_subset (rx->local_cset_size, tc->cset, e->params.cset)) { struct rx_nfa_state *n = e->dest; struct rx_possible_future *pf; /* ....search the partial epsilon closures of the destination * of that edge for a path that involves the same set of * side effects we are solving for. * If we find such a RX_POSSIBLE_FUTURE, we add members to the * stateset we are computing. */ for (pf = rx_state_possible_futures (rx, n); pf; pf = pf->next) if (pf->effects == df->effects) { struct rx_superset * old_sol; old_sol = solution; solution = rx_superstate_eclosure_union (rx, solution, pf->destset); if (!solution) return 0; rx_protect_superset (rx, solution); rx_release_superset (rx, old_sol); } } } /* It is possible that the RX_DISTINCT_FUTURE we are working on has * the empty set of NFA states as its definition. In that case, this * is a failure point. */ if (solution == nil_set) { df->future_frame.inx = (void *) rx_backtrack; df->future_frame.data = 0; df->future_frame.data_2 = 0; return 1; } dest = rx_superstate (rx, solution); rx_release_superset (rx, solution); if (!dest) return 0; { struct rx_distinct_future *dft; dft = df; df->prev_same_dest->next_same_dest = 0; while (dft) { dft->future = dest; dft->future_frame.inx = rx->instruction_table[rx_next_char]; dft->future_frame.data = (void *) dest->transitions; dft->future_frame.data_2 = (void *) dest->contents->is_final; dft = dft->next_same_dest; } df->prev_same_dest->next_same_dest = df; } if (!dest->transition_refs) dest->transition_refs = df; else { struct rx_distinct_future *dft = dest->transition_refs->next_same_dest; dest->transition_refs->next_same_dest = df->next_same_dest; df->next_same_dest->prev_same_dest = dest->transition_refs; df->next_same_dest = dft; dft->prev_same_dest = df; } return 1; } /* This takes a superstate and a character, and computes some edges * from the superstate NFA. In particular, this computes all edges * that lead from SUPERSTATE given CHR. This function also * computes the set of characters that share this edge set. * This returns 0 on allocation error. * The character set and list of edges are returned through * the paramters CSETOUT and DFOUT. } */ #ifdef __STDC__ static int compute_super_edge (struct rx *rx, struct rx_distinct_future **dfout, rx_Bitset csetout, struct rx_superstate *superstate, unsigned char chr) #else static int compute_super_edge (rx, dfout, csetout, superstate, chr) struct rx *rx; struct rx_distinct_future **dfout; rx_Bitset csetout; struct rx_superstate *superstate; unsigned char chr; #endif { struct rx_superset *stateset = superstate->contents; /* To compute the set of characters that share edges with CHR, * we start with the full character set, and subtract. */ rx_bitset_universe (rx->local_cset_size, csetout); *dfout = 0; /* Iterate over the NFA states in the superstate state-set. */ while (stateset->car) { struct rx_nfa_edge *e; for (e = stateset->car->edges; e; e = e->next) if (e->type == ne_cset) { if (!RX_bitset_member (e->params.cset, chr)) /* An edge that doesn't apply at least tells us some characters * that don't share the same edge set as CHR. */ rx_bitset_difference (rx->local_cset_size, csetout, e->params.cset); else { /* If we find an NFA edge that applies, we make sure there * are corresponding edges in the superstate NFA. */ { struct rx_distinct_future * saved; saved = *dfout; *dfout = include_futures (rx, *dfout, e->dest, superstate); if (!*dfout) { struct rx_distinct_future * df; df = saved; if (df) df->next_same_super_edge[1]->next_same_super_edge[0] = 0; while (df) { struct rx_distinct_future *dft; dft = df; df = df->next_same_super_edge[0]; if (dft->future && dft->future->transition_refs == dft) { dft->future->transition_refs = dft->next_same_dest; if (dft->future->transition_refs == dft) dft->future->transition_refs = 0; } dft->next_same_dest->prev_same_dest = dft->prev_same_dest; dft->prev_same_dest->next_same_dest = dft->next_same_dest; rx_cache_free (rx->cache, sizeof (*dft), (char *)dft); } return 0; } } /* We also trim the character set a bit. */ rx_bitset_intersection (rx->local_cset_size, csetout, e->params.cset); } } stateset = stateset->cdr; } return 1; } /* This is a constructor for RX_SUPER_EDGE structures. These are * wrappers for lists of superstate NFA edges that share character sets labels. * If a transition class contains more than one rx_distinct_future (superstate * edge), then it represents a non-determinism in the superstate NFA. */ #ifdef __STDC__ static struct rx_super_edge * rx_super_edge (struct rx *rx, struct rx_superstate *super, rx_Bitset cset, struct rx_distinct_future *df) #else static struct rx_super_edge * rx_super_edge (rx, super, cset, df) struct rx *rx; struct rx_superstate *super; rx_Bitset cset; struct rx_distinct_future *df; #endif { struct rx_super_edge *tc; int tc_size; tc_size = ( sizeof (struct rx_super_edge) + rx_sizeof_bitset (rx->local_cset_size)); tc = ((struct rx_super_edge *) rx_cache_malloc (rx->cache, tc_size)); if (!tc) return 0; tc->next = super->edges; super->edges = tc; tc->rx_backtrack_frame.inx = rx->instruction_table[rx_backtrack_point]; tc->rx_backtrack_frame.data = 0; tc->rx_backtrack_frame.data_2 = (void *) tc; tc->options = df; tc->cset = (rx_Bitset) ((char *) tc + sizeof (*tc)); rx_bitset_assign (rx->local_cset_size, tc->cset, cset); if (df) { struct rx_distinct_future * dfp = df; df->next_same_super_edge[1]->next_same_super_edge[0] = 0; while (dfp) { dfp->edge = tc; dfp = dfp->next_same_super_edge[0]; } df->next_same_super_edge[1]->next_same_super_edge[0] = df; } return tc; } /* There are three kinds of cache miss. The first occurs when a * transition is taken that has never been computed during the * lifetime of the source superstate. That cache miss is handled by * calling COMPUTE_SUPER_EDGE. The second kind of cache miss * occurs when the destination superstate of a transition doesn't * exist. SOLVE_DESTINATION is used to construct the destination superstate. * Finally, the third kind of cache miss occurs when the destination * superstate of a transition is in a `semi-free state'. That case is * handled by UNFREE_SUPERSTATE. * * The function of HANDLE_CACHE_MISS is to figure out which of these * cases applies. */ #ifdef __STDC__ static void install_partial_transition (struct rx_superstate *super, struct rx_inx *answer, RX_subset set, int offset) #else static void install_partial_transition (super, answer, set, offset) struct rx_superstate *super; struct rx_inx *answer; RX_subset set; int offset; #endif { int start = offset; int end = start + 32; RX_subset pos = 1; struct rx_inx * transitions = super->transitions; while (start < end) { if (set & pos) transitions[start] = *answer; pos <<= 1; ++start; } } #ifdef __STDC__ struct rx_inx * rx_handle_cache_miss (struct rx *rx, struct rx_superstate *super, unsigned char chr, void *data) #else struct rx_inx * rx_handle_cache_miss (rx, super, chr, data) struct rx *rx; struct rx_superstate *super; unsigned char chr; void *data; #endif { int offset = chr / RX_subset_bits; struct rx_distinct_future *df = data; if (!df) /* must be the shared_cache_miss_frame */ { /* Perhaps this is just a transition waiting to be filled. */ struct rx_super_edge *tc; RX_subset mask = rx_subset_singletons [chr % RX_subset_bits]; for (tc = super->edges; tc; tc = tc->next) if (tc->cset[offset] & mask) { struct rx_inx * answer; df = tc->options; answer = ((tc->options->next_same_super_edge[0] != tc->options) ? &tc->rx_backtrack_frame : (df->effects ? &df->side_effects_frame : &df->future_frame)); install_partial_transition (super, answer, tc->cset [offset], offset * 32); return answer; } /* Otherwise, it's a flushed or newly encountered edge. */ { char cset_space[1024]; /* this limit is far from unreasonable */ rx_Bitset trcset; struct rx_inx *answer; if (rx_sizeof_bitset (rx->local_cset_size) > sizeof (cset_space)) return 0; /* If the arbitrary limit is hit, always fail */ /* cleanly. */ trcset = (rx_Bitset)cset_space; rx_lock_superstate (rx, super); if (!compute_super_edge (rx, &df, trcset, super, chr)) { rx_unlock_superstate (rx, super); return 0; } if (!df) /* We just computed the fail transition. */ { static struct rx_inx shared_fail_frame = { 0, 0, (void *)rx_backtrack, 0 }; answer = &shared_fail_frame; } else { tc = rx_super_edge (rx, super, trcset, df); if (!tc) { rx_unlock_superstate (rx, super); return 0; } answer = ((tc->options->next_same_super_edge[0] != tc->options) ? &tc->rx_backtrack_frame : (df->effects ? &df->side_effects_frame : &df->future_frame)); } install_partial_transition (super, answer, trcset[offset], offset * 32); rx_unlock_superstate (rx, super); return answer; } } else if (df->future) /* A cache miss on an edge with a future? Must be * a semi-free destination. */ { if (df->future->is_semifree) refresh_semifree_superstate (rx->cache, df->future); return &df->future_frame; } else /* no future superstate on an existing edge */ { rx_lock_superstate (rx, super); if (!solve_destination (rx, df)) { rx_unlock_superstate (rx, super); return 0; } if (!df->effects && (df->edge->options->next_same_super_edge[0] == df->edge->options)) install_partial_transition (super, &df->future_frame, df->edge->cset[offset], offset * 32); rx_unlock_superstate (rx, super); return &df->future_frame; } } dist/rx/rxunfa.c100660 765 144 13230 6071416221 12502 0ustar lordusers/* classes: src_files */ /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "rxall.h" #include "rx.h" #include "rxunfa.h" #include "rxnfa.h" #ifdef __STDC__ static int unfa_equal (void * va, void * vb) #else static int unfa_equal (va, vb) void * va; void * vb; #endif { return rx_rexp_equal ((struct rexp_node *)va, (struct rexp_node *)vb); } struct rx_hash_rules unfa_rules = { unfa_equal, 0, 0, 0, 0 }; #ifdef __STDC__ static struct rx_cached_rexp * canonical_unfa (struct rx_hash * table, struct rexp_node * rexp, int cset_size) #else static struct rx_cached_rexp * canonical_unfa (table, rexp, cset_size) struct rx_hash * table; struct rexp_node * rexp; int cset_size; #endif { struct rx_hash_item * it; it = rx_hash_store (table, rx_rexp_hash (rexp, 0), rexp, &unfa_rules); if (it->binding == 0) { static int rx_id = 0; struct rx * new_rx; struct rx_cached_rexp * cr; if (it->data == (void *)rexp) rx_save_rexp (rexp); cr = (struct rx_cached_rexp *)malloc (sizeof (*cr)); bzero (cr, sizeof (*cr)); if (!cr) return 0; it->binding = (void *)cr; cr->unfa.nfa = 0; cr->unfa.exp = rexp; cr->hash_item = it; rx_save_rexp (rexp); } return (struct rx_cached_rexp *)it->binding; } #ifdef __STDC__ struct rx * rx_unfa_rx (struct rx_cached_rexp * cr, struct rexp_node * exp, int cset_size) #else struct rx * rx_unfa_rx (cr, exp, cset_size) struct rx_cached_rexp * cr; struct rexp_node * exp; int cset_size; #endif { struct rx * new_rx; if (cr->unfa.nfa) return cr->unfa.nfa; new_rx = rx_make_rx (cset_size); if (!new_rx) return 0; { struct rx_nfa_state * start; struct rx_nfa_state * end; start = end = 0; if (!rx_build_nfa (new_rx, exp, &start, &end)) { free (new_rx); return 0; } new_rx->start_nfa_states = start; end->is_final = 1; start->is_start = 1; { struct rx_nfa_state * s; int x; x = 0; for (s = new_rx->nfa_states; s; s = s->next) s->id = x++; } } cr->unfa.nfa = new_rx; return new_rx; } #ifdef __STDC__ struct rx_unfaniverse * rx_make_unfaniverse (int delay) #else struct rx_unfaniverse * rx_make_unfaniverse (delay) int delay; #endif { struct rx_unfaniverse * it; it = (struct rx_unfaniverse *)malloc (sizeof (*it)); if (!it) return 0; bzero (it, sizeof (*it)); it->delay = delay; return it; } #ifdef __STDC__ void rx_free_unfaniverse (struct rx_unfaniverse * it) #else void rx_free_unfaniverse (it) struct rx_unfaniverse * it; #endif { } #ifdef __STDC__ struct rx_unfa * rx_unfa (struct rx_unfaniverse * unfaniverse, struct rexp_node * exp, int cset_size) #else struct rx_unfa * rx_unfa (unfaniverse, exp, cset_size) struct rx_unfaniverse * unfaniverse; struct rexp_node * exp; int cset_size; #endif { struct rx_cached_rexp * cr; if (exp && exp->cr) cr = exp->cr; else { cr = canonical_unfa (&unfaniverse->table, exp, cset_size); if (exp) exp->cr = cr; } if (!cr) return 0; if (cr->next) { if (unfaniverse->free_queue == cr) { unfaniverse->free_queue = cr->next; if (unfaniverse->free_queue == cr) unfaniverse->free_queue = 0; } cr->next->prev = cr->prev; cr->prev->next = cr->next; cr->next = 0; cr->prev = 0; --unfaniverse->delayed; } ++cr->unfa.refs; cr->unfa.cset_size = cset_size; cr->unfa.verse = unfaniverse; rx_unfa_rx (cr, exp, cset_size); return &cr->unfa; } #ifdef __STDC__ void rx_free_unfa (struct rx_unfa * unfa) #else void rx_free_unfa (unfa) struct rx_unfa * unfa; #endif { struct rx_cached_rexp * cr; cr = (struct rx_cached_rexp *)unfa; if (!cr) return; if (!--cr->unfa.refs) { if (!unfa->verse->free_queue) { unfa->verse->free_queue = cr; cr->next = cr->prev = cr; } else { cr->next = unfa->verse->free_queue; cr->prev = unfa->verse->free_queue->prev; cr->next->prev = cr; cr->prev->next = cr; } ++unfa->verse->delayed; while (unfa->verse->delayed > unfa->verse->delay) { struct rx_cached_rexp * it; it = unfa->verse->free_queue; unfa->verse->free_queue = it->next; if (!--unfa->verse->delayed) unfa->verse->free_queue = 0; it->prev->next = it->next; it->next->prev = it->prev; if (it->unfa.exp) it->unfa.exp->cr = 0; rx_free_rexp ((struct rexp_node *)it->hash_item->data); rx_hash_free (it->hash_item, &unfa_rules); rx_free_rx (it->unfa.nfa); rx_free_rexp (it->unfa.exp); free (it); if (it == cr) break; } } else return; } dist/rx/_rx.h100660 765 144 12763 6071416222 12007 0ustar lordusers/* classes: h_files */ #ifndef _RXH #define _RXH /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include #include "rxhash.h" #include "rxcset.h" struct rx_cache; struct rx_superset; struct rx; struct rx_se_list; /* Suppose that from some NFA state and next character, more than one * path through side-effect edges is possible. In what order should * the paths be tried? A function of type rx_se_list_order answers * that question. It compares two lists of side effects, and says * which list comes first. */ #ifdef __STDC__ typedef int (*rx_se_list_order) (struct rx *, struct rx_se_list *, struct rx_se_list *); #else typedef int (*rx_se_list_order) (); #endif /* Struct RX holds an NFA and cache state for the corresponding super NFA. */ struct rx { /* The compiler assigns a unique id to every pattern. * Like sequence numbers in X, there is a subtle bug here * if you use Rx in a system that runs for a long time. * But, because of the way the caches work out, it is almost * impossible to trigger the Rx version of this bug. * * The id is used to validate superstates found in a cache * of superstates. It isn't sufficient to let a superstate * point back to the rx for which it was compiled -- the caller * may be re-using a `struct rx' in which case the superstate * is not really valid. So instead, superstates are validated * by checking the sequence number of the pattern for which * they were built. */ int rx_id; /* This is memory mgt. state for superstates. This may be * shared by more than one struct rx. */ struct rx_cache * cache; /* Every nfa defines the size of its own character set. * Each superstate has an array of this size, with each element * a `struct rx_inx'. So, don't make this number too large. * In particular, don't make it 2^16. */ int local_cset_size; /* Lists of side effects as stored in the NFA are `hash consed'..meaning * that lists with the same elements are ==. During compilation, * this table facilitates hash-consing. */ struct rx_hash se_list_memo; /* Lists of NFA states are also hashed. */ struct rx_hash set_list_memo; /* The compiler and matcher must build a number of instruction frames. * The format of these frames is fixed (c.f. struct rx_inx). The values * of the instruction opcodes is not fixed. * * An enumerated type (enum rx_opcode) defines the set of instructions * that the compiler or matcher might generate. When filling an instruction * frame, the INX field is found by indexing this instruction table * with an opcode: */ void ** instruction_table; /* The list of all states in an NFA. * The NEXT field of NFA states links this list. */ struct rx_nfa_state *nfa_states; struct rx_nfa_state *start_nfa_states; struct rx_superset * start_set; /* This orders the search through super-nfa paths. * See the comment near the typedef of rx_se_list_order. */ rx_se_list_order se_list_cmp; int next_nfa_id; }; /* Type for byte offsets within the string. POSIX mandates this. */ typedef int regoff_t; /* This is the structure we store register match data in. See * regex.texinfo for a full description of what registers match. */ struct re_registers { unsigned num_regs; regoff_t *start; regoff_t *end; }; /* If `regs_allocated' is REGS_UNALLOCATED in the pattern buffer, * `re_match_2' returns information about at least this many registers * the first time a `regs' structure is passed. * * Also, this is the greatest number of backreferenced subexpressions * allowed in a pattern being matched without caller-supplied registers. */ #ifndef RE_NREGS #define RE_NREGS 30 #endif #ifndef emacs #define CHARBITS 8 #define CHAR_SET_SIZE (1 << CHARBITS) #define Sword 1 #define SYNTAX(c) re_syntax_table[c] extern char re_syntax_table[CHAR_SET_SIZE]; #endif /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we * use `alloca' instead of `malloc' for the backtracking stack. * * Emacs will die miserably if we don't do this. */ #ifdef REGEX_MALLOC #define REGEX_ALLOCATE malloc #else /* not REGEX_MALLOC */ #define REGEX_ALLOCATE alloca #endif /* not REGEX_MALLOC */ #undef MAX #undef MIN #define MAX(a, b) ((a) > (b) ? (a) : (b)) #define MIN(a, b) ((a) < (b) ? (a) : (b)) extern void * rx_id_instruction_table[]; extern struct rx_cache * rx_default_cache; #ifdef __STDC__ #else /* STDC */ #endif /* STDC */ #endif /* _RXH */ dist/rx/rx.h100660 765 144 2657 6071417412 11633 0ustar lordusers/* classes: h_files */ #ifndef RXH #define RXH /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "rxhash.h" extern const char rx_version_string[]; #ifdef __STDC__ extern struct rx * rx_make_rx (int cset_size); extern void rx_free_rx (struct rx * rx); #else /* STDC */ extern struct rx * rx_make_rx (); extern void rx_free_rx (); #endif /* STDC */ #endif /* RXH */ dist/rx/rxanal.h100660 765 144 5623 6071416222 12461 0ustar lordusers/* classes: h_files */ #ifndef RXANALH #define RXANALH /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "rxcset.h" #include "rxnode.h" #include "rxsuper.h" enum rx_answers { rx_yes = 0, rx_no = 1, rx_maybe = 2, rx_bogus = -1, rx_start_state_with_too_many_futures = rx_bogus - 1, /* n < 0 -- error */ }; struct rx_classical_system { struct rx * rx; struct rx_superstate * state; }; #ifdef __STDC__ extern int rx_posix_analyze_rexp (struct rexp_node *** subexps, int * n_subexps, struct rexp_node * node, int id); extern int rx_fill_in_fastmap (int cset_size, unsigned char * map, struct rexp_node * exp); extern int rx_is_anchored_p (struct rexp_node * exp); extern enum rx_answers rx_start_superstate (struct rx_classical_system * frame); extern enum rx_answers rx_match_here_p (struct rx_classical_system * frame, unsigned char * burst, int len); extern enum rx_answers rx_fit_p (struct rx_classical_system * frame, unsigned char * burst, int len); extern enum rx_answers rx_longest (int * last_matched, int base_addr, struct rx_classical_system * frame, unsigned char * burst, int len); extern enum rx_answers rx_advance (struct rx_classical_system * frame, unsigned char * burst, int len); extern void rx_terminate_system (struct rx_classical_system * frame); extern void rx_init_system (struct rx_classical_system * frame, struct rx * rx); #else /* STDC */ extern int rx_posix_analyze_rexp (); extern int rx_fill_in_fastmap (); extern int rx_is_anchored_p (); extern enum rx_answers rx_start_superstate (); extern enum rx_answers rx_match_here_p (); extern enum rx_answers rx_fit_p (); extern enum rx_answers rx_longest (); extern enum rx_answers rx_advance (); extern void rx_terminate_system (); extern void rx_init_system (); #endif /* STDC */ #endif /* RXANALH */ dist/rx/rxbasic.h100660 765 144 3334 6071416222 12624 0ustar lordusers/* classes: h_files */ #ifndef RXBASICH #define RXBASICH /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "rxcontext.h" #include "rxnode.h" #include "rxspencer.h" #ifdef __STDC__ extern struct rx_solutions * rx_basic_make_solutions (struct rx_registers * regs, struct rexp_node * expression, struct rexp_node ** subexps, int start, int end, struct rx_context_rules * rules, char * str); extern void rx_basic_free_solutions (struct rx_solutions * solns); #else /* STDC */ extern struct rx_solutions * rx_basic_make_solutions (); extern void rx_basic_free_solutions (); #endif /* STDC */ #endif /* RXBASICH */ dist/rx/rxbitset.h100660 765 144 7615 6071421366 13050 0ustar lordusers/* classes: h_files */ #ifndef RXBITSETH #define RXBITSETH /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ typedef unsigned long RX_subset; #define RX_subset_bits (8 * sizeof (RX_subset)) #define RX_subset_mask (RX_subset_bits - 1) typedef RX_subset * rx_Bitset; #ifdef __STDC__ typedef void (*rx_bitset_iterator) (rx_Bitset, int member_index); #else typedef void (*rx_bitset_iterator) (); #endif /* Return the index of the word containing the Nth bit. */ #define rx_bitset_subset(N) ((N) / RX_subset_bits) /* Return the conveniently-sized supset containing the Nth bit. */ #define rx_bitset_subset_val(B,N) ((B)[rx_bitset_subset(N)]) /* Genericly combine the word containing the Nth bit with a 1 bit mask * of the Nth bit position within that word. */ #define RX_bitset_access(B,N,OP) \ ((B)[rx_bitset_subset(N)] OP rx_subset_singletons[(N) & RX_subset_mask]) #define RX_bitset_member(B,N) RX_bitset_access(B, N, &) #define RX_bitset_enjoin(B,N) RX_bitset_access(B, N, |=) #define RX_bitset_remove(B,N) RX_bitset_access(B, N, &= ~) #define RX_bitset_toggle(B,N) RX_bitset_access(B, N, ^= ) /* How many words are needed for N bits? */ #define rx_bitset_numb_subsets(N) (((N) + RX_subset_bits - 1) / RX_subset_bits) /* How much memory should be allocated for a bitset with N bits? */ #define rx_sizeof_bitset(N) (rx_bitset_numb_subsets(N) * sizeof(RX_subset)) extern RX_subset rx_subset_singletons[]; #ifdef __STDC__ extern int rx_bitset_is_equal (int size, rx_Bitset a, rx_Bitset b); extern int rx_bitset_is_subset (int size, rx_Bitset a, rx_Bitset b); extern int rx_bitset_empty (int size, rx_Bitset set); extern void rx_bitset_null (int size, rx_Bitset b); extern void rx_bitset_universe (int size, rx_Bitset b); extern void rx_bitset_complement (int size, rx_Bitset b); extern void rx_bitset_assign (int size, rx_Bitset a, rx_Bitset b); extern void rx_bitset_union (int size, rx_Bitset a, rx_Bitset b); extern void rx_bitset_intersection (int size, rx_Bitset a, rx_Bitset b); extern void rx_bitset_difference (int size, rx_Bitset a, rx_Bitset b); extern void rx_bitset_revdifference (int size, rx_Bitset a, rx_Bitset b); extern void rx_bitset_xor (int size, rx_Bitset a, rx_Bitset b); extern unsigned long rx_bitset_hash (int size, rx_Bitset b); extern int rx_bitset_population (int size, rx_Bitset a); #else /* STDC */ extern int rx_bitset_is_equal (); extern int rx_bitset_is_subset (); extern int rx_bitset_empty (); extern void rx_bitset_null (); extern void rx_bitset_universe (); extern void rx_bitset_complement (); extern void rx_bitset_assign (); extern void rx_bitset_union (); extern void rx_bitset_intersection (); extern void rx_bitset_difference (); extern void rx_bitset_revdifference (); extern void rx_bitset_xor (); extern unsigned long rx_bitset_hash (); extern int rx_bitset_population (); #endif /* STDC */ #endif /* RXBITSETH */ dist/rx/rxcontext.h100660 765 144 2750 6071416222 13230 0ustar lordusers/* classes: h_files */ #ifndef RXCONTEXTH #define RXCONTEXTH /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ struct rx_context_rules { unsigned int newline_anchor:1;/* If true, an anchor at a newline matches.*/ unsigned int not_bol:1; /* If set, the anchors ('^' and '$') don't */ unsigned int not_eol:1; /* match at the ends of the string. */ }; #ifdef __STDC__ #else /* STDC */ #endif /* STDC */ #endif /* RXCONTEXTH */ dist/rx/rxcset.h100660 765 144 3001 6071416221 12467 0ustar lordusers/* classes: h_files */ #ifndef RXCSETH #define RXCSETH /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ /* lord Sun May 7 12:34:11 1995 */ #include "rxbitset.h" #ifdef __STDC__ extern rx_Bitset rx_cset (int size); extern rx_Bitset rx_copy_cset (int size, rx_Bitset a); extern void rx_free_cset (rx_Bitset c); #else /* STDC */ extern rx_Bitset rx_cset (); extern rx_Bitset rx_copy_cset (); extern void rx_free_cset (); #endif /* STDC */ #endif /* RXCSETH */ dist/rx/rxgnucomp.h100660 765 144 22523 6071532113 13232 0ustar lordusers/* classes: h_files */ #ifndef RXGNUCOMPH #define RXGNUCOMPH /* Copyright (C) 1992, 1993, 1994, 1995 Free Software Foundation, Inc. * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU Library General Public License as published by * the Free Software Foundation; either version 2, or (at your option) * any later version. * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU Library General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this software; see the file COPYING. If not, write to * the Free Software Foundation, 59 Temple Place - Suite 330, * Boston, MA 02111-1307, USA. */ #include "rxcset.h" #include "rxnode.h" /* This is an array of error messages corresponding to the error codes. */ extern const char *rx_error_msg[]; /* If any error codes are removed, changed, or added, update the * `rx_error_msg' table. */ typedef enum { REG_NOERROR = 0, /* Success. */ REG_NOMATCH, /* Didn't find a match (for regexec). */ /* POSIX regcomp return error codes. (In the order listed in the standard.) */ REG_BADPAT, /* Invalid pattern. */ REG_ECOLLATE, /* Not implemented. */ REG_ECTYPE, /* Invalid character class name. */ REG_EESCAPE, /* Trailing backslash. */ REG_ESUBREG, /* Invalid back reference. */ REG_EBRACK, /* Unmatched left bracket. */ REG_EPAREN, /* Parenthesis imbalance. */ REG_EBRACE, /* Unmatched \{. */ REG_BADBR, /* Invalid contents of \{\}. */ REG_ERANGE, /* Invalid range end. */ REG_ESPACE, /* Ran out of memory. */ REG_BADRPT, /* No preceding re for repetition op. */ /* Error codes we've added. */ REG_EEND, /* Premature end. */ REG_ESIZE, /* Compiled pattern bigger than 2^16 bytes. */ REG_ERPAREN /* Unmatched ) or \); not returned from regcomp. */ } reg_errcode_t; /* {Syntax Bits} */ /* The following bits are used to determine the regexp syntax we recognize. The set/not-set meanings are chosen so that Emacs syntax remains the value 0. The bits are given in alphabetical order, and the definitions shifted by one from the previous bit; thus, when we add or remove a bit, only one other definition need change. */ /* If this bit is not set, then \ inside a bracket expression is literal. If set, then such a \ quotes the following character. */ #define RE_BACKSLASH_ESCAPE_IN_LISTS (1) /* If this bit is not set, then + and ? are operators, and \+ and \? are literals. If set, then \+ and \? are operators and + and ? are literals. */ #define RE_BK_PLUS_QM (RE_BACKSLASH_ESCAPE_IN_LISTS << 1) /* If this bit is set, then character classes are supported. They are: [:alpha:], [:upper:], [:lower:], [:digit:], [:alnum:], [:xdigit:], [:space:], [:print:], [:punct:], [:graph:], and [:cntrl:]. If not set, then character classes are not supported. */ #define RE_CHAR_CLASSES (RE_BK_PLUS_QM << 1) /* If this bit is set, then ^ and $ are always anchors (outside bracket expressions, of course). If this bit is not set, then it depends: ^ is an anchor if it is at the beginning of a regular expression or after an open-group or an alternation operator; $ is an anchor if it is at the end of a regular expression, or before a close-group or an alternation operator. This bit could be (re)combined with RE_CONTEXT_INDEP_OPS, because POSIX draft 11.2 says that * etc. in leading positions is undefined. We already implemented a previous draft which made those constructs invalid, though, so we haven't changed the code back. */ #define RE_CONTEXT_INDEP_ANCHORS (RE_CHAR_CLASSES << 1) /* If this bit is set, then special characters are always special regardless of where they are in the pattern. If this bit is not set, then special characters are special only in some contexts; otherwise they are ordinary. Specifically, * + ? and intervals are only special when not after the beginning, open-group, or alternation operator. */ #define RE_CONTEXT_INDEP_OPS (RE_CONTEXT_INDEP_ANCHORS << 1) /* If this bit is set, then *, +, ?, and { cannot be first in an re or immediately after an alternation or begin-group operator. */ #define RE_CONTEXT_INVALID_OPS (RE_CONTEXT_INDEP_OPS << 1) /* If this bit is set, then . matches newline. If not set, then it doesn't. */ #define RE_DOT_NEWLINE (RE_CONTEXT_INVALID_OPS << 1) /* If this bit is set, then . doesn't match NUL. If not set, then it does. */ #define RE_DOT_NOT_NULL (RE_DOT_NEWLINE << 1) /* If this bit is set, nonmatching lists [^...] do not match newline. If not set, they do. */ #define RE_HAT_LISTS_NOT_NEWLINE (RE_DOT_NOT_NULL << 1) /* If this bit is set, either \{...\} or {...} defines an interval, depending on RE_NO_BK_BRACES. If not set, \{, \}, {, and } are literals. */ #define RE_INTERVALS (RE_HAT_LISTS_NOT_NEWLINE << 1) /* If this bit is set, +, ? and | aren't recognized as operators. If not set, they are. */ #define RE_LIMITED_OPS (RE_INTERVALS << 1) /* If this bit is set, newline is an alternation operator. If not set, newline is literal. */ #define RE_NEWLINE_ALT (RE_LIMITED_OPS << 1) /* If this bit is set, then `{...}' defines an interval, and \{ and \} are literals. If not set, then `\{...\}' defines an interval. */ #define RE_NO_BK_BRACES (RE_NEWLINE_ALT << 1) /* If this bit is set, (...) defines a group, and \( and \) are literals. If not set, \(...\) defines a group, and ( and ) are literals. */ #define RE_NO_BK_PARENS (RE_NO_BK_BRACES << 1) /* If this bit is set, then \ matches . If not set, then \ is a back-reference. */ #define RE_NO_BK_REFS (RE_NO_BK_PARENS << 1) /* If this bit is set, then | is an alternation operator, and \| is literal. If not set, then \| is an alternation operator, and | is literal. */ #define RE_NO_BK_VBAR (RE_NO_BK_REFS << 1) /* If this bit is set, then an ending range point collating higher than the starting range point, as in [z-a], is invalid. If not set, then when ending range point collates higher than the starting range point, the range is ignored. */ #define RE_NO_EMPTY_RANGES (RE_NO_BK_VBAR << 1) /* If this bit is set, then an unmatched ) is ordinary. If not set, then an unmatched ) is invalid. */ #define RE_UNMATCHED_RIGHT_PAREN_ORD (RE_NO_EMPTY_RANGES << 1) /* Define combinations of the above bits for the standard possibilities. (The [[[ comments delimit what gets put into the Texinfo file, so don't delete them!) */ /* [[[begin syntaxes]]] */ #define RE_SYNTAX_EMACS 0 #define RE_SYNTAX_AWK \ (RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DOT_NOT_NULL \ | RE_NO_BK_PARENS | RE_NO_BK_REFS \ | RE_NO_BK_VBAR | RE_NO_EMPTY_RANGES \ | RE_UNMATCHED_RIGHT_PAREN_ORD) #define RE_SYNTAX_POSIX_AWK \ (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS) #define RE_SYNTAX_GREP \ (RE_BK_PLUS_QM | RE_CHAR_CLASSES \ | RE_HAT_LISTS_NOT_NEWLINE | RE_INTERVALS \ | RE_NEWLINE_ALT) #define RE_SYNTAX_EGREP \ (RE_CHAR_CLASSES | RE_CONTEXT_INDEP_ANCHORS \ | RE_CONTEXT_INDEP_OPS | RE_HAT_LISTS_NOT_NEWLINE \ | RE_NEWLINE_ALT | RE_NO_BK_PARENS \ | RE_NO_BK_VBAR) #define RE_SYNTAX_POSIX_EGREP \ (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES) #define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC /* Syntax bits common to both basic and extended POSIX regex syntax. */ #define _RE_SYNTAX_POSIX_COMMON \ (RE_CHAR_CLASSES | RE_DOT_NEWLINE | RE_DOT_NOT_NULL \ | RE_INTERVALS | RE_NO_EMPTY_RANGES) #define RE_SYNTAX_POSIX_BASIC \ (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM) /* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes RE_LIMITED_OPS, i.e., \? \+ \| are not recognized. Actually, this isn't minimal, since other operators, such as \`, aren't disabled. */ #define RE_SYNTAX_POSIX_MINIMAL_BASIC \ (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS) #define RE_SYNTAX_POSIX_EXTENDED \ (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \ | RE_CONTEXT_INDEP_OPS | RE_NO_BK_BRACES \ | RE_NO_BK_PARENS | RE_NO_BK_VBAR \ | RE_UNMATCHED_RIGHT_PAREN_ORD) /* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INVALID_OPS replaces RE_CONTEXT_INDEP_OPS and RE_NO_BK_REFS is added. */ #define RE_SYNTAX_POSIX_MINIMAL_EXTENDED \ (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \ | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES \ | RE_NO_BK_PARENS | RE_NO_BK_REFS \ | RE_NO_BK_VBAR | RE_UNMATCHED_RIGHT_PAREN_ORD) /* [[[end syntaxes]]] */ /* Maximum number of duplicates an interval can allow. Some systems (erroneously) define this in other header files, but we want our value, so remove any previous define. */ #ifdef RE_DUP_MAX #undef RE_DUP_MAX #endif #define RE_DUP_MAX ((1 << 15) - 1) #ifdef __STDC__ extern reg_errcode_t rx_parse (struct rexp_node ** rexp_p, const char *pattern, int size, unsigned long syntax, int cset_size, unsigned char *translate); #else /* STDC */ extern reg_errcode_t rx_parse (); #endif /* STDC */ #endif /* RXGNUCOMPH */ dist/rx/rxhash.h100660 765 144 6403 6071416221 12465 0ustar lordusers/* classes: h_files */ #ifndef RXHASHH #define RXHASHH /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ /* * Tom Lord (lord@cygnus.com, lord@gnu.ai.mit.edu) */ #include "rxbitset.h" /* giant inflatable hash trees */ struct rx_hash_item { struct rx_hash_item * next_same_hash; struct rx_hash * table; unsigned long hash; void * data; void * binding; }; struct rx_hash { struct rx_hash * parent; int refs; RX_subset nested_p; void ** children[16]; }; struct rx_hash_rules; /* rx_hash_eq should work like the == operator. */ #ifdef __STDC__ typedef int (*rx_hash_eq)(void *, void *); typedef struct rx_hash * (*rx_alloc_hash)(struct rx_hash_rules *); typedef void (*rx_free_hash)(struct rx_hash *, struct rx_hash_rules *); typedef struct rx_hash_item * (*rx_alloc_hash_item)(struct rx_hash_rules *, void *); typedef void (*rx_free_hash_item)(struct rx_hash_item *, struct rx_hash_rules *); typedef void (*rx_hash_freefn) (struct rx_hash_item * it); #else typedef int (*rx_hash_eq)(); typedef struct rx_hash * (*rx_alloc_hash)(); typedef void (*rx_free_hash)(); typedef struct rx_hash_item * (*rx_alloc_hash_item)(); typedef void (*rx_free_hash_item)(); typedef void (*rx_hash_freefn) (); #endif struct rx_hash_rules { rx_hash_eq eq; rx_alloc_hash hash_alloc; rx_free_hash free_hash; rx_alloc_hash_item hash_item_alloc; rx_free_hash_item free_hash_item; }; #ifdef __STDC__ extern struct rx_hash_item * rx_hash_find (struct rx_hash * table, unsigned long hash, void * value, struct rx_hash_rules * rules); extern struct rx_hash_item * rx_hash_store (struct rx_hash * table, unsigned long hash, void * value, struct rx_hash_rules * rules); extern void rx_hash_free (struct rx_hash_item * it, struct rx_hash_rules * rules); extern void rx_free_hash_table (struct rx_hash * tab, rx_hash_freefn freefn, struct rx_hash_rules * rules); extern int rx_count_hash_nodes (struct rx_hash * st); #else /* STDC */ extern struct rx_hash_item * rx_hash_find (); extern struct rx_hash_item * rx_hash_store (); extern void rx_hash_free (); extern void rx_free_hash_table (); extern int rx_count_hash_nodes (); #endif /* STDC */ #endif /* RXHASHH */ dist/rx/rxnfa.h100660 765 144 15230 6071416221 12324 0ustar lordusers/* classes: h_files */ #ifndef RXNFAH #define RXNFAH /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ /* * Tom Lord (lord@cygnus.com, lord@gnu.ai.mit.edu) */ #include "_rx.h" #include "rxnode.h" /* NFA * * A syntax tree is compiled into an NFA. This page defines the structure * of that NFA. */ struct rx_nfa_state { /* These are kept in a list as the NFA is being built. * Here is the link. */ struct rx_nfa_state *next; /* After the NFA is built, states are given integer id's. * States whose outgoing transitions are all either epsilon or * side effect edges are given ids less than 0. Other states * are given successive non-negative ids starting from 0. * * Here is the id for this state: */ int id; /* The list of NFA edges that go from this state to some other. */ struct rx_nfa_edge *edges; /* If you land in this state, then you implicitly land * in all other states reachable by only epsilon translations. * Call the set of maximal loop-less paths to such states the * epsilon closure of this state. * * There may be other states that are reachable by a mixture of * epsilon and side effect edges. Consider the set of maximal loop-less * paths of that sort from this state. Call it the epsilon-side-effect * closure of the state. * * The epsilon closure of the state is a subset of the epsilon-side- * effect closure. It consists of all the paths that contain * no side effects -- only epsilon edges. * * The paths in the epsilon-side-effect closure can be partitioned * into equivalance sets. Two paths are equivalant if they have the * same set of side effects, in the same order. The epsilon-closure * is one of these equivalance sets. Let's call these equivalance * sets: observably equivalant path sets. That name is chosen * because equivalance of two paths means they cause the same side * effects -- so they lead to the same subsequent observations other * than that they may wind up in different target states. * * The superstate nfa, which is derived from this nfa, is based on * the observation that all of the paths in an observably equivalant * path set can be explored at the same time, provided that the * matcher keeps track not of a single nfa state, but of a set of * states. In particular, after following all the paths in an * observably equivalant set, you wind up at a set of target states. * That set of target states corresponds to one state in the * superstate NFA. * * Staticly, before matching begins, it is convenient to analyze the * nfa. Each state is labeled with a list of the observably * equivalant path sets who's union covers all the * epsilon-side-effect paths beginning in this state. This list is * called the possible futures of the state. * * A trivial example is this NFA: * s1 * A ---> B * * s2 * ---> C * * epsilon s1 * ---------> D ------> E * * * In this example, A has two possible futures. * One invokes the side effect `s1' and contains two paths, * one ending in state B, the other in state E. * The other invokes the side effect `s2' and contains only * one path, landing in state C. * * Here is a list of the possible futures of this state: */ struct rx_possible_future *futures; int futures_computed:1; /* There are exactly two distinguished states in every NFA: */ unsigned int is_final:1; unsigned int is_start:1; /* These are used internally during NFA construction... */ unsigned int eclosure_needed:1; unsigned int mark:1; }; /* An edge in an NFA is typed: */ enum rx_nfa_etype { /* A cset edge is labled with a set of characters one of which * must be matched for the edge to be taken. */ ne_cset, /* An epsilon edge is taken whenever its starting state is * reached. */ ne_epsilon, /* A side effect edge is taken whenever its starting state is * reached. Side effects may cause the match to fail or the * position of the matcher to advance. */ ne_side_effect }; struct rx_nfa_edge { struct rx_nfa_edge *next; enum rx_nfa_etype type; struct rx_nfa_state *dest; union { rx_Bitset cset; void * side_effect; } params; }; /* A possible future consists of a list of side effects * and a set of destination states. Below are their * representations. These structures are hash-consed so * that lists with the same elements share a representation * (their addresses are ==). */ struct rx_nfa_state_set { struct rx_nfa_state * car; struct rx_nfa_state_set * cdr; }; struct rx_se_list { void * car; struct rx_se_list * cdr; }; struct rx_possible_future { struct rx_possible_future *next; struct rx_se_list * effects; struct rx_nfa_state_set * destset; }; #ifdef __STDC__ extern struct rx_nfa_state * rx_nfa_state (struct rx *rx); extern struct rx_nfa_edge * rx_nfa_edge (struct rx *rx, enum rx_nfa_etype type, struct rx_nfa_state *start, struct rx_nfa_state *dest); extern int rx_build_nfa (struct rx *rx, struct rexp_node *rexp, struct rx_nfa_state **start, struct rx_nfa_state **end); extern struct rx_possible_future * rx_state_possible_futures (struct rx * rx, struct rx_nfa_state * n); extern void rx_free_nfa (struct rx *rx); #else /* STDC */ extern struct rx_nfa_state * rx_nfa_state (); extern struct rx_nfa_edge * rx_nfa_edge (); extern int rx_build_nfa (); extern struct rx_possible_future * rx_state_possible_futures (); extern void rx_free_nfa (); #endif /* STDC */ #endif /* RXNFAH */ dist/rx/rxnode.h100660 765 144 6217 6071416221 12472 0ustar lordusers/* classes: h_files */ #ifndef RXNODEH #define RXNODEH /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ /* * Tom Lord (lord@cygnus.com, lord@gnu.ai.mit.edu) */ #include "rxbitset.h" #include "rxcset.h" enum rexp_node_type { r_cset, /* Match from a character set. `a' or `[a-z]'*/ r_concat, /* Concat two subexpressions. `ab' */ r_alternate, /* Choose one of two subexpressions. `a\|b' */ r_opt, /* Optional subexpression. `a?' */ r_star, /* Repeated subexpression. `a*' */ r_plus, /* Nontrivially repeated subexpression. `a+' */ r_interval, /* Counted subexpression. `a{4,1000}' */ r_parens, r_context }; struct rexp_node { int refs; enum rexp_node_type type; struct { int cset_size; rx_Bitset cset; int intval; int intval2; struct { struct rexp_node *left; struct rexp_node *right; } pair; } params; int id; int len; int observed; struct rexp_node * simplified; struct rx_cached_rexp * cr; }; #ifdef __STDC__ extern struct rexp_node * rexp_node (int type); extern struct rexp_node * rx_mk_r_cset (int type, int size, rx_Bitset b); extern struct rexp_node * rx_mk_r_int (int type, int intval); extern struct rexp_node * rx_mk_r_binop (int type, struct rexp_node * a, struct rexp_node * b); extern struct rexp_node * rx_mk_r_monop (int type, struct rexp_node * a); extern void rx_free_rexp (struct rexp_node * node); extern void rx_save_rexp (struct rexp_node * node); extern struct rexp_node * rx_copy_rexp (int cset_size, struct rexp_node *node); extern int rx_rexp_equal (struct rexp_node * a, struct rexp_node * b); extern unsigned long rx_rexp_hash (struct rexp_node * node, unsigned long seed); #else /* STDC */ extern struct rexp_node * rexp_node (); extern struct rexp_node * rx_mk_r_cset (); extern struct rexp_node * rx_mk_r_int (); extern struct rexp_node * rx_mk_r_binop (); extern struct rexp_node * rx_mk_r_monop (); extern void rx_free_rexp (); extern void rx_save_rexp (); extern struct rexp_node * rx_copy_rexp (); extern int rx_rexp_equal (); extern unsigned long rx_rexp_hash (); #endif /* STDC */ #endif /* RXNODEH */ dist/rx/inst-rxposix.h100660 765 144 7240 6071422250 13656 0ustar lordusers/* classes: h_files */ #ifndef INST_RXPOSIXH #define INST_RXPOSIXH /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ struct rx_posix_regex { struct rexp_node * pattern; struct rexp_node ** subexps; int n_subexps; unsigned char * translate; unsigned int newline_anchor:1;/* If true, an anchor at a newline matches.*/ unsigned int no_sub:1; /* If set, don't return register offsets. */ unsigned int is_anchored:1; unsigned char fastmap[256]; }; typedef struct rx_posix_regex regex_t; /* * POSIX `cflags' bits (i.e., information for `regcomp'). */ /* If this bit is set, then use extended regular expression syntax. * If not set, then use basic regular expression syntax. */ #define REG_EXTENDED 1 /* If this bit is set, then ignore case when matching. * If not set, then case is significant. */ #define REG_ICASE (REG_EXTENDED << 1) /* If this bit is set, then anchors do not match at newline * characters in the string. * If not set, then anchors do match at newlines. */ #define REG_NEWLINE (REG_ICASE << 1) /* If this bit is set, then report only success or fail in regexec. * If not set, then returns differ between not matching and errors. */ #define REG_NOSUB (REG_NEWLINE << 1) /* * POSIX `eflags' bits (i.e., information for regexec). */ /* If this bit is set, then the beginning-of-line operator doesn't match * the beginning of the string (presumably because it's not the * beginning of a line). * If not set, then the beginning-of-line operator does match the * beginning of the string. */ #define REG_NOTBOL 1 /* Like REG_NOTBOL, except for the end-of-line. */ #define REG_NOTEOL (1 << 1) #ifdef __STDC__ extern int regcomp (regex_t * preg, const char * pattern, int cflags); extern size_t regerror (int errcode, const regex_t *preg, char *errbuf, size_t errbuf_size); extern int rx_regmatch (regmatch_t pmatch[], regex_t *preg, struct rx_context_rules * rules, int start, int end, char *string); extern int rx_regexec (regmatch_t pmatch[], regex_t *preg, struct rx_context_rules * rules, int start, int end, char *string); extern int regnexec (regex_t *preg, int len, char *string, size_t nmatch, regmatch_t pmatch[], int eflags); extern int regexec (regex_t *preg, char *string, size_t nmatch, regmatch_t pmatch[], int eflags); extern void regfree (regex_t *preg); #else /* STDC */ extern int regcomp (); extern size_t regerror (); extern int rx_regmatch (); extern int rx_regexec (); extern int regnexec (); extern int regexec (); extern void regfree (); #endif /* STDC */ #endif /* INST_RXPOSIXH */ dist/rx/rxposix.h100660 765 144 2356 6071422513 12710 0ustar lordusers/* classes: h_files */ #ifndef RXPOSIXH #define RXPOSIXH /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "rxspencer.h" #include "rxcontext.h" #include "inst-rxposix.h" #endif RXPOSIXH dist/rx/rxproto.h100660 765 144 2445 6071416221 12707 0ustar lordusers/* classes: h_files */ #ifndef RXPROTOH #define RXPROTOH /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #ifdef __STDC__ #define P(X) X #else #define P(X) () #endif #ifdef __STDC__ #else /* STDC */ #endif /* STDC */ #endif /* RXPROTOH */ dist/rx/rxsimp.h100660 765 144 2673 6071416221 12517 0ustar lordusers/* classes: h_files */ #ifndef RXSIMPH #define RXSIMPH /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "rxcset.h" #include "rxnode.h" #ifdef __STDC__ extern int rx_simple_rexp (struct rexp_node ** answer, int cset_size, struct rexp_node *node, struct rexp_node ** subexps); #else /* STDC */ extern int rx_simple_rexp (); #endif /* STDC */ #endif /* RXSIMPH */ dist/rx/rxspencer.h100660 765 144 5426 6071416221 13205 0ustar lordusers/* classes: h_files */ #ifndef RXSPENCERH #define RXSPENCERH /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "rxproto.h" #include "rxnode.h" #include "rxunfa.h" #include "rxanal.h" typedef struct rx_registers { regoff_t rm_so; /* Byte offset from string's start to substring's start. */ regoff_t rm_eo; /* Byte offset from string's start to substring's end. */ } regmatch_t; typedef enum rx_answers (*rx_vmfn) P((void * closure, unsigned char ** burst, int * len, int * offset, int start, int end, int need)); typedef enum rx_answers (*rx_contextfn) P((void * closure, int type, int start, int end, struct rx_registers * regs)); struct rx_solutions { int step; int cset_size; struct rexp_node * exp; struct rexp_node ** subexps; struct rx_registers * regs; int start; int end; rx_vmfn vmfn; rx_contextfn contextfn; void * closure; int current_pos; struct rx_unfaniverse * verse; struct rx_unfa * dfa; struct rx_classical_system match_engine; int split_guess; struct rx_solutions * left; struct rx_solutions * right; int interval_x; }; #ifdef __STDC__ extern struct rx_solutions * rx_make_solutions (struct rx_registers * regs, struct rx_unfaniverse * verse, struct rexp_node * expression, struct rexp_node ** subexps, int cset_size, int start, int end, rx_vmfn vmfn, rx_contextfn contextfn, void * closure); extern void rx_free_solutions (struct rx_solutions * solns); extern enum rx_answers rx_next_solution (struct rx_solutions * solns); #else /* STDC */ extern struct rx_solutions * rx_make_solutions (); extern void rx_free_solutions (); extern enum rx_answers rx_next_solution (); #endif /* STDC */ #endif /* RXSPENCERH */ dist/rx/rxstr.h100660 765 144 3362 6071416221 12353 0ustar lordusers/* classes: h_files */ #ifndef RXSTRH #define RXSTRH /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "rxspencer.h" #include "rxcontext.h" struct rx_str_closure { struct rx_context_rules rules; unsigned char * str; int len; }; #ifdef __STDC__ extern enum rx_answers rx_str_vmfn (void * closure, unsigned char ** burstp, int * lenp, int * offsetp, int start, int end, int need); extern enum rx_answers rx_str_contextfn (void * closure, int type, int start, int end, struct rx_registers * regs); #else /* STDC */ extern enum rx_answers rx_str_vmfn (); extern enum rx_answers rx_str_contextfn (); #endif /* STDC */ #endif /* RXSTRH */ dist/rx/rxsuper.h100660 765 144 40050 6071416221 12714 0ustar lordusers/* classes: h_files */ #ifndef RXSUPERH #define RXSUPERH /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ /* lord Sun May 7 12:40:17 1995 */ #include "rxnfa.h" /* This begins the description of the superstate NFA. * * The superstate NFA corresponds to the NFA in these ways: * * Superstate states correspond to sets of NFA states (nfa_states(SUPER)), * * Superstate edges correspond to NFA paths. * * The superstate has no epsilon transitions; * every edge has a character label, and a (possibly empty) side * effect label. The side effect label corresponds to a list of * side effects that occur in the NFA. These parts are referred * to as: superedge_character(EDGE) and superedge_sides(EDGE). * * For a superstate edge EDGE starting in some superstate SUPER, * the following is true (in pseudo-notation :-): * * exists DEST in nfa_states s.t. * exists nfaEDGE in nfa_edges s.t. * origin (nfaEDGE) == DEST * && origin (nfaEDGE) is a member of nfa_states(SUPER) * && exists PF in possible_futures(dest(nfaEDGE)) s.t. * sides_of_possible_future (PF) == superedge_sides (EDGE) * * also: * * let SUPER2 := superedge_destination(EDGE) * nfa_states(SUPER2) * == union of all nfa state sets S s.t. * exists PF in possible_futures(dest(nfaEDGE)) s.t. * sides_of_possible_future (PF) == superedge_sides (EDGE) * && S == dests_of_possible_future (PF) } * * Or in english, every superstate is a set of nfa states. A given * character and a superstate implies many transitions in the NFA -- * those that begin with an edge labeled with that character from a * state in the set corresponding to the superstate. * * The destinations of those transitions each have a set of possible * futures. A possible future is a list of side effects and a set of * destination NFA states. Two sets of possible futures can be * `merged' by combining all pairs of possible futures that have the * same side effects. A pair is combined by creating a new future * with the same side effect but the union of the two destination sets. * In this way, all the possible futures suggested by a superstate * and a character can be merged into a set of possible futures where * no two elements of the set have the same set of side effects. * * The destination of a possible future, being a set of NFA states, * corresponds to a supernfa state. So, the merged set of possible * futures we just created can serve as a set of edges in the * supernfa. * * The representation of the superstate nfa and the nfa is critical. * The nfa has to be compact, but has to facilitate the rapid * computation of missing superstates. The superstate nfa has to * be fast to interpret, lazilly constructed, and bounded in space. * * To facilitate interpretation, the superstate data structures are * peppered with `instruction frames'. There is an instruction set * defined below which matchers using the supernfa must be able to * interpret. * * We'd like to make it possible but not mandatory to use code * addresses to represent instructions (c.f. gcc's computed goto). * Therefore, we define an enumerated type of opcodes, and when * writing one of these instructions into a data structure, use * the opcode as an index into a table of instruction values. * * Below are the opcodes that occur in the superstate nfa. * * The descriptions of the opcodes refer to data structures that are * described further below. */ enum rx_opcode { /* * BACKTRACK_POINT is invoked when a character transition in * a superstate leads to more than one edge. In that case, * the edges have to be explored independently using a backtracking * strategy. * * A BACKTRACK_POINT instruction is stored in a superstate's * transition table for some character when it is known that that * character crosses more than one edge. On encountering this * instruction, the matcher saves enough state to backtrack to this * point later in the match. */ rx_backtrack_point = 0, /* data is (struct transition_class *) */ /* * RX_DO_SIDE_EFFECTS evaluates the side effects of an epsilon path. * There is one occurence of this instruction per rx_distinct_future. * This instruction is skipped if a rx_distinct_future has no side effects. */ rx_do_side_effects = rx_backtrack_point + 1, /* data is (struct rx_distinct_future *) */ /* * RX_CACHE_MISS instructions are stored in rx_distinct_futures whose * destination superstate has been reclaimed (or was never built). * It recomputes the destination superstate. * RX_CACHE_MISS is also stored in a superstate transition table before * any of its edges have been built. */ rx_cache_miss = rx_do_side_effects + 1, /* data is (struct rx_distinct_future *) */ /* * RX_NEXT_CHAR is called to consume the next character and take the * corresponding transition. This is the only instruction that uses * the DATA field of the instruction frame instead of DATA_2. * The comments about rx_inx explain this further. */ rx_next_char = rx_cache_miss + 1, /* data is (struct superstate *) */ /* RX_BACKTRACK indicates that a transition fails. Don't * confuse this with rx_backtrack_point. */ rx_backtrack = rx_next_char + 1, /* no data */ /* * RX_ERROR_INX is stored only in places that should never be executed. */ rx_error_inx = rx_backtrack + 1, /* Not supposed to occur. */ rx_num_instructions = rx_error_inx + 1 }; /* The heart of the matcher is a `word-code-interpreter' * (like a byte-code interpreter, except that instructions * are a full word wide). * * Instructions are not stored in a vector of code, instead, * they are scattered throughout the data structures built * by the regexp compiler and the matcher. One word-code instruction, * together with the arguments to that instruction, constitute * an instruction frame (struct rx_inx). * * This structure type is padded by hand to a power of 2 because * in one of the dominant cases, we dispatch by indexing a table * of instruction frames. If that indexing can be accomplished * by just a shift of the index, we're happy. * * Instructions take at most one argument, but there are two * slots in an instruction frame that might hold that argument. * These are called data and data_2. The data slot is only * used for one instruction (RX_NEXT_CHAR). For all other * instructions, data should be set to 0. * * RX_NEXT_CHAR is the most important instruction by far. * By reserving the data field for its exclusive use, * instruction dispatch is sped up in that case. There is * no need to fetch both the instruction and the data, * only the data is needed. In other words, a `cycle' begins * by fetching the field data. If that is non-0, then it must * be the destination state of a next_char transition, so * make that value the current state, advance the match position * by one character, and start a new cycle. On the other hand, * if data is 0, fetch the instruction and do a more complicated * dispatch on that. */ struct rx_inx { void * data; void * data_2; void * inx; void * fnord; }; #ifndef RX_TAIL_ARRAY #define RX_TAIL_ARRAY 1 #endif /* A superstate corresponds to a set of nfa states. Those sets are * represented by STRUCT RX_SUPERSET. The constructors * guarantee that only one (shared) structure is created for a given set. */ struct rx_superset { int is_final; int refs; /* This is a reference counted structure. */ /* We keep these sets in a cache because (in an unpredictable way), * the same set is often created again and again. But that is also * problematic -- compatibility with POSIX and GNU regex requires * that we not be able to tell when a program discards a particular * NFA (thus invalidating the supersets created from it). * * But when a cache hit appears to occur, we will have in hand the * nfa for which it may have happened. That is why every nfa is given * its own sequence number. On a cache hit, the cache is validated * by comparing the nfa sequence number to this field: */ int id; struct rx_nfa_state * car; /* May or may not be a valid addr. */ struct rx_superset * cdr; /* If the corresponding superstate exists: */ struct rx_superstate * superstate; /* There is another bookkeeping problem. It is expensive to * compute the starting nfa state set for an nfa. So, once computed, * it is cached in the `struct rx'. * * But, the state set can be flushed from the superstate cache. * When that happens, we can't know if the corresponding `struct rx' * is still alive or if it has been freed or re-used by the program. * So, the cached pointer to this set in a struct rx might be invalid * and we need a way to validate it. * * Fortunately, even if this set is flushed from the cache, it is * not freed. It just goes on the free-list of supersets. * So we can still examine it. * * So to validate a starting set memo, check to see if the * starts_for field still points back to the struct rx in question, * and if the ID matches the rx sequence number. */ struct rx * starts_for; /* This is used to link into a hash bucket so these objects can * be `hash-consed'. */ struct rx_hash_item hash_item; }; #define rx_protect_superset(RX,CON) (++(CON)->refs) /* The terminology may be confusing (rename this structure?). * Every character occurs in at most one rx_super_edge per super-state. * But, that structure might have more than one option, indicating a point * of non-determinism. * * In other words, this structure holds a list of superstate edges * sharing a common starting state and character label. The edges * are in the field OPTIONS. All superstate edges sharing the same * starting state and character are in this list. */ struct rx_super_edge { struct rx_super_edge *next; struct rx_inx rx_backtrack_frame; int cset_size; rx_Bitset cset; struct rx_distinct_future *options; }; /* A superstate is a set of nfa states (RX_SUPERSET) along * with a transition table. Superstates are built on demand and reclaimed * without warning. To protect a superstate from this ghastly fate, * use LOCK_SUPERSTATE. */ struct rx_superstate { int rx_id; /* c.f. the id field of rx_superset */ int locks; /* protection from reclamation */ /* Within a superstate cache, all the superstates are kept in a big * queue. The tail of the queue is the state most likely to be * reclaimed. The *recyclable fields hold the queue position of * this state. */ struct rx_superstate * next_recyclable; struct rx_superstate * prev_recyclable; /* The supernfa edges that exist in the cache and that have * this state as their destination are kept in this list: */ struct rx_distinct_future * transition_refs; /* The list of nfa states corresponding to this superstate: */ struct rx_superset * contents; /* The list of edges in the cache beginning from this state. */ struct rx_super_edge * edges; /* A tail of the recyclable queue is marked as semifree. A semifree * state has no incoming next_char transitions -- any transition * into a semifree state causes a complex dispatch with the side * effect of rescuing the state from its semifree state into a * fully free state at the head of the queue. * * An alternative to this might be to make next_char more expensive, * and to move a state to the head of the recyclable queue whenever * it is entered. That way, popular states would never be recycled. * * But unilaterally making next_char more expensive actually loses. * So, incoming transitions are only made expensive for states near * the tail of the recyclable queue. The more cache contention * there is, the more frequently a state will have to prove itself * and be moved back to the front of the queue. If there is less * contention, then popular states just aggregate in the front of * the queue and stay there. */ int is_semifree; /* This keeps track of the size of the transition table for this * state. There is a half-hearted attempt to support variable sized * superstates. */ int trans_size; /* Indexed by characters... */ struct rx_inx transitions[RX_TAIL_ARRAY]; }; /* A list of distinct futures define the edges that leave from a * given superstate on a given character. c.f. rx_super_edge. */ struct rx_distinct_future { struct rx_distinct_future * next_same_super_edge[2]; struct rx_distinct_future * next_same_dest; struct rx_distinct_future * prev_same_dest; struct rx_superstate * present; /* source state */ struct rx_superstate * future; /* destination state */ struct rx_super_edge * edge; /* The future_frame holds the instruction that should be executed * after all the side effects are done, when it is time to complete * the transition to the next state. * * Normally this is a next_char instruction, but it may be a * cache_miss instruction as well, depending on whether or not * the superstate is in the cache and semifree. * * If this is the only future for a given superstate/char, and * if there are no side effects to be performed, this frame is * not used (directly) at all. Instead, its contents are copied * into the transition table of the starting state of this dist. future * (a sort of goto elimination). */ struct rx_inx future_frame; struct rx_inx side_effects_frame; struct rx_se_list * effects; }; #define rx_lock_superstate(R,S) ((S)->locks++) #define rx_unlock_superstate(R,S) (--(S)->locks) struct rx_cache; #ifdef __STDC__ typedef void (*rx_morecore_fn)(struct rx_cache *); #else typedef void (*rx_morecore_fn)(); #endif struct rx_cache { struct rx_hash_rules superset_hash_rules; struct rx_superstate * lru_superstate; struct rx_superstate * semifree_superstate; struct rx_superset * empty_superset; int superstates; int semifree_superstates; int hits; int misses; int bytes_allowed; int bytes_used; int local_cset_size; void ** instruction_table; struct rx_hash superset_table; }; extern struct rx_cache * rx_default_cache; #ifdef __STDC__ extern void rx_release_superset (struct rx *rx, struct rx_superset *set); extern struct rx_superset * rx_superset_cons (struct rx * rx, struct rx_nfa_state *car, struct rx_superset *cdr); extern struct rx_superset * rx_superstate_eclosure_union (struct rx * rx, struct rx_superset *set, struct rx_nfa_state_set *ecl) ; extern struct rx_superstate * rx_superstate (struct rx *rx, struct rx_superset *set); extern struct rx_inx * rx_handle_cache_miss (struct rx *rx, struct rx_superstate *super, unsigned char chr, void *data) ; #else /* STDC */ extern void rx_release_superset (); extern struct rx_superset * rx_superset_cons (); extern struct rx_superset * rx_superstate_eclosure_union extern struct rx_superstate * rx_superstate (); extern struct rx_inx * rx_handle_cache_miss #endif /* STDC */ #endif /* RXSUPERH */ dist/rx/rxunfa.h100660 765 144 4264 6071416221 12476 0ustar lordusers#ifndef RXUNFAH #define RXUNFAH /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #include "_rx.h" struct rx_unfaniverse { int delay; int delayed; struct rx_hash table; struct rx_cached_rexp * free_queue; }; struct rx_unfa { int refs; struct rexp_node * exp; struct rx * nfa; int cset_size; struct rx_unfaniverse * verse; }; struct rx_cached_rexp { struct rx_unfa unfa; struct rx_cached_rexp * next; struct rx_cached_rexp * prev; struct rx_hash_item * hash_item; }; #ifdef __STDC__ extern struct rx * rx_unfa_rx (struct rx_cached_rexp * cr, struct rexp_node * exp, int cset_size); extern struct rx_unfaniverse * rx_make_unfaniverse (int delay); extern void rx_free_unfaniverse (struct rx_unfaniverse * it); extern struct rx_unfa * rx_unfa (struct rx_unfaniverse * unfaniverse, struct rexp_node * exp, int cset_size); extern void rx_free_unfa (struct rx_unfa * unfa); #else /* STDC */ extern struct rx * rx_unfa_rx (); extern struct rx_unfaniverse * rx_make_unfaniverse (); extern void rx_free_unfaniverse (); extern struct rx_unfa * rx_unfa (); extern void rx_free_unfa (); #endif /* STDC */ #endif /* RXUNFAH */ dist/rx/testcases.h100660 765 144 14261 6070131331 13202 0ustar lordusers {0, "(.*)*\\1", "xx"}, {0, "^", ""}, {0, "$", ""}, {0, "^$", ""}, {0, "^a$", "a"}, {0, "abc", "abc"}, {1, "abc", "xbc"}, {1, "abc", "axc"}, {1, "abc", "abx"}, {0, "abc", "xabcy"}, {0, "abc", "ababc"}, {0, "ab*c", "abc"}, {0, "ab*bc", "abc"}, {0, "ab*bc", "abbc"}, {0, "ab*bc", "abbbbc"}, {0, "ab+bc", "abbc"}, {1, "ab+bc", "abc"}, {1, "ab+bc", "abq"}, {0, "ab+bc", "abbbbc"}, {0, "ab?bc", "abbc"}, {0, "ab?bc", "abc"}, {1, "ab?bc", "abbbbc"}, {0, "ab?c", "abc"}, {0, "^abc$", "abc"}, {1, "^abc$", "abcc"}, {0, "^abc", "abcc"}, {1, "^abc$", "aabc"}, {0, "abc$", "aabc"}, {0, "^", "abc"}, {0, "$", "abc"}, {0, "a.c", "abc"}, {0, "a.c", "axc"}, {0, "a.*c", "axyzc"}, {1, "a.*c", "axyzd"}, {1, "a[bc]d", "abc"}, {0, "a[bc]d", "abd"}, {1, "a[b-d]e", "abd"}, {0, "a[b-d]e", "ace"}, {0, "a[b-d]", "aac"}, {0, "a[-b]", "a-"}, {0, "a[b-]", "a-"}, {1, "a[b-a]", "-"}, {2, "a[]b", "-"}, {2, "a[", "-"}, {0, "a]", "a]"}, {0, "a[]]b", "a]b"}, {0, "a[^bc]d", "aed"}, {1, "a[^bc]d", "abd"}, {0, "a[^-b]c", "adc"}, {1, "a[^-b]c", "a-c"}, {1, "a[^]b]c", "a]c"}, {0, "a[^]b]c", "adc"}, {0, "ab|cd", "abc"}, {0, "ab|cd", "abcd"}, {0, "()ef", "def"}, {0, "()*", "-"}, {1, "*a", "-"}, {0, "^*", "-"}, {0, "$*", "-"}, {1, "(*)b", "-"}, {1, "$b", "b"}, {2, "a\\", "-"}, {0, "a\\(b", "a(b"}, {0, "a\\(*b", "ab"}, {0, "a\\(*b", "a((b"}, {1, "a\\x", "a\\x"}, {1, "abc)", "-"}, {2, "(abc", "-"}, {0, "((a))", "abc"}, {0, "(a)b(c)", "abc"}, {0, "a+b+c", "aabbabc"}, {0, "a**", "-"}, {0, "a*?", "-"}, {0, "(a*)*", "-"}, {0, "(a*)+", "-"}, {0, "(a|)*", "-"}, {0, "(a*|b)*", "-"}, {0, "(a+|b)*", "ab"}, {0, "(a+|b)+", "ab"}, {0, "(a+|b)?", "ab"}, {0, "[^ab]*", "cde"}, {0, "(^)*", "-"}, {0, "(ab|)*", "-"}, {2, ")(", "-"}, {1, "abc", ""}, {1, "abc", ""}, {0, "a*", ""}, {0, "([abc])*d", "abbbcd"}, {0, "([abc])*bcd", "abcd"}, {0, "a|b|c|d|e", "e"}, {0, "(a|b|c|d|e)f", "ef"}, {0, "((a*|b))*", "-"}, {0, "abcd*efg", "abcdefg"}, {0, "ab*", "xabyabbbz"}, {0, "ab*", "xayabbbz"}, {0, "(ab|cd)e", "abcde"}, {0, "[abhgefdc]ij", "hij"}, {1, "^(ab|cd)e", "abcde"}, {0, "(abc|)ef", "abcdef"}, {0, "(a|b)c*d", "abcd"}, {0, "(ab|ab*)bc", "abc"}, {0, "a([bc]*)c*", "abc"}, {0, "a([bc]*)(c*d)", "abcd"}, {0, "a([bc]+)(c*d)", "abcd"}, {0, "a([bc]*)(c+d)", "abcd"}, {0, "a[bcd]*dcdcde", "adcdcde"}, {1, "a[bcd]+dcdcde", "adcdcde"}, {0, "(ab|a)b*c", "abc"}, {0, "((a)(b)c)(d)", "abcd"}, {0, "[A-Za-z_][A-Za-z0-9_]*", "alpha"}, {0, "^a(bc+|b[eh])g|.h$", "abh"}, {0, "(bc+d$|ef*g.|h?i(j|k))", "effgz"}, {0, "(bc+d$|ef*g.|h?i(j|k))", "ij"}, {1, "(bc+d$|ef*g.|h?i(j|k))", "effg"}, {1, "(bc+d$|ef*g.|h?i(j|k))", "bcdd"}, {0, "(bc+d$|ef*g.|h?i(j|k))", "reffgz"}, {1, "((((((((((a))))))))))", "-"}, {0, "(((((((((a)))))))))", "a"}, {1, "multiple words of text", "uh-uh"}, {0, "multiple words", "multiple words, yeah"}, {0, "(.*)c(.*)", "abcde"}, {1, "\\((.*),", "(.*)\\)"}, {1, "[k]", "ab"}, {0, "abcd", "abcd"}, {0, "a(bc)d", "abcd"}, {0, "a[-]?c", "ac"}, {0, "(....).*\\1", "beriberi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Muammar Qaddafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Mo'ammar Gadhafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Muammar Kaddafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Muammar Qadhafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Moammar El Kadhafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Muammar Gadafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Mu'ammar al-Qadafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Moamer El Kazzafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Moamar al-Gaddafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Mu'ammar Al Qathafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Muammar Al Qathafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Mo'ammar el-Gadhafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Moamar El Kadhafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Muammar al-Qadhafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Mu'ammar al-Qadhdhafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Mu'ammar Qadafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Moamar Gaddafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Mu'ammar Qadhdhafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Muammar Khaddafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Muammar al-Khaddafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Mu'amar al-Kadafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Muammar Ghaddafy"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Muammar Ghadafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Muammar Ghaddafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Muamar Kaddafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Muammar Quathafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Muammar Gheddafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Muamar Al-Kaddafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Moammar Khadafy "}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Moammar Qudhafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Mu'ammar al-Qaddafi"}, {0, "M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]", "Mulazim Awwal Mu'ammar Muhammad Abu Minyar al-Qadhafi"}, dist/rx/rxall.h100660 765 144 2425 6071416222 12313 0ustar lordusers/* classes: h_files */ #ifndef RXALLH #define RXALLH /*********************************************************** Copyright 1995 by Tom Lord All Rights Reserved Permission to use, copy, modify, and distribute this software and its documentation for any purpose and without fee is hereby granted, provided that the above copyright notice appear in all copies and that both that copyright notice and this permission notice appear in supporting documentation, and that the name of the copyright holder not be used in advertising or publicity pertaining to distribution of the software without specific, written prior permission. Tom Lord DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL TOM LORD BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. ******************************************************************/ #if 0 #include #include "malloc.h" #endif #ifdef __STDC__ #else /* STDC */ #endif /* STDC */ #endif /* RXALLH */ dist/rx/configure.in100660 765 144 552 6071412230 13303 0ustar lordusersdnl Process this file with autoconf to produce a configure script. AC_INIT(rx.c) AC_PROG_CC AC_PROG_CPP AC_PROG_RANLIB AC_PROG_INSTALL AC_STDC_HEADERS opt_objs= opt_cflags= if test -d ../libguile ; then opt_objs=guilergx.o opt_cflags="-I$srcdir/../libguile -I../libguile -DRX_WANT_RX_DEFS" fi AC_SUBST(opt_objs) AC_SUBST(opt_cflags) AC_OUTPUT(Makefile) dist/rx/Makefile.in100660 765 144 12626 6071535141 13113 0ustar lordusers# Makefile for librx # Copyright (C) 1994 Free Software Foundation, Inc. # # This file is part of GNU Rx # # GNU Rx is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by # the Free Software Foundation; either version 2, or (at your option) # any later version. # # GNU Rx is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with GNU Rx; see the file COPYING.LIB. If not, write to # the Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. SHELL = /bin/sh srcdir = @srcdir@ VPATH = @srcdir@ prefix = @prefix@ exec_prefix = $(prefix) bindir = $(exec_prefix)/bin libdir = $(exec_prefix)/lib includedir = $(prefix)/include infodir = $(prefix)/info #### Start of system configuration section. #### CC = @CC@ INSTALL = @INSTALL@ DEFS = @DEFS@ CFLAGS = -g @opt_cflags@ -I. -I$(srcdir) AR = ar AR_FLAGS = rc RANLIB = @RANLIB@ opt_objs=@opt_objs@ source= hashrexp.c runtests.c rx.c rxanal.c rxbasic.c rxbitset.c rxcset.c rxdbug.c rxgnucomp.c rxhash.c rxnfa.c rxnode.c rxposix.c rxsimp.c rxspencer.c rxstr.c rxsuper.c rxunfa.c headers=_rx.h rx.h rxanal.h rxbasic.h rxbitset.h rxcontext.h rxcset.h rxgnucomp.h rxhash.h rxnfa.h rxnode.h inst-rxposix.h rxposix.h rxproto.h rxsimp.h rxspencer.h rxstr.h rxsuper.h rxunfa.h testcases.h rxall.h ancillery=configure.in Makefile.in configure ../doc/rx.texi ../doc/rx.info \ COPYING.LIB ChangeLog PLUGIN TESTS TESTS2C.sed distfiles=$(source) $(headers) $(ancillery) libobjs= hashrexp.o rx.o rxanal.o rxbasic.o rxbitset.o rxcset.o rxdbug.o rxgnucomp.o rxhash.o rxnfa.o rxnode.o rxposix.o rxsimp.o rxspencer.o rxstr.o rxsuper.o rxunfa.o .SUFFIXES: .SUFFIXES: .o .c .h .ps .dvi .info .texinfo .scm .cd .x .c.x: $(CC) $(CFLAGS) -DGSCM_MAGIC_SNARFER -E $< | grep "^%%%" | sed -e "s/^%%%//" > $@ ; \ .c.o: $(CC) -c $(CFLAGS) $(DEFS) -I$(srcdir) $< all: librx.a install: all test -d $(libdir) || mkdir $(libdir) test -d $(includedir) || mkdir $(includedir) $(INSTALL) librx.a $(libdir)/librx.a $(RANLIB) $(libdir)/librx.a $(INSTALL) $(srcdir)/inst-rxposix.h $(includedir)/rxposix.h uninstall: rm $(libdir)/librx.a rm $(includedir)/rx.h clean: -rm -f $(libobjs) librx.a distclean: clean -rm Makefile config.status mostlyclean: clean realclean: distclean TAGS: etags $(source) info: install-info: clean-info: dvi: check: SUBDIR=. manifest: for file in $(distfiles); do echo $(SUBDIR)/$$file; done dist: $(distfiles) echo rx-`sed -e '/version_string/!d' -e 's/[^0-9.]*\([0-9.]*\).*/\1/' -e q rx.c` > .fname rm -rf `cat .fname` mkdir `cat .fname` ln $(distfiles) `cat .fname` tar chzf `cat .fname`.tar.gz `cat .fname` rm -rf `cat .fname` .fname librx.a: $(libobjs) rm -f librx.a $(AR) $(AR_FLAGS) librx.a $(libobjs) $(RANLIB) librx.a depends: gcc -MM $(source) $(srcdir)/testcases.h: TESTS TESTS2C.sed sed -f $(srcdir)/TESTS2C.sed $(srcdir)/TESTS > $(srcdir)/testcases.h runtests: runtests.o librx.a rxdbug.o $(CC) -g -o runtests runtests.o rxdbug.o librx.a ### hashrexp.o : hashrexp.c rxall.h /usr/include/malloc.h rxnode.h rxbitset.h rxcset.h \ rxhash.h runtests.o : runtests.c rxall.h /usr/include/malloc.h rxposix.h rxspencer.h \ rxproto.h rxnode.h rxbitset.h rxcset.h rxunfa.h _rx.h rxhash.h rxanal.h rxsuper.h \ rxnfa.h rxcontext.h testcases.h rx.o : rx.c rxall.h /usr/include/malloc.h rxhash.h rxbitset.h rxnfa.h _rx.h \ rxcset.h rxnode.h rxanal.o : rxanal.c rxall.h /usr/include/malloc.h rxanal.h rxcset.h rxbitset.h \ rxnode.h rxsuper.h rxnfa.h _rx.h rxhash.h rxbasic.o : rxbasic.c rxall.h /usr/include/malloc.h rxbasic.h rxcontext.h rxnode.h \ rxbitset.h rxcset.h rxspencer.h rxproto.h rxunfa.h _rx.h rxhash.h rxanal.h rxsuper.h \ rxnfa.h rxstr.h rxbitset.o : rxbitset.c rxall.h /usr/include/malloc.h rxbitset.h rxcset.o : rxcset.c rxall.h /usr/include/malloc.h rxcset.h rxbitset.h rxdbug.o : rxdbug.c rxall.h /usr/include/malloc.h rxgnucomp.h rxcset.h rxbitset.h \ rxnode.h rxnfa.h _rx.h rxhash.h rxgnucomp.o : rxgnucomp.c rxall.h /usr/include/malloc.h rxgnucomp.h rxcset.h \ rxbitset.h rxnode.h rxhash.o : rxhash.c rxall.h /usr/include/malloc.h rxhash.h rxbitset.h rxnfa.o : rxnfa.c rxall.h /usr/include/malloc.h rxnfa.h _rx.h rxhash.h rxbitset.h \ rxcset.h rxnode.h rxnode.o : rxnode.c rxall.h /usr/include/malloc.h rxnode.h rxbitset.h rxcset.h rxposix.o : rxposix.c rxall.h /usr/include/malloc.h rxposix.h rxspencer.h rxproto.h \ rxnode.h rxbitset.h rxcset.h rxunfa.h _rx.h rxhash.h rxanal.h rxsuper.h rxnfa.h \ rxcontext.h rxgnucomp.h rxbasic.h rxsimp.h rxsimp.o : rxsimp.c rxall.h /usr/include/malloc.h rxsimp.h rxcset.h rxbitset.h \ rxnode.h rxspencer.o : rxspencer.c rxall.h /usr/include/malloc.h rxspencer.h rxproto.h \ rxnode.h rxbitset.h rxcset.h rxunfa.h _rx.h rxhash.h rxanal.h rxsuper.h rxnfa.h rxstr.o : rxstr.c rxall.h /usr/include/malloc.h rxstr.h rxspencer.h rxproto.h \ rxnode.h rxbitset.h rxcset.h rxunfa.h _rx.h rxhash.h rxanal.h rxsuper.h rxnfa.h \ rxcontext.h rxsuper.o : rxsuper.c rxall.h /usr/include/malloc.h rxsuper.h rxnfa.h _rx.h \ rxhash.h rxbitset.h rxcset.h rxnode.h rxunfa.o : rxunfa.c rxall.h /usr/include/malloc.h rx.h rxhash.h rxbitset.h rxunfa.h \ _rx.h rxcset.h rxnfa.h rxnode.h dist/rx/configure100770 765 144 70357 6071412311 12752 0ustar lordusers#! /bin/sh # Guess values for system-dependent variables and create Makefiles. # Generated automatically using autoconf version 2.3 # Copyright (C) 1992, 1993, 1994 Free Software Foundation, Inc. # # This configure script is free software; the Free Software Foundation # gives unlimited permission to copy, distribute and modify it. # Defaults: ac_help= ac_default_prefix=/usr/local # Any additions from configure.in: # Initialize some variables set by options. # The variables have the same names as the options, with # dashes changed to underlines. build=NONE cache_file=./config.cache exec_prefix=NONE host=NONE no_create= nonopt=NONE no_recursion= prefix=NONE program_prefix=NONE program_suffix=NONE program_transform_name=s,x,x, silent= site= srcdir= target=NONE verbose= x_includes=NONE x_libraries=NONE # Initialize some other variables. subdirs= ac_prev= for ac_option do # If the previous option needs an argument, assign it. if test -n "$ac_prev"; then eval "$ac_prev=\$ac_option" ac_prev= continue fi case "$ac_option" in -*=*) ac_optarg=`echo "$ac_option" | sed 's/[-_a-zA-Z0-9]*=//'` ;; *) ac_optarg= ;; esac # Accept the important Cygnus configure options, so we can diagnose typos. case "$ac_option" in -build | --build | --buil | --bui | --bu | --b) ac_prev=build ;; -build=* | --build=* | --buil=* | --bui=* | --bu=* | --b=*) build="$ac_optarg" ;; -cache-file | --cache-file | --cache-fil | --cache-fi \ | --cache-f | --cache- | --cache | --cach | --cac | --ca | --c) ac_prev=cache_file ;; -cache-file=* | --cache-file=* | --cache-fil=* | --cache-fi=* \ | --cache-f=* | --cache-=* | --cache=* | --cach=* | --cac=* | --ca=* | --c=*) cache_file="$ac_optarg" ;; -disable-* | --disable-*) ac_feature=`echo $ac_option|sed -e 's/-*disable-//'` # Reject names that are not valid shell variable names. if test -n "`echo $ac_feature| sed 's/[-a-zA-Z0-9_]//g'`"; then { echo "configure: error: $ac_feature: invalid feature name" 1>&2; exit 1; } fi ac_feature=`echo $ac_feature| sed 's/-/_/g'` eval "enable_${ac_feature}=no" ;; -enable-* | --enable-*) ac_feature=`echo $ac_option|sed -e 's/-*enable-//' -e 's/=.*//'` # Reject names that are not valid shell variable names. if test -n "`echo $ac_feature| sed 's/[-_a-zA-Z0-9]//g'`"; then { echo "configure: error: $ac_feature: invalid feature name" 1>&2; exit 1; } fi ac_feature=`echo $ac_feature| sed 's/-/_/g'` case "$ac_option" in *=*) ;; *) ac_optarg=yes ;; esac eval "enable_${ac_feature}='$ac_optarg'" ;; -exec-prefix | --exec_prefix | --exec-prefix | --exec-prefi \ | --exec-pref | --exec-pre | --exec-pr | --exec-p | --exec- \ | --exec | --exe | --ex) ac_prev=exec_prefix ;; -exec-prefix=* | --exec_prefix=* | --exec-prefix=* | --exec-prefi=* \ | --exec-pref=* | --exec-pre=* | --exec-pr=* | --exec-p=* | --exec-=* \ | --exec=* | --exe=* | --ex=*) exec_prefix="$ac_optarg" ;; -gas | --gas | --ga | --g) # Obsolete; use --with-gas. with_gas=yes ;; -help | --help | --hel | --he) # Omit some internal or obsolete options to make the list less imposing. # This message is too long to be a string in the A/UX 3.1 sh. cat << EOF Usage: configure [options] [host] Options: [defaults in brackets after descriptions] Configuration: --cache-file=FILE cache test results in FILE --help print this message --no-create do not create output files --quiet, --silent do not print \`checking...' messages --version print the version of autoconf that created configure Directory and file names: --prefix=PREFIX install architecture-independent files in PREFIX [$ac_default_prefix] --exec-prefix=PREFIX install architecture-dependent files in PREFIX [same as prefix] --srcdir=DIR find the sources in DIR [configure dir or ..] --program-prefix=PREFIX prepend PREFIX to installed program names --program-suffix=SUFFIX append SUFFIX to installed program names --program-transform-name=PROGRAM run sed PROGRAM on installed program names Host type: --build=BUILD configure for building on BUILD [BUILD=HOST] --host=HOST configure for HOST [guessed] --target=TARGET configure for TARGET [TARGET=HOST] Features and packages: --disable-FEATURE do not include FEATURE (same as --enable-FEATURE=no) --enable-FEATURE[=ARG] include FEATURE [ARG=yes] --with-PACKAGE[=ARG] use PACKAGE [ARG=yes] --without-PACKAGE do not use PACKAGE (same as --with-PACKAGE=no) --x-includes=DIR X include files are in DIR --x-libraries=DIR X library files are in DIR --enable and --with options recognized:$ac_help EOF exit 0 ;; -host | --host | --hos | --ho) ac_prev=host ;; -host=* | --host=* | --hos=* | --ho=*) host="$ac_optarg" ;; -nfp | --nfp | --nf) # Obsolete; use --without-fp. with_fp=no ;; -no-create | --no-create | --no-creat | --no-crea | --no-cre \ | --no-cr | --no-c) no_create=yes ;; -no-recursion | --no-recursion | --no-recursio | --no-recursi \ | --no-recurs | --no-recur | --no-recu | --no-rec | --no-re | --no-r) no_recursion=yes ;; -prefix | --prefix | --prefi | --pref | --pre | --pr | --p) ac_prev=prefix ;; -prefix=* | --prefix=* | --prefi=* | --pref=* | --pre=* | --pr=* | --p=*) prefix="$ac_optarg" ;; -program-prefix | --program-prefix | --program-prefi | --program-pref \ | --program-pre | --program-pr | --program-p) ac_prev=program_prefix ;; -program-prefix=* | --program-prefix=* | --program-prefi=* \ | --program-pref=* | --program-pre=* | --program-pr=* | --program-p=*) program_prefix="$ac_optarg" ;; -program-suffix | --program-suffix | --program-suffi | --program-suff \ | --program-suf | --program-su | --program-s) ac_prev=program_suffix ;; -program-suffix=* | --program-suffix=* | --program-suffi=* \ | --program-suff=* | --program-suf=* | --program-su=* | --program-s=*) program_suffix="$ac_optarg" ;; -program-transform-name | --program-transform-name \ | --program-transform-nam | --program-transform-na \ | --program-transform-n | --program-transform- \ | --program-transform | --program-transfor \ | --program-transfo | --program-transf \ | --program-trans | --program-tran \ | --progr-tra | --program-tr | --program-t) ac_prev=program_transform_name ;; -program-transform-name=* | --program-transform-name=* \ | --program-transform-nam=* | --program-transform-na=* \ | --program-transform-n=* | --program-transform-=* \ | --program-transform=* | --program-transfor=* \ | --program-transfo=* | --program-transf=* \ | --program-trans=* | --program-tran=* \ | --progr-tra=* | --program-tr=* | --program-t=*) program_transform_name="$ac_optarg" ;; -q | -quiet | --quiet | --quie | --qui | --qu | --q \ | -silent | --silent | --silen | --sile | --sil) silent=yes ;; -site | --site | --sit) ac_prev=site ;; -site=* | --site=* | --sit=*) site="$ac_optarg" ;; -srcdir | --srcdir | --srcdi | --srcd | --src | --sr) ac_prev=srcdir ;; -srcdir=* | --srcdir=* | --srcdi=* | --srcd=* | --src=* | --sr=*) srcdir="$ac_optarg" ;; -target | --target | --targe | --targ | --tar | --ta | --t) ac_prev=target ;; -target=* | --target=* | --targe=* | --targ=* | --tar=* | --ta=* | --t=*) target="$ac_optarg" ;; -v | -verbose | --verbose | --verbos | --verbo | --verb) verbose=yes ;; -version | --version | --versio | --versi | --vers) echo "configure generated by autoconf version 2.3" exit 0 ;; -with-* | --with-*) ac_package=`echo $ac_option|sed -e 's/-*with-//' -e 's/=.*//'` # Reject names that are not valid shell variable names. if test -n "`echo $ac_package| sed 's/[-_a-zA-Z0-9]//g'`"; then { echo "configure: error: $ac_package: invalid package name" 1>&2; exit 1; } fi ac_package=`echo $ac_package| sed 's/-/_/g'` case "$ac_option" in *=*) ;; *) ac_optarg=yes ;; esac eval "with_${ac_package}='$ac_optarg'" ;; -without-* | --without-*) ac_package=`echo $ac_option|sed -e 's/-*without-//'` # Reject names that are not valid shell variable names. if test -n "`echo $ac_package| sed 's/[-a-zA-Z0-9_]//g'`"; then { echo "configure: error: $ac_package: invalid package name" 1>&2; exit 1; } fi ac_package=`echo $ac_package| sed 's/-/_/g'` eval "with_${ac_package}=no" ;; --x) # Obsolete; use --with-x. with_x=yes ;; -x-includes | --x-includes | --x-include | --x-includ | --x-inclu \ | --x-incl | --x-inc | --x-in | --x-i) ac_prev=x_includes ;; -x-includes=* | --x-includes=* | --x-include=* | --x-includ=* | --x-inclu=* \ | --x-incl=* | --x-inc=* | --x-in=* | --x-i=*) x_includes="$ac_optarg" ;; -x-libraries | --x-libraries | --x-librarie | --x-librari \ | --x-librar | --x-libra | --x-libr | --x-lib | --x-li | --x-l) ac_prev=x_libraries ;; -x-libraries=* | --x-libraries=* | --x-librarie=* | --x-librari=* \ | --x-librar=* | --x-libra=* | --x-libr=* | --x-lib=* | --x-li=* | --x-l=*) x_libraries="$ac_optarg" ;; -*) { echo "configure: error: $ac_option: invalid option; use --help to show usage" 1>&2; exit 1; } ;; *) if test -n "`echo $ac_option| sed 's/[-a-z0-9.]//g'`"; then echo "configure: warning: $ac_option: invalid host type" 1>&2 fi if test "x$nonopt" != xNONE; then { echo "configure: error: can only configure for one host and one target at a time" 1>&2; exit 1; } fi nonopt="$ac_option" ;; esac done if test -n "$ac_prev"; then { echo "configure: error: missing argument to --`echo $ac_prev | sed 's/_/-/g'`" 1>&2; exit 1; } fi trap 'rm -fr conftest* confdefs* core core.* *.core $ac_clean_files; exit 1' 1 2 15 # File descriptor usage: # 0 standard input # 1 file creation # 2 errors and warnings # 3 some systems may open it to /dev/tty # 4 used on the Kubota Titan # 6 checking for... messages and results # 5 compiler messages saved in config.log if test "$silent" = yes; then exec 6>/dev/null else exec 6>&1 fi exec 5>./config.log echo "\ This file contains any messages produced by compilers while running configure, to aid debugging if configure makes a mistake. " 1>&5 # Strip out --no-create and --no-recursion so they do not pile up. # Also quote any args containing shell metacharacters. ac_configure_args= for ac_arg do case "$ac_arg" in -no-create | --no-create | --no-creat | --no-crea | --no-cre \ | --no-cr | --no-c) ;; -no-recursion | --no-recursion | --no-recursio | --no-recursi \ | --no-recurs | --no-recur | --no-recu | --no-rec | --no-re | --no-r) ;; *" "*|*" "*|*[\[\]\~\#\$\^\&\*\(\)\{\}\\\|\;\<\>\?]*) ac_configure_args="$ac_configure_args '$ac_arg'" ;; *) ac_configure_args="$ac_configure_args $ac_arg" ;; esac done # NLS nuisances. # Only set LANG and LC_ALL to C if already set. # These must not be set unconditionally because not all systems understand # e.g. LANG=C (notably SCO). if test "${LC_ALL+set}" = set; then LC_ALL=C; export LC_ALL; fi if test "${LANG+set}" = set; then LANG=C; export LANG; fi # confdefs.h avoids OS command line length limits that DEFS can exceed. rm -rf conftest* confdefs.h # AIX cpp loses on an empty file, so make sure it contains at least a newline. echo > confdefs.h # A filename unique to this package, relative to the directory that # configure is in, which we can look for to find out if srcdir is correct. ac_unique_file=rx.c # Find the source files, if location was not specified. if test -z "$srcdir"; then ac_srcdir_defaulted=yes # Try the directory containing this script, then its parent. ac_prog=$0 ac_confdir=`echo $ac_prog|sed 's%/[^/][^/]*$%%'` test "x$ac_confdir" = "x$ac_prog" && ac_confdir=. srcdir=$ac_confdir if test ! -r $srcdir/$ac_unique_file; then srcdir=.. fi else ac_srcdir_defaulted=no fi if test ! -r $srcdir/$ac_unique_file; then if test "$ac_srcdir_defaulted" = yes; then { echo "configure: error: can not find sources in $ac_confdir or .." 1>&2; exit 1; } else { echo "configure: error: can not find sources in $srcdir" 1>&2; exit 1; } fi fi srcdir=`echo "${srcdir}" | sed 's%\([^/]\)/*$%\1%'` # Prefer explicitly selected file to automatically selected ones. if test -z "$CONFIG_SITE"; then if test "x$prefix" != xNONE; then CONFIG_SITE="$prefix/share/config.site $prefix/etc/config.site" else CONFIG_SITE="$ac_default_prefix/share/config.site $ac_default_prefix/etc/config.site" fi fi for ac_site_file in $CONFIG_SITE; do if test -r "$ac_site_file"; then echo "loading site script $ac_site_file" . "$ac_site_file" fi done if test -r "$cache_file"; then echo "loading cache $cache_file" . $cache_file else echo "creating cache $cache_file" > $cache_file fi ac_ext=c # CFLAGS is not in ac_cpp because -g, -O, etc. are not valid cpp options. ac_cpp='$CPP $CPPFLAGS' ac_compile='${CC-cc} -c $CFLAGS $CPPFLAGS conftest.$ac_ext 1>&5 2>&5' ac_link='${CC-cc} -o conftest $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS 1>&5 2>&5' if (echo "testing\c"; echo 1,2,3) | grep c >/dev/null; then # Stardent Vistra SVR4 grep lacks -e, says ghazi@caip.rutgers.edu. if (echo -n testing; echo 1,2,3) | sed s/-n/xn/ | grep xn >/dev/null; then ac_n= ac_c=' ' ac_t=' ' else ac_n=-n ac_c= ac_t= fi else ac_n= ac_c='\c' ac_t= fi # Extract the first word of "gcc", so it can be a program name with args. set dummy gcc; ac_word=$2 echo $ac_n "checking for $ac_word""... $ac_c" 1>&6 if eval "test \"`echo '$''{'ac_cv_prog_CC'+set}'`\" = set"; then echo $ac_n "(cached) $ac_c" 1>&6 else if test -n "$CC"; then ac_cv_prog_CC="$CC" # Let the user override the test. else IFS="${IFS= }"; ac_save_ifs="$IFS"; IFS="${IFS}:" for ac_dir in $PATH; do test -z "$ac_dir" && ac_dir=. if test -f $ac_dir/$ac_word; then ac_cv_prog_CC="gcc" break fi done IFS="$ac_save_ifs" test -z "$ac_cv_prog_CC" && ac_cv_prog_CC="cc" fi fi CC="$ac_cv_prog_CC" if test -n "$CC"; then echo "$ac_t""$CC" 1>&6 else echo "$ac_t""no" 1>&6 fi echo $ac_n "checking whether we are using GNU C""... $ac_c" 1>&6 if eval "test \"`echo '$''{'ac_cv_prog_gcc'+set}'`\" = set"; then echo $ac_n "(cached) $ac_c" 1>&6 else cat > conftest.c <&5 | egrep yes >/dev/null 2>&1; then ac_cv_prog_gcc=yes else ac_cv_prog_gcc=no fi fi echo "$ac_t""$ac_cv_prog_gcc" 1>&6 if test $ac_cv_prog_gcc = yes; then GCC=yes if test "${CFLAGS+set}" != set; then echo $ac_n "checking whether ${CC-cc} accepts -g""... $ac_c" 1>&6 if eval "test \"`echo '$''{'ac_cv_prog_gcc_g'+set}'`\" = set"; then echo $ac_n "(cached) $ac_c" 1>&6 else echo 'void f(){}' > conftest.c if test -z "`${CC-cc} -g -c conftest.c 2>&1`"; then ac_cv_prog_gcc_g=yes else ac_cv_prog_gcc_g=no fi rm -f conftest* fi echo "$ac_t""$ac_cv_prog_gcc_g" 1>&6 if test $ac_cv_prog_gcc_g = yes; then CFLAGS="-g -O" else CFLAGS="-O" fi fi else GCC= test "${CFLAGS+set}" = set || CFLAGS="-g" fi echo $ac_n "checking how to run the C preprocessor""... $ac_c" 1>&6 # On Suns, sometimes $CPP names a directory. if test -n "$CPP" && test -d "$CPP"; then CPP= fi if test -z "$CPP"; then if eval "test \"`echo '$''{'ac_cv_prog_CPP'+set}'`\" = set"; then echo $ac_n "(cached) $ac_c" 1>&6 else # This must be in double quotes, not single quotes, because CPP may get # substituted into the Makefile and "${CC-cc}" will confuse make. CPP="${CC-cc} -E" # On the NeXT, cc -E runs the code through the compiler's parser, # not just through cpp. cat > conftest.$ac_ext < Syntax Error EOF eval "$ac_cpp conftest.$ac_ext >/dev/null 2>conftest.out" ac_err=`grep -v '^ *+' conftest.out` if test -z "$ac_err"; then : else echo "$ac_err" >&5 rm -rf conftest* CPP="${CC-cc} -E -traditional-cpp" cat > conftest.$ac_ext < Syntax Error EOF eval "$ac_cpp conftest.$ac_ext >/dev/null 2>conftest.out" ac_err=`grep -v '^ *+' conftest.out` if test -z "$ac_err"; then : else echo "$ac_err" >&5 rm -rf conftest* CPP=/lib/cpp fi rm -f conftest* fi rm -f conftest* ac_cv_prog_CPP="$CPP" fi fi CPP="$ac_cv_prog_CPP" echo "$ac_t""$CPP" 1>&6 # Extract the first word of "ranlib", so it can be a program name with args. set dummy ranlib; ac_word=$2 echo $ac_n "checking for $ac_word""... $ac_c" 1>&6 if eval "test \"`echo '$''{'ac_cv_prog_RANLIB'+set}'`\" = set"; then echo $ac_n "(cached) $ac_c" 1>&6 else if test -n "$RANLIB"; then ac_cv_prog_RANLIB="$RANLIB" # Let the user override the test. else IFS="${IFS= }"; ac_save_ifs="$IFS"; IFS="${IFS}:" for ac_dir in $PATH; do test -z "$ac_dir" && ac_dir=. if test -f $ac_dir/$ac_word; then ac_cv_prog_RANLIB="ranlib" break fi done IFS="$ac_save_ifs" test -z "$ac_cv_prog_RANLIB" && ac_cv_prog_RANLIB=":" fi fi RANLIB="$ac_cv_prog_RANLIB" if test -n "$RANLIB"; then echo "$ac_t""$RANLIB" 1>&6 else echo "$ac_t""no" 1>&6 fi ac_aux_dir= for ac_dir in $srcdir $srcdir/.. $srcdir/../..; do if test -f $ac_dir/install-sh; then ac_aux_dir=$ac_dir ac_install_sh="$ac_aux_dir/install-sh -c" break elif test -f $ac_dir/install.sh; then ac_aux_dir=$ac_dir ac_install_sh="$ac_aux_dir/install.sh -c" break fi done if test -z "$ac_aux_dir"; then { echo "configure: error: can not find install-sh or install.sh in $srcdir $srcdir/.. $srcdir/../.." 1>&2; exit 1; } fi ac_config_guess=$ac_aux_dir/config.guess ac_config_sub=$ac_aux_dir/config.sub ac_configure=$ac_aux_dir/configure # This should be Cygnus configure. # Find a good install program. We prefer a C program (faster), # so one script is as good as another. But avoid the broken or # incompatible versions: # SysV /etc/install, /usr/sbin/install # SunOS /usr/etc/install # IRIX /sbin/install # AIX /bin/install # AFS /usr/afsws/bin/install, which mishandles nonexistent args # SVR4 /usr/ucb/install, which tries to use the nonexistent group "staff" # ./install, which can be erroneously created by make from ./install.sh. echo $ac_n "checking for a BSD compatible install""... $ac_c" 1>&6 if test -z "$INSTALL"; then if eval "test \"`echo '$''{'ac_cv_path_install'+set}'`\" = set"; then echo $ac_n "(cached) $ac_c" 1>&6 else IFS="${IFS= }"; ac_save_ifs="$IFS"; IFS="${IFS}:" for ac_dir in $PATH; do # Account for people who put trailing slashes in PATH elements. case "$ac_dir/" in /|./|.//|/etc/*|/usr/sbin/*|/usr/etc/*|/sbin/*|/usr/afsws/bin/*|/usr/ucb/*) ;; *) # OSF1 and SCO ODT 3.0 have their own names for install. for ac_prog in ginstall installbsd scoinst install; do if test -f $ac_dir/$ac_prog; then if test $ac_prog = install && grep dspmsg $ac_dir/$ac_prog >/dev/null 2>&1; then # AIX install. It has an incompatible calling convention. # OSF/1 installbsd also uses dspmsg, but is usable. : else ac_cv_path_install="$ac_dir/$ac_prog -c" break 2 fi fi done ;; esac done IFS="$ac_save_ifs" # As a last resort, use the slow shell script. test -z "$ac_cv_path_install" && ac_cv_path_install="$ac_install_sh" fi INSTALL="$ac_cv_path_install" fi echo "$ac_t""$INSTALL" 1>&6 # Use test -z because SunOS4 sh mishandles braces in ${var-val}. # It thinks the first close brace ends the variable substitution. test -z "$INSTALL_PROGRAM" && INSTALL_PROGRAM='${INSTALL}' test -z "$INSTALL_DATA" && INSTALL_DATA='${INSTALL} -m 644' # If we cannot run a trivial program, we must be cross compiling. echo $ac_n "checking whether cross-compiling""... $ac_c" 1>&6 if eval "test \"`echo '$''{'ac_cv_c_cross'+set}'`\" = set"; then echo $ac_n "(cached) $ac_c" 1>&6 else if test "$cross_compiling" = yes; then ac_cv_c_cross=yes else cat > conftest.$ac_ext </dev/null; then ac_cv_c_cross=no else ac_cv_c_cross=yes fi fi rm -fr conftest* fi cross_compiling=$ac_cv_c_cross echo "$ac_t""$ac_cv_c_cross" 1>&6 echo $ac_n "checking for ANSI C header files""... $ac_c" 1>&6 if eval "test \"`echo '$''{'ac_cv_header_stdc'+set}'`\" = set"; then echo $ac_n "(cached) $ac_c" 1>&6 else cat > conftest.$ac_ext < #include #include #include EOF eval "$ac_cpp conftest.$ac_ext >/dev/null 2>conftest.out" ac_err=`grep -v '^ *+' conftest.out` if test -z "$ac_err"; then rm -rf conftest* ac_cv_header_stdc=yes else echo "$ac_err" >&5 rm -rf conftest* ac_cv_header_stdc=no fi rm -f conftest* if test $ac_cv_header_stdc = yes; then # SunOS 4.x string.h does not declare mem*, contrary to ANSI. cat > conftest.$ac_ext < EOF if (eval "$ac_cpp conftest.$ac_ext") 2>&5 | egrep "memchr" >/dev/null 2>&1; then : else rm -rf conftest* ac_cv_header_stdc=no fi rm -f conftest* fi if test $ac_cv_header_stdc = yes; then # ISC 2.0.2 stdlib.h does not declare free, contrary to ANSI. cat > conftest.$ac_ext < EOF if (eval "$ac_cpp conftest.$ac_ext") 2>&5 | egrep "free" >/dev/null 2>&1; then : else rm -rf conftest* ac_cv_header_stdc=no fi rm -f conftest* fi if test $ac_cv_header_stdc = yes; then # /bin/cc in Irix-4.0.5 gets non-ANSI ctype macros unless using -ansi. if test "$cross_compiling" = yes; then ac_cv_header_stdc=no else cat > conftest.$ac_ext < #define ISLOWER(c) ('a' <= (c) && (c) <= 'z') #define TOUPPER(c) (ISLOWER(c) ? 'A' + ((c) - 'a') : (c)) #define XOR(e, f) (((e) && !(f)) || (!(e) && (f))) int main () { int i; for (i = 0; i < 256; i++) if (XOR (islower (i), ISLOWER (i)) || toupper (i) != TOUPPER (i)) exit(2); exit (0); } EOF eval $ac_link if test -s conftest && (./conftest; exit) 2>/dev/null; then : else ac_cv_header_stdc=no fi fi rm -fr conftest* fi fi echo "$ac_t""$ac_cv_header_stdc" 1>&6 if test $ac_cv_header_stdc = yes; then cat >> confdefs.h <<\EOF #define STDC_HEADERS 1 EOF fi opt_objs= opt_cflags= if test -d ../libguile ; then opt_objs=guilergx.o opt_cflags="-I$srcdir/../libguile -I../libguile -DRX_WANT_RX_DEFS" fi trap '' 1 2 15 cat > confcache <<\EOF # This file is a shell script that caches the results of configure # tests run on this system so they can be shared between configure # scripts and configure runs. It is not useful on other systems. # If it contains results you don't want to keep, you may remove or edit it. # # By default, configure uses ./config.cache as the cache file, # creating it if it does not exist already. You can give configure # the --cache-file=FILE option to use a different cache file; that is # what configure does when it calls configure scripts in # subdirectories, so they share the cache. # Giving --cache-file=/dev/null disables caching, for debugging configure. # config.status only pays attention to the cache file if you give it the # --recheck option to rerun configure. # EOF # Ultrix sh set writes to stderr and can't be redirected directly, # and sets the high bit in the cache file unless we assign to the vars. (set) 2>&1 | sed -n "s/^\([a-zA-Z0-9_]*_cv_[a-zA-Z0-9_]*\)=\(.*\)/\1=\${\1='\2'}/p" \ >> confcache if cmp -s $cache_file confcache; then : else if test -w $cache_file; then echo "updating cache $cache_file" cat confcache > $cache_file else echo "not updating unwritable cache $cache_file" fi fi rm -f confcache trap 'rm -fr conftest* confdefs* core core.* *.core $ac_clean_files; exit 1' 1 2 15 test "x$prefix" = xNONE && prefix=$ac_default_prefix # Let make expand exec_prefix. test "x$exec_prefix" = xNONE && exec_prefix='${prefix}' # Any assignment to VPATH causes Sun make to only execute # the first set of double-colon rules, so remove it if not needed. # If there is a colon in the path, we need to keep it. if test "x$srcdir" = x.; then ac_vpsub='/^[ ]*VPATH[ ]*=[^:]*$/d' fi trap 'rm -f $CONFIG_STATUS conftest*; exit 1' 1 2 15 # Transform confdefs.h into DEFS. # Protect against shell expansion while executing Makefile rules. # Protect against Makefile macro expansion. cat > conftest.defs <<\EOF s%#define \([A-Za-z_][A-Za-z0-9_]*\) \(.*\)%-D\1=\2%g s%[ `~#$^&*(){}\\|;'"<>?]%\\&%g s%\[%\\&%g s%\]%\\&%g s%\$%$$%g EOF DEFS=`sed -f conftest.defs confdefs.h | tr '\012' ' '` rm -f conftest.defs # Without the "./", some shells look in PATH for config.status. : ${CONFIG_STATUS=./config.status} echo creating $CONFIG_STATUS rm -f $CONFIG_STATUS cat > $CONFIG_STATUS </dev/null | sed 1q`: # # $0 $ac_configure_args # # Compiler output produced by configure, useful for debugging # configure, is in ./config.log if it exists. ac_cs_usage="Usage: $CONFIG_STATUS [--recheck] [--version] [--help]" for ac_option do case "\$ac_option" in -recheck | --recheck | --rechec | --reche | --rech | --rec | --re | --r) echo "running \${CONFIG_SHELL-/bin/sh} $0 $ac_configure_args --no-create --no-recursion" exec \${CONFIG_SHELL-/bin/sh} $0 $ac_configure_args --no-create --no-recursion ;; -version | --version | --versio | --versi | --vers | --ver | --ve | --v) echo "$CONFIG_STATUS generated by autoconf version 2.3" exit 0 ;; -help | --help | --hel | --he | --h) echo "\$ac_cs_usage"; exit 0 ;; *) echo "\$ac_cs_usage"; exit 1 ;; esac done ac_given_srcdir=$srcdir ac_given_INSTALL="$INSTALL" trap 'rm -fr `echo "Makefile" | sed "s/:[^ ]*//g"` conftest*; exit 1' 1 2 15 # Protect against being on the right side of a sed subst in config.status. sed 's/%@/@@/; s/@%/@@/; s/%g$/@g/; /@g$/s/[\\\\&%]/\\\\&/g; s/@@/%@/; s/@@/@%/; s/@g$/%g/' > conftest.subs <<\CEOF $ac_vpsub $extrasub s%@CFLAGS@%$CFLAGS%g s%@CPPFLAGS@%$CPPFLAGS%g s%@CXXFLAGS@%$CXXFLAGS%g s%@DEFS@%$DEFS%g s%@LDFLAGS@%$LDFLAGS%g s%@LIBS@%$LIBS%g s%@exec_prefix@%$exec_prefix%g s%@prefix@%$prefix%g s%@program_transform_name@%$program_transform_name%g s%@CC@%$CC%g s%@CPP@%$CPP%g s%@RANLIB@%$RANLIB%g s%@INSTALL_PROGRAM@%$INSTALL_PROGRAM%g s%@INSTALL_DATA@%$INSTALL_DATA%g s%@opt_objs@%$opt_objs%g s%@opt_cflags@%$opt_cflags%g CEOF EOF cat >> $CONFIG_STATUS <> $CONFIG_STATUS <<\EOF for ac_file in .. $CONFIG_FILES; do if test "x$ac_file" != x..; then # Support "outfile[:infile]", defaulting infile="outfile.in". case "$ac_file" in *:*) ac_file_in=`echo "$ac_file"|sed 's%.*:%%'` ac_file=`echo "$ac_file"|sed 's%:.*%%'` ;; *) ac_file_in="${ac_file}.in" ;; esac # Adjust relative srcdir, etc. for subdirectories. # Remove last slash and all that follows it. Not all systems have dirname. ac_dir=`echo $ac_file|sed 's%/[^/][^/]*$%%'` if test "$ac_dir" != "$ac_file" && test "$ac_dir" != .; then # The file is in a subdirectory. test ! -d "$ac_dir" && mkdir "$ac_dir" ac_dir_suffix="/`echo $ac_dir|sed 's%^\./%%'`" # A "../" for each directory in $ac_dir_suffix. ac_dots=`echo $ac_dir_suffix|sed 's%/[^/]*%../%g'` else ac_dir_suffix= ac_dots= fi case "$ac_given_srcdir" in .) srcdir=. if test -z "$ac_dots"; then top_srcdir=. else top_srcdir=`echo $ac_dots|sed 's%/$%%'`; fi ;; /*) srcdir="$ac_given_srcdir$ac_dir_suffix"; top_srcdir="$ac_given_srcdir" ;; *) # Relative path. srcdir="$ac_dots$ac_given_srcdir$ac_dir_suffix" top_srcdir="$ac_dots$ac_given_srcdir" ;; esac case "$ac_given_INSTALL" in [/$]*) INSTALL="$ac_given_INSTALL" ;; *) INSTALL="$ac_dots$ac_given_INSTALL" ;; esac echo creating "$ac_file" rm -f "$ac_file" configure_input="Generated automatically from `echo $ac_file_in|sed 's%.*/%%'` by configure." case "$ac_file" in *Makefile*) ac_comsub="1i\\ # $configure_input" ;; *) ac_comsub= ;; esac sed -e "$ac_comsub s%@configure_input@%$configure_input%g s%@srcdir@%$srcdir%g s%@top_srcdir@%$top_srcdir%g s%@INSTALL@%$INSTALL%g " -f conftest.subs $ac_given_srcdir/$ac_file_in > $ac_file fi; done rm -f conftest.subs exit 0 EOF chmod +x $CONFIG_STATUS rm -fr confdefs* $ac_clean_files test "$no_create" = yes || ${CONFIG_SHELL-/bin/sh} $CONFIG_STATUS || exit 1 dist/rx/../doc/rx.texi100660 765 144 111765 6071535072 13401 0ustar lordusers\input texinfo @c -*-texinfo-*- @c %**start of header @setfilename rx.info @settitle Rx @setchapternewpage odd @c %**end of header @c This title page illustrates only one of the @c two methods of forming a title page. @titlepage @title Rx @subtitle A Dynamic Automata Library @author Tom Lord @center with excerpts from @emph{The GNU C Library} reference manual @center by Sandra Loosemore @center with @center Richard M. Stallman, Roland McGrath, and Andrew Oram @c The following two commands @c start the copyright page. @page @vskip 0pt plus 1filll Copyright @copyright{} 1995 Cygnus Support Permission is granted to make and distribute verbatim copies of this manual provided the copyright notice and this permission notice are preserved on all copies. Permission is granted to copy and distribute modified versions of this manual under the conditions for verbatim copying, provided that the entire resulting derived work is distributed under the terms of a permission notice identical to this one. Permission is granted to copy and distribute translations of this manual into another language, under the above conditions for modified versions, except that this permission notice may be stated in a translation approved by the author. @end titlepage @node Top, Posix Entry Points, (dir), (dir) @top Rx This document describes Rx. This document applies to version "Rx-diSpencer" of the library named librx.a @menu * Posix Entry Points:: The normal way to call regexp functions. * Performance Considerations:: Special considerations when using Rx. * Semantic Tests:: A test-suite for Posix matcher semantics. * General Functions:: Rx's native entry points. * Rx Theory:: A quick tour of the implementation. * Improvements to Make:: The Rx TODO list. @end menu @node Posix Entry Points, Performance Considerations, Top, Top @section Regular Expression Matching This section is excerpted from @emph{The GNU C Library} reference manual by Sandra Loosemore with Richard M. Stallman, Roland McGrath, and Andrew Oram. The GNU C library supports the standard POSIX.2 interface. Programs using this interface should include the header file @file{rxposix.h}. @menu * POSIX Regexp Compilation:: Using @code{regcomp} to prepare to match. * Flags for POSIX Regexps:: Syntax variations for @code{regcomp}. * Matching POSIX Regexps:: Using @code{regexec} to match the compiled pattern that you get from @code{regcomp}. * Regexp Subexpressions:: Finding which parts of the string were matched. * Subexpression Complications:: Find points of which parts were matched. * Regexp Cleanup:: Freeing storage; reporting errors. @end menu @node POSIX Regexp Compilation @subsection POSIX Regular Expression Compilation Before you can actually match a regular expression, you must @dfn{compile} it. This is not true compilation---it produces a special data structure, not machine instructions. But it is like ordinary compilation in that its purpose is to enable you to ``execute'' the pattern fast. (@xref{Matching POSIX Regexps}, for how to use the compiled regular expression for matching.) There is a special data type for compiled regular expressions: @comment regex.h @comment POSIX.2 @deftp {Data Type} regex_t This type of object holds a compiled regular expression. It is actually a structure. It has just one field that your programs should look at: @table @code @item re_nsub This field holds the number of parenthetical subexpressions in the regular expression that was compiled. @end table There are several other fields, but we don't describe them here, because only the functions in the library should use them. @end deftp After you create a @code{regex_t} object, you can compile a regular expression into it by calling @code{regcomp}. @comment regex.h @comment POSIX.2 @deftypefun int regcomp (regex_t *@var{compiled}, const char *@var{pattern}, int @var{cflags}) The function @code{regcomp} ``compiles'' a regular expression into a data structure that you can use with @code{regexec} to match against a string. The compiled regular expression format is designed for efficient matching. @code{regcomp} stores it into @code{*@var{compiled}}. It's up to you to allocate an object of type @code{regex_t} and pass its address to @code{regcomp}. The argument @var{cflags} lets you specify various options that control the syntax and semantics of regular expressions. @xref{Flags for POSIX Regexps}. If you use the flag @code{REG_NOSUB}, then @code{regcomp} omits from the compiled regular expression the information necessary to record how subexpressions actually match. In this case, you might as well pass @code{0} for the @var{matchptr} and @var{nmatch} arguments when you call @code{regexec}. If you don't use @code{REG_NOSUB}, then the compiled regular expression does have the capacity to record how subexpressions match. Also, @code{regcomp} tells you how many subexpressions @var{pattern} has, by storing the number in @code{@var{compiled}->re_nsub}. You can use that value to decide how long an array to allocate to hold information about subexpression matches. @code{regcomp} returns @code{0} if it succeeds in compiling the regular expression; otherwise, it returns a nonzero error code (see the table below). You can use @code{regerror} to produce an error message string describing the reason for a nonzero value; see @ref{Regexp Cleanup}. @end deftypefun Here are the possible nonzero values that @code{regcomp} can return: @table @code @comment regex.h @comment POSIX.2 @item REG_BADBR There was an invalid @samp{\@{@dots{}\@}} construct in the regular expression. A valid @samp{\@{@dots{}\@}} construct must contain either a single number, or two numbers in increasing order separated by a comma. @comment regex.h @comment POSIX.2 @item REG_BADPAT There was a syntax error in the regular expression. @comment regex.h @comment POSIX.2 @item REG_BADRPT A repetition operator such as @samp{?} or @samp{*} appeared in a bad position (with no preceding subexpression to act on). @comment regex.h @comment POSIX.2 @item REG_ECOLLATE The regular expression referred to an invalid collating element (one not defined in the current locale for string collation). @comment regex.h @comment POSIX.2 @item REG_ECTYPE The regular expression referred to an invalid character class name. @comment regex.h @comment POSIX.2 @item REG_EESCAPE The regular expression ended with @samp{\}. @comment regex.h @comment POSIX.2 @item REG_ESUBREG There was an invalid number in the @samp{\@var{digit}} construct. @comment regex.h @comment POSIX.2 @item REG_EBRACK There were unbalanced square brackets in the regular expression. @comment regex.h @comment POSIX.2 @item REG_EPAREN An extended regular expression had unbalanced parentheses, or a basic regular expression had unbalanced @samp{\(} and @samp{\)}. @comment regex.h @comment POSIX.2 @item REG_EBRACE The regular expression had unbalanced @samp{\@{} and @samp{\@}}. @comment regex.h @comment POSIX.2 @item REG_ERANGE One of the endpoints in a range expression was invalid. @comment regex.h @comment POSIX.2 @item REG_ESPACE @code{regcomp} ran out of memory. @end table @node Flags for POSIX Regexps @subsection Flags for POSIX Regular Expressions These are the bit flags that you can use in the @var{cflags} operand when compiling a regular expression with @code{regcomp}. @table @code @comment regex.h @comment POSIX.2 @item REG_EXTENDED Treat the pattern as an extended regular expression, rather than as a basic regular expression. @comment regex.h @comment POSIX.2 @item REG_ICASE Ignore case when matching letters. @comment regex.h @comment POSIX.2 @item REG_NOSUB Don't bother storing the contents of the @var{matches-ptr} array. @comment regex.h @comment POSIX.2 @item REG_NEWLINE Treat a newline in @var{string} as dividing @var{string} into multiple lines, so that @samp{$} can match before the newline and @samp{^} can match after. Also, don't permit @samp{.} to match a newline, and don't permit @samp{[^@dots{}]} to match a newline. Otherwise, newline acts like any other ordinary character. @end table @node Matching POSIX Regexps @subsection Matching a Compiled POSIX Regular Expression Once you have compiled a regular expression, as described in @ref{POSIX Regexp Compilation}, you can match it against strings using @code{regexec}. A match anywhere inside the string counts as success, unless the regular expression contains anchor characters (@samp{^} or @samp{$}). @comment regex.h @comment POSIX.2 @deftypefun int regexec (regex_t *@var{compiled}, char *@var{string}, size_t @var{nmatch}, regmatch_t @var{matchptr} @t{[]}, int @var{eflags}) This function tries to match the compiled regular expression @code{*@var{compiled}} against @var{string}. @code{regexec} returns @code{0} if the regular expression matches; otherwise, it returns a nonzero value. See the table below for what nonzero values mean. You can use @code{regerror} to produce an error message string describing the reason for a nonzero value; see @ref{Regexp Cleanup}. The argument @var{eflags} is a word of bit flags that enable various options. If you want to get information about what part of @var{string} actually matched the regular expression or its subexpressions, use the arguments @var{matchptr} and @var{nmatch}. Otherwise, pass @code{0} for @var{nmatch}, and @code{NULL} for @var{matchptr}. @xref{Regexp Subexpressions}. @end deftypefun You must match the regular expression with the same set of current locales that were in effect when you compiled the regular expression. The function @code{regexec} accepts the following flags in the @var{eflags} argument: @table @code @comment regex.h @comment POSIX.2 @item REG_NOTBOL Do not regard the beginning of the specified string as the beginning of a line; more generally, don't make any assumptions about what text might precede it. @comment regex.h @comment POSIX.2 @item REG_NOTEOL Do not regard the end of the specified string as the end of a line; more generally, don't make any assumptions about what text might follow it. @end table Here are the possible nonzero values that @code{regexec} can return: @table @code @comment regex.h @comment POSIX.2 @item REG_NOMATCH The pattern didn't match the string. This isn't really an error. @comment regex.h @comment POSIX.2 @item REG_ESPACE @code{regexec} ran out of memory. @end table @node Regexp Subexpressions @subsection Match Results with Subexpressions When @code{regexec} matches parenthetical subexpressions of @var{pattern}, it records which parts of @var{string} they match. It returns that information by storing the offsets into an array whose elements are structures of type @code{regmatch_t}. The first element of the array (index @code{0}) records the part of the string that matched the entire regular expression. Each other element of the array records the beginning and end of the part that matched a single parenthetical subexpression. @comment regex.h @comment POSIX.2 @deftp {Data Type} regmatch_t This is the data type of the @var{matcharray} array that you pass to @code{regexec}. It containes two structure fields, as follows: @table @code @item rm_so The offset in @var{string} of the beginning of a substring. Add this value to @var{string} to get the address of that part. @item rm_eo The offset in @var{string} of the end of the substring. @end table @end deftp @comment regex.h @comment POSIX.2 @deftp {Data Type} regoff_t @code{regoff_t} is an alias for another signed integer type. The fields of @code{regmatch_t} have type @code{regoff_t}. @end deftp The @code{regmatch_t} elements correspond to subexpressions positionally; the first element (index @code{1}) records where the first subexpression matched, the second element records the second subexpression, and so on. The order of the subexpressions is the order in which they begin. When you call @code{regexec}, you specify how long the @var{matchptr} array is, with the @var{nmatch} argument. This tells @code{regexec} how many elements to store. If the actual regular expression has more than @var{nmatch} subexpressions, then you won't get offset information about the rest of them. But this doesn't alter whether the pattern matches a particular string or not. If you don't want @code{regexec} to return any information about where the subexpressions matched, you can either supply @code{0} for @var{nmatch}, or use the flag @code{REG_NOSUB} when you compile the pattern with @code{regcomp}. @node Subexpression Complications @subsection Complications in Subexpression Matching Sometimes a subexpression matches a substring of no characters. This happens when @samp{f\(o*\)} matches the string @samp{fum}. (It really matches just the @samp{f}.) In this case, both of the offsets identify the point in the string where the null substring was found. In this example, the offsets are both @code{1}. Sometimes the entire regular expression can match without using some of its subexpressions at all---for example, when @samp{ba\(na\)*} matches the string @samp{ba}, the parenthetical subexpression is not used. When this happens, @code{regexec} stores @code{-1} in both fields of the element for that subexpression. Sometimes matching the entire regular expression can match a particular subexpression more than once---for example, when @samp{ba\(na\)*} matches the string @samp{bananana}, the parenthetical subexpression matches three times. When this happens, @code{regexec} usually stores the offsets of the last part of the string that matched the subexpression. In the case of @samp{bananana}, these offsets are @code{6} and @code{8}. But the last match is not always the one that is chosen. It's more accurate to say that the last @emph{opportunity} to match is the one that takes precedence. What this means is that when one subexpression appears within another, then the results reported for the inner subexpression reflect whatever happened on the last match of the outer subexpression. For an example, consider @samp{\(ba\(na\)*s \)*} matching the string @samp{bananas bas }. The last time the inner expression actually matches is near the end of the first word. But it is @emph{considered} again in the second word, and fails to match there. @code{regexec} reports nonuse of the ``na'' subexpression. Another place where this rule applies is when the regular expression @w{@samp{\(ba\(na\)*s \|nefer\(ti\)* \)*}} matches @samp{bananas nefertiti}. The ``na'' subexpression does match in the first word, but it doesn't match in the second word because the other alternative is used there. Once again, the second repetition of the outer subexpression overrides the first, and within that second repetition, the ``na'' subexpression is not used. So @code{regexec} reports nonuse of the ``na'' subexpression. @node Regexp Cleanup @subsection POSIX Regexp Matching Cleanup When you are finished using a compiled regular expression, you can free the storage it uses by calling @code{regfree}. @comment regex.h @comment POSIX.2 @deftypefun void regfree (regex_t *@var{compiled}) Calling @code{regfree} frees all the storage that @code{*@var{compiled}} points to. This includes various internal fields of the @code{regex_t} structure that aren't documented in this manual. @code{regfree} does not free the object @code{*@var{compiled}} itself. @end deftypefun You should always free the space in a @code{regex_t} structure with @code{regfree} before using the structure to compile another regular expression. When @code{regcomp} or @code{regexec} reports an error, you can use the function @code{regerror} to turn it into an error message string. @comment regex.h @comment POSIX.2 @deftypefun size_t regerror (int @var{errcode}, regex_t *@var{compiled}, char *@var{buffer}, size_t @var{length}) This function produces an error message string for the error code @var{errcode}, and stores the string in @var{length} bytes of memory starting at @var{buffer}. For the @var{compiled} argument, supply the same compiled regular expression structure that @code{regcomp} or @code{regexec} was working with when it got the error. Alternatively, you can supply @code{NULL} for @var{compiled}; you will still get a meaningful error message, but it might not be as detailed. If the error message can't fit in @var{length} bytes (including a terminating null character), then @code{regerror} truncates it. The string that @code{regerror} stores is always null-terminated even if it has been truncated. The return value of @code{regerror} is the minimum length needed to store the entire error message. If this is less than @var{length}, then the error message was not truncated, and you can use it. Otherwise, you should call @code{regerror} again with a larger buffer. Here is a function which uses @code{regerror}, but always dynamically allocates a buffer for the error message: @smallexample char *get_regerror (int errcode, regex_t *compiled) @{ size_t length = regerror (errcode, compiled, NULL, 0); char *buffer = xmalloc (length); (void) regerror (errcode, compiled, buffer, length); return buffer; @} @end smallexample @end deftypefun @node Performance Considerations, Semantic Tests, Posix Entry Points, Top @section Performance Considerations @cindex Performance Considerations In order to use the Rx implementation of Posix regexp functions most effectively, it may help to know a bit about performance tuning in Rx. @menu * Avoiding Redundant Compilations :: Compiling Regexps is costly. * Controlling Memory Use:: You can limit Rx's memory use. @end menu @node Avoiding Redundant Compilations , Controlling Memory Use, Performance Considerations, Performance Considerations @subsection Avoiding Redundant Compilations Rx implements the Posix regexp entry points @code{regcomp}, @code{regerror}, @code{regexec}, and @code{regfree}. Some special considerations apply when using the Rx versions. First, @code{regcomp} is a comparatively expensive operation. Addiitonally, @code{regexec} tends to perform better when a compiled expression is compiled once and used twice than when the pattern is compiled twice and used twice. Internally, Rx caches that information about a pattern which is constructed dynamicly by regexec. You can save on compilation costs and maximize the effectiveness of the Rx cache by re-using compiled expressions when it is convenient. For example, for reasons unrelated to Rx, it has long been the practice in GNU emacs to always save one compiled regular expression (the most recent). Before compiling a new expression, Emacs checks to see if the new pattern is the same as the one that is already compiled. This is the sort of optimization that Rx likes. (It may even win, in some cases, to cache more than a single compiled regexp.) In the long-run, Rx should be modified to cache compilations on its own. (@xref{Improvements to Make}.) Sometimes programmers write programs with many regexps, all known staticly at compile time. This can be a problem with Rx because when the program starts up, or as it runs, all of those staticly known regexps have to be compiled, which may be too time consuming. One easy fix for programs with many static regexps is to use @code{flex} or another lexical-analyzer generator program instead. Lexer-generators are optimized for the case of compiling staticly known regexps. There are still reasons why staticly known regexps may not be convertable to @code{flex} input -- for example, the Posix pattern language is more general than @code{flex}'s. In the long-run, it may be a good idea to extend Rx to support static compilation of regexps. (@xref{Improvements to Make}.) @node Controlling Memory Use, , Avoiding Redundant Compilations , Performance Considerations @subsection Controlling Memory Use The size of the cache used by regexec is subject to control by programmers. Additionally, by using lower level entry points, programmers can create multiple, distinct Rx caches. This part of the code is subject to change and so is not yet documented. See "default_cache" in rxsuper.c and the parameters at the top of "rxbasic.c" to find some parameters you can tune or hints about creating alternative regexec caches. @node Semantic Tests, General Functions, Performance Considerations, Top @section Semantic Tests @cindex Semantic tests The distribution of @emph{Rx} includes a test-suite, aimed at testing the conformance of an implementation of the Posix regexp functions to the standard. Currently, the tests come from two sources: a test-suite started by Henry Spencer, and the "Khadafy" test suite by Scott Anderson. The tests come in three files: @itemize @bullet @item runtests.c -- The driver program. @item TESTS -- The list of test cases @item TESTS2C.sed -- A script to convert test cases into C. @end itemize @menu * Running the Tests:: Invoking the test suite. * Adding New Tests:: Extending the test suite. @end menu @node Running the Tests, Adding New Tests, Semantic Tests, Semantic Tests @subsection Running the Tests To build the test program, use @code{make runtests}. (Detailed configuration and build instructions can be found in INSTALL). Invoked with no arguments, @code{runtests} runs all test-cases. For each test case, a sequence number is printed. If there is a problem with that case, more information is printed. Output like this: @example #0 #1 #2 ... @end example indicates a successful run. (The output format is likely to change in the future to make it easier to automate testing of Rx.) With a single numeric argument, runtests executes just that test and no others. This is handy when debugging. Sometimes a bug will occur when all tests are run, but disappear when just the problematic test is run. Usually this has to do with memory or cache corruption. @node Adding New Tests, , Running the Tests, Semantic Tests @subsection Adding New Tests This list of tests used by @code{runtests} is found in the file @code{TESTS}. Each line of that file is a list of colon separated fields similar to: @example 1:^(ab|cd)e:abcde @end example The first field is the expected return value of regexec, or '2' meaning that the pattern is not valid. The second field is the regular expression being tested. The third field is the string to which the pattern is to be compared. The file @code{TESTS2C.sed} is used to convert the test cases into initializers for an array of structures (which is automated in the Makefile). To add a new test case, add lines to @code{TESTS} and recompile @code{runtests}. @node General Functions, Rx Theory, Semantic Tests, Top @section General Functions @cindex Generalized entry points Here is a whirlwind tour of the lower-level Rx entry points and data structures. This is not intended to be complete, but rather to be a help to anyone reading the code itself. These are presented in the same order in which they might typically be used. @menu * Compiling Expressions:: Strings->Syntax Trees * Posixly Correct Matches:: Checking for a Posixly correct match. * Hairy Matches:: Matching with multi-part strings. * NFA Functions:: Forget Posix, eat NFAs raw. @end menu @node Compiling Expressions, Posixly Correct Matches, General Functions, General Functions @subsection Compiling Expressions @deftypefn {Internal Function} reg_errcode_t rx_parse (struct rexp_node ** rexp_p, __const__ char *pattern, int size, unsigned long syntax, int cset_size, unsigned char *translate); Compile a regular expression. @var{rexp_p} will return a syntax tree for the pattern. See @code{"rxnode.h"} to learn about syntax trees. Minimally, you should know that syntax trees are reference counted using @code{rx_save_rexp} and @code{rx_free_rexp}. @var{pattern} and @var{size} are the expression and its length. The expression need not be 0-terminated. The bits set in @var{syntax} control details of the language understood by the parser. The meaning of each bit is described in @code{"rxgnucomp.h"}. @var{cset_size} is usually 256. It should not be much larger than that or space-performance will suffer badly. @var{translate} is a translation table -- an array of characters. Characters in the pattern are looked up in @var{translate} as they are read. Additionally, the pattern is compiled presuming that the same translation table will be applied to characters in the string being searched (meaning that the pattern is compiled in such a way that the effect will be as if every character in the target string is looked up in the translation table, even though the expense of that lookup is usually avoided). A return of 0 indicates success, otherwise, the error can be looked up in the @code{rx_error_msg} array. @end deftypefn @deftypefn {Internal Function} int rx_posix_analyze_rexp (struct rexp_node *** subexps, int * n_subexps, struct rexp_node * node, int id) Staticly analyze a regular expression in preparation for matching. This function fills in some fields of the nodes of a syntax tree. @code{subexps} and @code{n_subexps} return a malloced array of pointers into the syntax tree, one per parenthesized subexpression. The caller must eventually free this array. @code{node} is the pattern to be analyzed. The tree returned by @code{rx_parse} is suitable for this. @code{id} should be 1. Ignore the return value. This function is not robust in the case that @code{malloc} or @code{realloc} returns 0 (which should be fixed). @end deftypefn You might expect that @code{rx_posix_analyze_rexp} and @code{rx_parse} should be combined. They are not because the actions of @code{rx_posix_analyze_rexp} make sense for any syntax tree, regardless of how it is constructed. @code{rx_parse} is just one way to build a syntax tree -- it is subject to replacement and trees can be built without using a parser at all (c.f. @code{"rxnode.h"}). @node Posixly Correct Matches, Hairy Matches, Compiling Expressions, General Functions @subsection Checking for Posixly Correct Matches with One-Piece Strings Looking for a match that is Posixly correct is conceptually tricky. Here is one way to think of it: The Posix expression languages, absent of any consideration of the meaning of "leftmost-longest", underspecify the return value of @code{regexec}. Consider the example: @example Pattern: foo\(.*\)\(.*\)bar Target string: fooxxxbar @end example Initially it is ambiguous how this should be matched. The ambiguity is exposed by considering the possible bindings for the subexpressions @code{\1} and @code{\2}: @example Possible outcomes: \1 == "xxx" \2 == "" \1 == "xx" \2 == "x" \1 == "x" \2 == "xx" \1 == "" \2 == "xxx" @end example Posix tells us that the first outcome is the prefered one if the illustrated pattern is used alone (this is a consequence of the "leftmost longest" rule). But Posix also says the answer may be different if the pattern is a sub-pattern of a larger pattern. For example: @example Pattern: foo\(.*\)\(.*\)bar\2 Target string: fooxxxbarxx @end example Now the only acceptable outcome is: @example \1 == "x" \2 == "xx" @end example Three entry points provide a relatively simple interface to this moderately complicated situation: @deftypefn {Internal Function} {struct rx_solutions *} rx_basic_make_solutions (struct rx_registers * regs, struct rexp_node * expression, struct rexp_node ** subexps, int start, int end, struct rx_context_rules * rules, char * str) @deftypefnx {Internal Function} void rx_basic_free_solutions (struct rx_solutions * solns); @deftypefnx {Internal Function} {enum rx_answers} rx_next_solution (struct rx_solutions * solns) @deftypefnx {Internal Function} void rx_basic_free_solutions (struct rx_solutions * solns) @code{make_solutions} returns a lazilly constructed stream of possible solutions for a given regexp and target string. @code{next_solution} constructs and returns the next solution from a list returned by @code{make_solutions}. Solutions are returned in order of preferability according to the Posix spec. So, the first solution is the leftmost-longest, and the last, the rightmost-shortest. @var{regs} is where subexpression extents are tracked. @var{expression} is the expression to solve. @var{subexps} is as returned by @code{rx_posix_analyze_rexp}. @var{start} and @var{end} are the integer addresses of the data to be compared to the pattern. A solution must match all characters starting at @var{start} and ending at @code{@var{end} - 1}. @var{rules} are some bits that control the precise meaning of "^" and "$". See @code{"rxcontext.h"}. @var{str} is the data to be matched. Note that anchoring operators can cause the matcher to look beyond @var{start} and @var{end} when testing a match. It may make sense to pass @var{start} > 0, for example. @code{make_solutions} returns a pointer to a solution list, or 0 if an allocation fails. @code{next_solution} returns a status which can be: @example @code{rx_yes} -- The pattern matches exactly. @code{rx_maybe} -- The pattern does not match, but might if more characters were added to the string. @code{rx_no} -- The pattern does not match and could not be made to match even by adding more characters. @end example In addition, @var{regs} is filled in by @code{next_solution}. Finally, @code{free_solutions} is used to release resources consumed by a solution list. @end deftypefn @node Hairy Matches, NFA Functions, Posixly Correct Matches, General Functions @subsection Hairy Matches and Suspendable Searches The functions @code{rx_basic_make_solutions} and @code{rx_basic_free_solutions} are used to find matches in a one-piece string. More general entry points are provided for multi-piece strings, which may or may not be entirely available in memory at one time: @deftypefn {Internal Function} {struct rx_solutions *} rx_make_solutions (struct rx_registers * regs, struct rx_unfaniverse * verse, struct rexp_node * expression, struct rexp_node ** subexps, int cset_size, int start, int end, rx_vmfn vmfn, rx_contextfn contextfn, void * closure) @deftypefnx {Internal Function} void rx_free_solutions (struct rx_solutions * solns) These functions are similar to their @code{_basic_} analogues, but substitute some new parameters for the one piece string: @var{vmfn} is a function provided by the caller that returns at least part of the string being matched. @var{contextfn} is a function provided by the caller that does the work of the backreference operators, "^", and "$". @var{closure} is data provided by the caller, passed through to vmfn and contextfn. @example typedef enum rx_answers (*rx_vmfn) P((void * closure, unsigned char ** burst, int * len, int * offset, int start, int end, int need)); typedef enum rx_answers (*rx_contextfn) P((void * closure, int type, int start, int end, struct rx_registers * regs)); @end example @code{vmfn} is asked for a range of characters from @var{start} to @var{end}. @var{need} is in that range and is the only position that must be returned by vmfn. @var{burst}, @var{len}, and @var{offset} are output parameters. @var{offset} will get the integer address of burst, and @var{len} the number of valid characters in the burst. @var{vmfn} may be asked for the empty string from @var{end} to @var{end} and should be able to handle that case. If @code{vmfn} provides the needed character, it should return @code{rx_yes}. If @code{vmfn} can't provide the character now, but might be able to later, it should return @code{rx_maybe}. This will cause @code{rx_next_solution} to quickly return @code{rx_maybe}. If @code{rx_next_solution} is called again, it will retry the call to vmfn and possibly resume the search. If @code{vmfn} can not provide the needed character, it should return @code{rx_no}. @code{rx_next_solution} will not normally ask for out-of-range characters, so usually there is no good reason to return @code{rx_no}, but the code is supposed to be robust in case you do. The @code{contextfn} also returns a @code{yes/no/maybe} status. It tests whether the characters in @var{start} to @var{end} satisfy the operator @var{type}. @code{type} is an ascii value; one of the characters in "^$123456789". @end deftypefn @node NFA Functions, , Hairy Matches, General Functions @subsection NFA Functions and Super-NFA Functions Not really documented yet. @example rx.h, rxnfa.h -- translating an expression into an NFA. rxsuper.h -- optimizing the NFA into a DFA, or at least an NFA with fewer branch points. rxanal.h -- how to use DFAs to find longest matches, matches of a specific length, the shortest match, etc. rxsimp.h -- Translating a Posix "regular expression" into a true regular expression that describes a superset language. @end example @node Rx Theory, Improvements to Make, General Functions, Top @section Rx Theory @cindex Rx Theory There are two match algorithms. One is for truly regular regexps (those that can be reduced to a dfa). The other is for non-regular regexps. The dfa algorithm implements the idea suggested in @cite{Compilers} by Aho, Sethi and Ullman: @quotation [One] approach [to pattern matching regular expressions] is to use a DFA, but avoid constructing all of the transition table by using a technique called "lazy transition evaluation". Here, transitions are computed at run time [when] actually needed. [T]ransitions are stored in a cache. [....] If the cache becomes full, we can erase some previously computed transition to make room for the new transition. @end quotation The implementation in Rx is generalized from that, but the above description covers what is used for Posix patterns. The non-dfa algorithm implements a "recursive decomposition" technique described in email by Henry Spencer. For a given pattern, this algorithm first checks to see if a simpler, superset language, DFA-pattern matches. If it does, then this algorithm does the detail-work to see if the non-DFA pattern matches. The detail work usually involves recursing on subpatterns. For example, a concatentation of two subexpressions matches a string if the string can be divided into two parts, each matching one subexpression, in the right order. More than one solution is often possible for a given pattern. This ambiguity is the subject of the "leftmost longest" rules in the spec, and the back-tracking oriented stream-of-solution functions @code{rx_make_solutions}, @code{rx_next_solution} and @code{rx_free_solutions}. @example rxspencer.[ch] -- The non-DFA algorithm rxanal.[ch] rxsuper.[ch] rxnfa.[ch] -- The DFA algorithm @end example @node Improvements to Make, , Rx Theory, Top @section Improvements to Make @cindex Improvements to make An approximation of the GNU entry points is needed. This can not easily be an exact approximation because the GNU entry points have no function for freeing a compiled expression -- they assume that compiled expressions can be freed using a single call to "free(3)". For that reason, do not use the names of the old GNU entry points ...use "rx_" instead. This version of Rx hasn't been profiled and tuned. It might be nice to have some C++ types to hide reference counting and provide handy constructors for rexp_nodes and nfa parts. Right now, the regexp->regular regexp mapping in rxsimp.[ch] is cached for a particular input expression. I think it would probably be a big win in many situations if it were cached, like unique nfas (c.f. rxunfa.[ch]), for equivalent expressions. It would speed up matching to lift the non-regular constructs to the top of the syntax tree. e.g., if you have: @example pattern: "abc$" syntax tree: (concat (cset "a") (concat (cset "b") (concat (cset "c") (operator "$")))) @end example it involves a lot more work (in "next_solution) than if you have the equivalent: @example same pattern: "abc$" syntax tree: (concat (concat (cset "a") (concat (cset "b") (cset "c"))) (operator "$")) @end example in fact, while doing that, notice that subtrees of nothing-but-concat can conceivably be optimized so that next_solution just uses "strcmp". For that matter, a tree which is a disjunction of conjunction (a tree of "|" whose leaves are trees of concat nodes) could be optimized to use a table method (Boyer/Moore?). Again, this would be for next_solution but might involve adding stuff to rx_posix_analyze_rexp. As more and more optimizations are added, programmers should be given simple control over which optimizations are important for a particular pattern. For example, packages in emacs often include a few regexps that are used heavily -- it would be worth the trouble to arrange for those regexps to be compiled specially. Static compilation of regexps is mostly a job for lex, but posix patterns are more general and have a nicer interface. What's more, it'd be nice to have the benefits of static compilation even for dynamicly loaded regexps (e.g., imagine if when an emacs lisp file is byte-compiled, the regexps it contains are also compiled). A lot of optimizations apply to just syntax trees. One approach to static compilation is to make an ugly but fast-reading syntax for trees, and a printer for that syntax. Then, after optimiztaions, a pattern could be stored as a string and later read back in, already optimized. A similar trick could be done for the nfa built in rxnfa.c. next_solution already considers the length of a fixed-length subexpression, but many more lenght optimizations are possible. I'd guess these will pay off well. @cindex TODO dist/rx/../doc/rx.info100660 765 144 110710 6071534721 13350 0ustar lordusersThis is Info file rx.info, produced by Makeinfo-1.63 from the input file rx.texi.  File: rx.info, Node: Top, Next: Posix Entry Points, Prev: (dir), Up: (dir) Rx ** This document describes Rx. This document applies to version "Rx-diSpencer" of the library named librx.a * Menu: * Posix Entry Points:: * Performance Considerations:: Special considerations when using Rx. * Semantic Tests:: A test-suite for Posix matcher semantics. * General Functions:: Rx's native entry points. * Rx Theory:: A quick tour of the implementation. * Improvements to Make:: The Rx TODO list.  File: rx.info, Node: Posix Entry Points, Next: Performance Considerations, Prev: Top, Up: Top Regular Expression Matching =========================== This section is excerpted from *The GNU C Library* reference manual by Sandra Loosemore with Richard M. Stallman, Roland McGrath, and Andrew Oram. The GNU C library supports the standard POSIX.2 interface. Programs using this interface should include the header file `rx.h'. * Menu: * POSIX Regexp Compilation:: Using `regcomp' to prepare to match. * Flags for POSIX Regexps:: Syntax variations for `regcomp'. * Matching POSIX Regexps:: Using `regexec' to match the compiled pattern that you get from `regcomp'. * Regexp Subexpressions:: Finding which parts of the string were matched. * Subexpression Complications:: Find points of which parts were matched. * Regexp Cleanup:: Freeing storage; reporting errors.  File: rx.info, Node: POSIX Regexp Compilation, Next: Flags for POSIX Regexps, Up: Posix Entry Points POSIX Regular Expression Compilation ------------------------------------ Before you can actually match a regular expression, you must "compile" it. This is not true compilation--it produces a special data structure, not machine instructions. But it is like ordinary compilation in that its purpose is to enable you to "execute" the pattern fast. (*Note Matching POSIX Regexps::, for how to use the compiled regular expression for matching.) There is a special data type for compiled regular expressions: - Data Type: regex_t This type of object holds a compiled regular expression. It is actually a structure. It has just one field that your programs should look at: `re_nsub' This field holds the number of parenthetical subexpressions in the regular expression that was compiled. There are several other fields, but we don't describe them here, because only the functions in the library should use them. After you create a `regex_t' object, you can compile a regular expression into it by calling `regcomp'. - Function: int regcomp (regex_t *COMPILED, const char *PATTERN, int CFLAGS) The function `regcomp' "compiles" a regular expression into a data structure that you can use with `regexec' to match against a string. The compiled regular expression format is designed for efficient matching. `regcomp' stores it into `*COMPILED'. It's up to you to allocate an object of type `regex_t' and pass its address to `regcomp'. The argument CFLAGS lets you specify various options that control the syntax and semantics of regular expressions. *Note Flags for POSIX Regexps::. If you use the flag `REG_NOSUB', then `regcomp' omits from the compiled regular expression the information necessary to record how subexpressions actually match. In this case, you might as well pass `0' for the MATCHPTR and NMATCH arguments when you call `regexec'. If you don't use `REG_NOSUB', then the compiled regular expression does have the capacity to record how subexpressions match. Also, `regcomp' tells you how many subexpressions PATTERN has, by storing the number in `COMPILED->re_nsub'. You can use that value to decide how long an array to allocate to hold information about subexpression matches. `regcomp' returns `0' if it succeeds in compiling the regular expression; otherwise, it returns a nonzero error code (see the table below). You can use `regerror' to produce an error message string describing the reason for a nonzero value; see *Note Regexp Cleanup::. Here are the possible nonzero values that `regcomp' can return: `REG_BADBR' There was an invalid `\{...\}' construct in the regular expression. A valid `\{...\}' construct must contain either a single number, or two numbers in increasing order separated by a comma. `REG_BADPAT' There was a syntax error in the regular expression. `REG_BADRPT' A repetition operator such as `?' or `*' appeared in a bad position (with no preceding subexpression to act on). `REG_ECOLLATE' The regular expression referred to an invalid collating element (one not defined in the current locale for string collation). `REG_ECTYPE' The regular expression referred to an invalid character class name. `REG_EESCAPE' The regular expression ended with `\'. `REG_ESUBREG' There was an invalid number in the `\DIGIT' construct. `REG_EBRACK' There were unbalanced square brackets in the regular expression. `REG_EPAREN' An extended regular expression had unbalanced parentheses, or a basic regular expression had unbalanced `\(' and `\)'. `REG_EBRACE' The regular expression had unbalanced `\{' and `\}'. `REG_ERANGE' One of the endpoints in a range expression was invalid. `REG_ESPACE' `regcomp' ran out of memory.  File: rx.info, Node: Flags for POSIX Regexps, Next: Matching POSIX Regexps, Prev: POSIX Regexp Compilation, Up: Posix Entry Points Flags for POSIX Regular Expressions ----------------------------------- These are the bit flags that you can use in the CFLAGS operand when compiling a regular expression with `regcomp'. `REG_EXTENDED' Treat the pattern as an extended regular expression, rather than as a basic regular expression. `REG_ICASE' Ignore case when matching letters. `REG_NOSUB' Don't bother storing the contents of the MATCHES-PTR array. `REG_NEWLINE' Treat a newline in STRING as dividing STRING into multiple lines, so that `$' can match before the newline and `^' can match after. Also, don't permit `.' to match a newline, and don't permit `[^...]' to match a newline. Otherwise, newline acts like any other ordinary character.  File: rx.info, Node: Matching POSIX Regexps, Next: Regexp Subexpressions, Prev: Flags for POSIX Regexps, Up: Posix Entry Points Matching a Compiled POSIX Regular Expression -------------------------------------------- Once you have compiled a regular expression, as described in *Note POSIX Regexp Compilation::, you can match it against strings using `regexec'. A match anywhere inside the string counts as success, unless the regular expression contains anchor characters (`^' or `$'). - Function: int regexec (regex_t *COMPILED, char *STRING, size_t NMATCH, regmatch_t MATCHPTR [], int EFLAGS) This function tries to match the compiled regular expression `*COMPILED' against STRING. `regexec' returns `0' if the regular expression matches; otherwise, it returns a nonzero value. See the table below for what nonzero values mean. You can use `regerror' to produce an error message string describing the reason for a nonzero value; see *Note Regexp Cleanup::. The argument EFLAGS is a word of bit flags that enable various options. If you want to get information about what part of STRING actually matched the regular expression or its subexpressions, use the arguments MATCHPTR and NMATCH. Otherwise, pass `0' for NMATCH, and `NULL' for MATCHPTR. *Note Regexp Subexpressions::. You must match the regular expression with the same set of current locales that were in effect when you compiled the regular expression. The function `regexec' accepts the following flags in the EFLAGS argument: `REG_NOTBOL' Do not regard the beginning of the specified string as the beginning of a line; more generally, don't make any assumptions about what text might precede it. `REG_NOTEOL' Do not regard the end of the specified string as the end of a line; more generally, don't make any assumptions about what text might follow it. Here are the possible nonzero values that `regexec' can return: `REG_NOMATCH' The pattern didn't match the string. This isn't really an error. `REG_ESPACE' `regexec' ran out of memory.  File: rx.info, Node: Regexp Subexpressions, Next: Subexpression Complications, Prev: Matching POSIX Regexps, Up: Posix Entry Points Match Results with Subexpressions --------------------------------- When `regexec' matches parenthetical subexpressions of PATTERN, it records which parts of STRING they match. It returns that information by storing the offsets into an array whose elements are structures of type `regmatch_t'. The first element of the array (index `0') records the part of the string that matched the entire regular expression. Each other element of the array records the beginning and end of the part that matched a single parenthetical subexpression. - Data Type: regmatch_t This is the data type of the MATCHARRAY array that you pass to `regexec'. It containes two structure fields, as follows: `rm_so' The offset in STRING of the beginning of a substring. Add this value to STRING to get the address of that part. `rm_eo' The offset in STRING of the end of the substring. - Data Type: regoff_t `regoff_t' is an alias for another signed integer type. The fields of `regmatch_t' have type `regoff_t'. The `regmatch_t' elements correspond to subexpressions positionally; the first element (index `1') records where the first subexpression matched, the second element records the second subexpression, and so on. The order of the subexpressions is the order in which they begin. When you call `regexec', you specify how long the MATCHPTR array is, with the NMATCH argument. This tells `regexec' how many elements to store. If the actual regular expression has more than NMATCH subexpressions, then you won't get offset information about the rest of them. But this doesn't alter whether the pattern matches a particular string or not. If you don't want `regexec' to return any information about where the subexpressions matched, you can either supply `0' for NMATCH, or use the flag `REG_NOSUB' when you compile the pattern with `regcomp'.  File: rx.info, Node: Subexpression Complications, Next: Regexp Cleanup, Prev: Regexp Subexpressions, Up: Posix Entry Points Complications in Subexpression Matching --------------------------------------- Sometimes a subexpression matches a substring of no characters. This happens when `f\(o*\)' matches the string `fum'. (It really matches just the `f'.) In this case, both of the offsets identify the point in the string where the null substring was found. In this example, the offsets are both `1'. Sometimes the entire regular expression can match without using some of its subexpressions at all--for example, when `ba\(na\)*' matches the string `ba', the parenthetical subexpression is not used. When this happens, `regexec' stores `-1' in both fields of the element for that subexpression. Sometimes matching the entire regular expression can match a particular subexpression more than once--for example, when `ba\(na\)*' matches the string `bananana', the parenthetical subexpression matches three times. When this happens, `regexec' usually stores the offsets of the last part of the string that matched the subexpression. In the case of `bananana', these offsets are `6' and `8'. But the last match is not always the one that is chosen. It's more accurate to say that the last *opportunity* to match is the one that takes precedence. What this means is that when one subexpression appears within another, then the results reported for the inner subexpression reflect whatever happened on the last match of the outer subexpression. For an example, consider `\(ba\(na\)*s \)*' matching the string `bananas bas '. The last time the inner expression actually matches is near the end of the first word. But it is *considered* again in the second word, and fails to match there. `regexec' reports nonuse of the "na" subexpression. Another place where this rule applies is when the regular expression `\(ba\(na\)*s \|nefer\(ti\)* \)*' matches `bananas nefertiti'. The "na" subexpression does match in the first word, but it doesn't match in the second word because the other alternative is used there. Once again, the second repetition of the outer subexpression overrides the first, and within that second repetition, the "na" subexpression is not used. So `regexec' reports nonuse of the "na" subexpression.  File: rx.info, Node: Regexp Cleanup, Prev: Subexpression Complications, Up: Posix Entry Points POSIX Regexp Matching Cleanup ----------------------------- When you are finished using a compiled regular expression, you can free the storage it uses by calling `regfree'. - Function: void regfree (regex_t *COMPILED) Calling `regfree' frees all the storage that `*COMPILED' points to. This includes various internal fields of the `regex_t' structure that aren't documented in this manual. `regfree' does not free the object `*COMPILED' itself. You should always free the space in a `regex_t' structure with `regfree' before using the structure to compile another regular expression. When `regcomp' or `regexec' reports an error, you can use the function `regerror' to turn it into an error message string. - Function: size_t regerror (int ERRCODE, regex_t *COMPILED, char *BUFFER, size_t LENGTH) This function produces an error message string for the error code ERRCODE, and stores the string in LENGTH bytes of memory starting at BUFFER. For the COMPILED argument, supply the same compiled regular expression structure that `regcomp' or `regexec' was working with when it got the error. Alternatively, you can supply `NULL' for COMPILED; you will still get a meaningful error message, but it might not be as detailed. If the error message can't fit in LENGTH bytes (including a terminating null character), then `regerror' truncates it. The string that `regerror' stores is always null-terminated even if it has been truncated. The return value of `regerror' is the minimum length needed to store the entire error message. If this is less than LENGTH, then the error message was not truncated, and you can use it. Otherwise, you should call `regerror' again with a larger buffer. Here is a function which uses `regerror', but always dynamically allocates a buffer for the error message: char *get_regerror (int errcode, regex_t *compiled) { size_t length = regerror (errcode, compiled, NULL, 0); char *buffer = xmalloc (length); (void) regerror (errcode, compiled, buffer, length); return buffer; }  File: rx.info, Node: Performance Considerations, Next: Semantic Tests, Prev: Posix Entry Points, Up: Top Performance Considerations ========================== In order to use the Rx implementation of Posix regexp functions, you should include `' in your program. In order to use the Rx implementation of Posix regexp functions most effectively, it may help to know a bit about performance tuning in Rx. * Menu: * Avoiding Redundant Compilations :: Compiling Regexps is costly. * Controlling Memory Use:: You can limit Rx's memory use.  File: rx.info, Node: Avoiding Redundant Compilations, Next: Controlling Memory Use, Prev: Performance Considerations, Up: Performance Considerations Avoiding Redundant Compilations ------------------------------- Rx implements the Posix regexp entry points `regcomp', `regerror', `regexec', and `regfree'. Some special considerations apply when using the Rx versions. First, `regcomp' is a comparatively expensive operation. Addiitonally, `regexec' tends to perform better when a compiled expression is compiled once and used twice than when the pattern is compiled twice and used twice. Internally, Rx caches that information about a pattern which is constructed dynamicly by regexec. You can save on compilation costs and maximize the effectiveness of the Rx cache by re-using compiled expressions when it is convenient. For example, for reasons unrelated to Rx, it has long been the practice in GNU emacs to always save one compiled regular expression (the most recent). Before compiling a new expression, Emacs checks to see if the new pattern is the same as the one that is already compiled. This is the sort of optimization that Rx likes. (It may even win, in some cases, to cache more than a single compiled regexp.) In the long-run, Rx should be modified to cache compilations on its own. (*Note Improvements to Make::.) Sometimes programmers write programs with many regexps, all known staticly at compile time. This can be a problem with Rx because when the program starts up, or as it runs, all of those staticly known regexps have to be compiled, which may be too time consuming. One easy fix for programs with many static regexps is to use `flex' or another lexical-analyzer generator program instead. Lexer-generators are optimized for the case of compiling staticly known regexps. There are still reasons why staticly known regexps may not be convertable to `flex' input - for example, the Posix pattern language is more general than `flex''s. In the long-run, it may be a good idea to extend Rx to support static compilation of regexps. (*Note Improvements to Make::.)  File: rx.info, Node: Controlling Memory Use, Prev: Avoiding Redundant Compilations, Up: Performance Considerations Controlling Memory Use ---------------------- The size of the cache used by regexec is subject to control by programmers. Additionally, by using lower level entry points, programmers can create multiple, distinct Rx caches. This part of the code is subject to change and so is not yet documented. See "default_cache" in rxsuper.c and the parameters at the top of "rxbasic.c" to find some parameters you can tune or hints about creating alternative regexec caches.  File: rx.info, Node: Semantic Tests, Next: General Functions, Prev: Performance Considerations, Up: Top Semantic Tests ============== The distribution of *Rx* includes a test-suite, aimed at testing the conformance of an implementation of the Posix regexp functions to the standard. Currently, the tests come from two sources: a test-suite started by Henry Spencer, and the "Khadafy" test suite by Scott Anderson. The tests come in three files: * runtests.c - The driver program. * TESTS - The list of test cases * TESTS2C.sed - A script to convert test cases into C. * Menu: * Running the Tests:: Invoking the test suite. * Adding New Tests:: Extending the test suite.  File: rx.info, Node: Running the Tests, Next: Adding New Tests, Prev: Semantic Tests, Up: Semantic Tests Running the Tests ----------------- To build the test program, use `make runtests'. (Detailed configuration and build instructions can be found in INSTALL). Invoked with no arguments, `runtests' runs all test-cases. For each test case, a sequence number is printed. If there is a problem with that case, more information is printed. Output like this: #0 #1 #2 ... indicates a successful run. (The output format is likely to change in the future to make it easier to automate testing of Rx.) With a single numeric argument, runtests executes just that test and no others. This is handy when debugging. Sometimes a bug will occur when all tests are run, but disappear when just the problematic test is run. Usually this has to do with memory or cache corruption.  File: rx.info, Node: Adding New Tests, Prev: Running the Tests, Up: Semantic Tests Adding New Tests ---------------- This list of tests used by `runtests' is found in the file `TESTS'. Each line of that file is a list of colon separated fields similar to: 1:^(ab|cd)e:abcde The first field is the expected return value of regexec, or '2' meaning that the pattern is not valid. The second field is the regular expression being tested. The third field is the string to which the pattern is to be compared. The file `TESTS2C.sed' is used to convert the test cases into initializers for an array of structures (which is automated in the Makefile). To add a new test case, add lines to `TESTS' and recompile `runtests'.  File: rx.info, Node: General Functions, Next: Rx Theory, Prev: Semantic Tests, Up: Top General Functions ================= Here is a whirlwind tour of the lower-level Rx entry points and data structures. This is not intended to be complete, but rather to be a help to anyone reading the code itself. These are presented in the same order in which they might typically be used. * Menu: * Compiling Expressions:: Strings->Syntax Trees * Posixly Correct Matches:: Checking for a Posixly correct match. * Hairy Matches:: Matching with multi-part strings. * NFA Functions:: Forget Posix, eat NFAs raw.  File: rx.info, Node: Compiling Expressions, Next: Posixly Correct Matches, Prev: General Functions, Up: General Functions Compiling Expressions --------------------- - Internal Function: reg_errcode_t rx_parse (struct rexp_node ** rexp_p, __const__ char *pattern, int size, unsigned long syntax, int cset_size, unsigned char *translate); Compile a regular expression. REXP_P will return a syntax tree for the pattern. See `"rxnode.h"' to learn about syntax trees. Minimally, you should know that syntax trees are reference counted using `rx_save_rexp' and `rx_free_rexp'. PATTERN and SIZE are the expression and its length. The expression need not be 0-terminated. The bits set in SYNTAX control details of the language understood by the parser. The meaning of each bit is described in `"rxgnucomp.h"'. CSET_SIZE is usually 256. It should not be much larger than that or space-performance will suffer badly. TRANSLATE is a translation table - an array of characters. Characters in the pattern are looked up in TRANSLATE as they are read. Additionally, the pattern is compiled presuming that the same translation table will be applied to characters in the string being searched (meaning that the pattern is compiled in such a way that the effect will be as if every character in the target string is looked up in the translation table, even though the expense of that lookup is usually avoided). A return of 0 indicates success, otherwise, the error can be looked up in the `rx_error_msg' array. - Internal Function: int rx_posix_analyze_rexp (struct rexp_node *** subexps, int * n_subexps, struct rexp_node * node, int id) Staticly analyze a regular expression in preparation for matching. This function fills in some fields of the nodes of a syntax tree. `subexps' and `n_subexps' return a malloced array of pointers into the syntax tree, one per parenthesized subexpression. The caller must eventually free this array. `node' is the pattern to be analyzed. The tree returned by `rx_parse' is suitable for this. `id' should be 1. Ignore the return value. This function is not robust in the case that `malloc' or `realloc' returns 0 (which should be fixed). You might expect that `rx_posix_analyze_rexp' and `rx_parse' should be combined. They are not because the actions of `rx_posix_analyze_rexp' make sense for any syntax tree, regardless of how it is constructed. `rx_parse' is just one way to build a syntax tree - it is subject to replacement and trees can be built without using a parser at all (c.f. `"rxnode.h"').  File: rx.info, Node: Posixly Correct Matches, Next: Hairy Matches, Prev: Compiling Expressions, Up: General Functions Checking for Posixly Correct Matches with One-Piece Strings ----------------------------------------------------------- Looking for a match that is Posixly correct is conceptually tricky. Here is one way to think of it: The Posix expression languages, absent of any consideration of the meaning of "leftmost-longest", underspecify the return value of `regexec'. Consider the example: Pattern: foo\(.*\)\(.*\)bar Target string: fooxxxbar Initially it is ambiguous how this should be matched. The ambiguity is exposed by considering the possible bindings for the subexpressions `\1' and `\2': Possible outcomes: \1 == "xxx" \2 == "" \1 == "xx" \2 == "x" \1 == "x" \2 == "xx" \1 == "" \2 == "xxx" Posix tells us that the first outcome is the prefered one if the illustrated pattern is used alone (this is a consequence of the "leftmost longest" rule). But Posix also says the answer may be different if the pattern is a sub-pattern of a larger pattern. For example: Pattern: foo\(.*\)\(.*\)bar\2 Target string: fooxxxbarxx Now the only acceptable outcome is: \1 == "x" \2 == "xx" Three entry points provide a relatively simple interface to this moderately complicated situation: - Internal Function: struct rx_solutions * rx_basic_make_solutions (struct rx_registers * regs, struct rexp_node * expression, struct rexp_node ** subexps, int start, int end, struct rx_context_rules * rules, char * str) - Internal Function: void rx_basic_free_solutions (struct rx_solutions * solns); - Internal Function: enum rx_answers rx_next_solution (struct rx_solutions * solns) - Internal Function: void rx_basic_free_solutions (struct rx_solutions * solns) `make_solutions' returns a lazilly constructed stream of possible solutions for a given regexp and target string. `next_solution' constructs and returns the next solution from a list returned by `make_solutions'. Solutions are returned in order of preferability according to the Posix spec. So, the first solution is the leftmost-longest, and the last, the rightmost-shortest. REGS is where subexpression extents are tracked. EXPRESSION is the expression to solve. SUBEXPS is as returned by `rx_posix_analyze_rexp'. START and END are the integer addresses of the data to be compared to the pattern. A solution must match all characters starting at START and ending at `END - 1'. RULES are some bits that control the precise meaning of "^" and "$". See `"rxcontext.h"'. STR is the data to be matched. Note that anchoring operators can cause the matcher to look beyond START and END when testing a match. It may make sense to pass START > 0, for example. `make_solutions' returns a pointer to a solution list, or 0 if an allocation fails. `next_solution' returns a status which can be: `rx_yes' -- The pattern matches exactly. `rx_maybe' -- The pattern does not match, but might if more characters were added to the string. `rx_no' -- The pattern does not match and could not be made to match even by adding more characters. In addition, REGS is filled in by `next_solution'. Finally, `free_solutions' is used to release resources consumed by a solution list.  File: rx.info, Node: Hairy Matches, Next: NFA Functions, Prev: Posixly Correct Matches, Up: General Functions Hairy Matches and Suspendable Searches -------------------------------------- The functions `rx_basic_make_solutions' and `rx_basic_free_solutions' are used to find matches in a one-piece string. More general entry points are provided for multi-piece strings, which may or may not be entirely available in memory at one time: - Internal Function: struct rx_solutions * rx_make_solutions (struct rx_registers * regs, struct rx_unfaniverse * verse, struct rexp_node * expression, struct rexp_node ** subexps, int cset_size, int start, int end, rx_vmfn vmfn, rx_contextfn contextfn, void * closure) - Internal Function: void rx_free_solutions (struct rx_solutions * solns) These functions are similar to their `_basic_' analogues, but substitute some new parameters for the one piece string: VMFN is a function provided by the caller that returns at least part of the string being matched. CONTEXTFN is a function provided by the caller that does the work of the backreference operators, "^", and "$". CLOSURE is data provided by the caller, passed through to vmfn and contextfn. typedef enum rx_answers (*rx_vmfn) P((void * closure, unsigned char ** burst, int * len, int * offset, int start, int end, int need)); typedef enum rx_answers (*rx_contextfn) P((void * closure, int type, int start, int end, struct rx_registers * regs)); `vmfn' is asked for a range of characters from START to END. NEED is in that range and is the only position that must be returned by vmfn. BURST, LEN, and OFFSET are output parameters. OFFSET will get the integer address of burst, and LEN the number of valid characters in the burst. VMFN may be asked for the empty string from END to END and should be able to handle that case. If `vmfn' provides the needed character, it should return `rx_yes'. If `vmfn' can't provide the character now, but might be able to later, it should return `rx_maybe'. This will cause `rx_next_solution' to quickly return `rx_maybe'. If `rx_next_solution' is called again, it will retry the call to vmfn and possibly resume the search. If `vmfn' can not provide the needed character, it should return `rx_no'. `rx_next_solution' will not normally ask for out-of-range characters, so usually there is no good reason to return `rx_no', but the code is supposed to be robust in case you do. The `contextfn' also returns a `yes/no/maybe' status. It tests whether the characters in START to END satisfy the operator TYPE. `type' is an ascii value; one of the characters in "^$123456789".  File: rx.info, Node: NFA Functions, Prev: Hairy Matches, Up: General Functions NFA Functions and Super-NFA Functions ------------------------------------- Not really documented yet. rx.h, rxnfa.h -- translating an expression into an NFA. rxsuper.h -- optimizing the NFA into a DFA, or at least an NFA with fewer branch points. rxanal.h -- how to use DFAs to find longest matches, matches of a specific length, the shortest match, etc. rxsimp.h -- Translating a Posix "regular expression" into a true regular expression that describes a superset language.  File: rx.info, Node: Rx Theory, Next: Improvements to Make, Prev: General Functions, Up: Top Rx Theory ========= There are two match algorithms. One is for truly regular regexps (those that can be reduced to a dfa). The other is for non-regular regexps. The dfa algorithm implements the idea suggested in `Compilers' by Aho, Sethi and Ullman: [One] approach [to pattern matching regular expressions] is to use a DFA, but avoid constructing all of the transition table by using a technique called "lazy transition evaluation". Here, transitions are computed at run time [when] actually needed. [T]ransitions are stored in a cache. [....] If the cache becomes full, we can erase some previously computed transition to make room for the new transition. The implementation in Rx is generalized from that, but the above description covers what is used for Posix patterns. The non-dfa algorithm implements a "recursive decomposition" technique described in email by Henry Spencer. For a given pattern, this algorithm first checks to see if a simpler, superset language, DFA-pattern matches. If it does, then this algorithm does the detail-work to see if the non-DFA pattern matches. The detail work usually involves recursing on subpatterns. For example, a concatentation of two subexpressions matches a string if the string can be divided into two parts, each matching one subexpression, in the right order. More than one solution is often possible for a given pattern. This ambiguity is the subject of the "leftmost longest" rules in the spec, and the back-tracking oriented stream-of-solution functions `rx_make_solutions', `rx_next_solution' and `rx_free_solutions'. rxspencer.[ch] -- The non-DFA algorithm rxanal.[ch] rxsuper.[ch] rxnfa.[ch] -- The DFA algorithm  File: rx.info, Node: Improvements to Make, Prev: Rx Theory, Up: Top Improvements to Make ==================== An approximation of the GNU entry points is needed. This can not easily be an exact approximation because the GNU entry points have no function for freeing a compiled expression - they assume that compiled expressions can be freed using a single call to "free(3)". For that reason, do not use the names of the old GNU entry points ...use "rx_" instead. This version of Rx hasn't been profiled and tuned. It might be nice to have some C++ types to hide reference counting and provide handy constructors for rexp_nodes and nfa parts. Right now, the regexp->regular regexp mapping in rxsimp.[ch] is cached for a particular input expression. I think it would probably be a big win in many situations if it were cached, like unique nfas (c.f. rxunfa.[ch]), for equivalent expressions. It would speed up matching to lift the non-regular constructs to the top of the syntax tree. e.g., if you have: pattern: "abc$" syntax tree: (concat (cset "a") (concat (cset "b") (concat (cset "c") (operator "$")))) it involves a lot more work (in "next_solution) than if you have the equivalent: same pattern: "abc$" syntax tree: (concat (concat (cset "a") (concat (cset "b") (cset "c"))) (operator "$")) in fact, while doing that, notice that subtrees of nothing-but-concat can conceivably be optimized so that next_solution just uses "strcmp". For that matter, a tree which is a disjunction of conjunction (a tree of "|" whose leaves are trees of concat nodes) could be optimized to use a table method (Boyer/Moore?). Again, this would be for next_solution but might involve adding stuff to rx_posix_analyze_rexp. As more and more optimizations are added, programmers should be given simple control over which optimizations are important for a particular pattern. For example, packages in emacs often include a few regexps that are used heavily - it would be worth the trouble to arrange for those regexps to be compiled specially. Static compilation of regexps is mostly a job for lex, but posix patterns are more general and have a nicer interface. What's more, it'd be nice to have the benefits of static compilation even for dynamicly loaded regexps (e.g., imagine if when an emacs lisp file is byte-compiled, the regexps it contains are also compiled). A lot of optimizations apply to just syntax trees. One approach to static compilation is to make an ugly but fast-reading syntax for trees, and a printer for that syntax. Then, after optimiztaions, a pattern could be stored as a string and later read back in, already optimized. A similar trick could be done for the nfa built in rxnfa.c. next_solution already considers the length of a fixed-length subexpression, but many more lenght optimizations are possible. I'd guess these will pay off well.  Tag Table: Node: Top83 Node: Posix Entry Points638 Node: POSIX Regexp Compilation1546 Node: Flags for POSIX Regexps5626 Node: Matching POSIX Regexps6530 Node: Regexp Subexpressions8688 Node: Subexpression Complications10740 Node: Regexp Cleanup13096 Node: Performance Considerations15419 Node: Avoiding Redundant Compilations15989 Node: Controlling Memory Use18120 Node: Semantic Tests18719 Node: Running the Tests19445 Node: Adding New Tests20369 Node: General Functions21121 Node: Compiling Expressions21775 Node: Posixly Correct Matches24531 Node: Hairy Matches28136 Node: NFA Functions31078 Node: Rx Theory31742 Node: Improvements to Make33591  End Tag Table dist/rx/COPYING.LIB100640 765 144 61261 6071532210 12475 0ustar lordusers GNU LIBRARY GENERAL PUBLIC LICENSE Version 2, June 1991 Copyright (C) 1991 Free Software Foundation, Inc. 675 Mass Ave, Cambridge, MA 02139, USA Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed. [This is the first released version of the library GPL. It is numbered 2 because it goes with version 2 of the ordinary GPL.] Preamble The licenses for most software are designed to take away your freedom to share and change it. By contrast, the GNU General Public Licenses are intended to guarantee your freedom to share and change free software--to make sure the software is free for all its users. This license, the Library General Public License, applies to some specially designated Free Software Foundation software, and to any other libraries whose authors decide to use it. You can use it for your libraries, too. When we speak of free software, we are referring to freedom, not price. Our General Public Licenses are designed to make sure that you have the freedom to distribute copies of free software (and charge for this service if you wish), that you receive source code or can get it if you want it, that you can change the software or use pieces of it in new free programs; and that you know you can do these things. To protect your rights, we need to make restrictions that forbid anyone to deny you these rights or to ask you to surrender the rights. These restrictions translate to certain responsibilities for you if you distribute copies of the library, or if you modify it. For example, if you distribute copies of the library, whether gratis or for a fee, you must give the recipients all the rights that we gave you. You must make sure that they, too, receive or can get the source code. If you link a program with the library, you must provide complete object files to the recipients so that they can relink them with the library, after making changes to the library and recompiling it. And you must show them these terms so they know their rights. Our method of protecting your rights has two steps: (1) copyright the library, and (2) offer you this license which gives you legal permission to copy, distribute and/or modify the library. Also, for each distributor's protection, we want to make certain that everyone understands that there is no warranty for this free library. If the library is modified by someone else and passed on, we want its recipients to know that what they have is not the original version, so that any problems introduced by others will not reflect on the original authors' reputations. Finally, any free program is threatened constantly by software patents. We wish to avoid the danger that companies distributing free software will individually obtain patent licenses, thus in effect transforming the program into proprietary software. To prevent this, we have made it clear that any patent must be licensed for everyone's free use or not licensed at all. Most GNU software, including some libraries, is covered by the ordinary GNU General Public License, which was designed for utility programs. This license, the GNU Library General Public License, applies to certain designated libraries. This license is quite different from the ordinary one; be sure to read it in full, and don't assume that anything in it is the same as in the ordinary license. The reason we have a separate public license for some libraries is that they blur the distinction we usually make between modifying or adding to a program and simply using it. Linking a program with a library, without changing the library, is in some sense simply using the library, and is analogous to running a utility program or application program. However, in a textual and legal sense, the linked executable is a combined work, a derivative of the original library, and the ordinary General Public License treats it as such. Because of this blurred distinction, using the ordinary General Public License for libraries did not effectively promote software sharing, because most developers did not use the libraries. We concluded that weaker conditions might promote sharing better. However, unrestricted linking of non-free programs would deprive the users of those programs of all benefit from the free status of the libraries themselves. This Library General Public License is intended to permit developers of non-free programs to use free libraries, while preserving your freedom as a user of such programs to change the free libraries that are incorporated in them. (We have not seen how to achieve this as regards changes in header files, but we have achieved it as regards changes in the actual functions of the Library.) The hope is that this will lead to faster development of free libraries. The precise terms and conditions for copying, distribution and modification follow. Pay close attention to the difference between a "work based on the library" and a "work that uses the library". The former contains code derived from the library, while the latter only works together with the library. Note that it is possible for a library to be covered by the ordinary General Public License rather than by this special one. GNU LIBRARY GENERAL PUBLIC LICENSE TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION 0. This License Agreement applies to any software library which contains a notice placed by the copyright holder or other authorized party saying it may be distributed under the terms of this Library General Public License (also called "this License"). Each licensee is addressed as "you". A "library" means a collection of software functions and/or data prepared so as to be conveniently linked with application programs (which use some of those functions and data) to form executables. The "Library", below, refers to any such software library or work which has been distributed under these terms. A "work based on the Library" means either the Library or any derivative work under copyright law: that is to say, a work containing the Library or a portion of it, either verbatim or with modifications and/or translated straightforwardly into another language. (Hereinafter, translation is included without limitation in the term "modification".) "Source code" for a work means the preferred form of the work for making modifications to it. For a library, complete source code means all the source code for all modules it contains, plus any associated interface definition files, plus the scripts used to control compilation and installation of the library. Activities other than copying, distribution and modification are not covered by this License; they are outside its scope. The act of running a program using the Library is not restricted, and output from such a program is covered only if its contents constitute a work based on the Library (independent of the use of the Library in a tool for writing it). Whether that is true depends on what the Library does and what the program that uses the Library does. 1. You may copy and distribute verbatim copies of the Library's complete source code as you receive it, in any medium, provided that you conspicuously and appropriately publish on each copy an appropriate copyright notice and disclaimer of warranty; keep intact all the notices that refer to this License and to the absence of any warranty; and distribute a copy of this License along with the Library. You may charge a fee for the physical act of transferring a copy, and you may at your option offer warranty protection in exchange for a fee. 2. You may modify your copy or copies of the Library or any portion of it, thus forming a work based on the Library, and copy and distribute such modifications or work under the terms of Section 1 above, provided that you also meet all of these conditions: a) The modified work must itself be a software library. b) You must cause the files modified to carry prominent notices stating that you changed the files and the date of any change. c) You must cause the whole of the work to be licensed at no charge to all third parties under the terms of this License. d) If a facility in the modified Library refers to a function or a table of data to be supplied by an application program that uses the facility, other than as an argument passed when the facility is invoked, then you must make a good faith effort to ensure that, in the event an application does not supply such function or table, the facility still operates, and performs whatever part of its purpose remains meaningful. (For example, a function in a library to compute square roots has a purpose that is entirely well-defined independent of the application. Therefore, Subsection 2d requires that any application-supplied function or table used by this function must be optional: if the application does not supply it, the square root function must still compute square roots.) These requirements apply to the modified work as a whole. If identifiable sections of that work are not derived from the Library, and can be reasonably considered independent and separate works in themselves, then this License, and its terms, do not apply to those sections when you distribute them as separate works. But when you distribute the same sections as part of a whole which is a work based on the Library, the distribution of the whole must be on the terms of this License, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it. Thus, it is not the intent of this section to claim rights or contest your rights to work written entirely by you; rather, the intent is to exercise the right to control the distribution of derivative or collective works based on the Library. In addition, mere aggregation of another work not based on the Library with the Library (or with a work based on the Library) on a volume of a storage or distribution medium does not bring the other work under the scope of this License. 3. You may opt to apply the terms of the ordinary GNU General Public License instead of this License to a given copy of the Library. To do this, you must alter all the notices that refer to this License, so that they refer to the ordinary GNU General Public License, version 2, instead of to this License. (If a newer version than version 2 of the ordinary GNU General Public License has appeared, then you can specify that version instead if you wish.) Do not make any other change in these notices. Once this change is made in a given copy, it is irreversible for that copy, so the ordinary GNU General Public License applies to all subsequent copies and derivative works made from that copy. This option is useful when you wish to copy part of the code of the Library into a program that is not a library. 4. You may copy and distribute the Library (or a portion or derivative of it, under Section 2) in object code or executable form under the terms of Sections 1 and 2 above provided that you accompany it with the complete corresponding machine-readable source code, which must be distributed under the terms of Sections 1 and 2 above on a medium customarily used for software interchange. If distribution of object code is made by offering access to copy from a designated place, then offering equivalent access to copy the source code from the same place satisfies the requirement to distribute the source code, even though third parties are not compelled to copy the source along with the object code. 5. A program that contains no derivative of any portion of the Library, but is designed to work with the Library by being compiled or linked with it, is called a "work that uses the Library". Such a work, in isolation, is not a derivative work of the Library, and therefore falls outside the scope of this License. However, linking a "work that uses the Library" with the Library creates an executable that is a derivative of the Library (because it contains portions of the Library), rather than a "work that uses the library". The executable is therefore covered by this License. Section 6 states terms for distribution of such executables. When a "work that uses the Library" uses material from a header file that is part of the Library, the object code for the work may be a derivative work of the Library even though the source code is not. Whether this is true is especially significant if the work can be linked without the Library, or if the work is itself a library. The threshold for this to be true is not precisely defined by law. If such an object file uses only numerical parameters, data structure layouts and accessors, and small macros and small inline functions (ten lines or less in length), then the use of the object file is unrestricted, regardless of whether it is legally a derivative work. (Executables containing this object code plus portions of the Library will still fall under Section 6.) Otherwise, if the work is a derivative of the Library, you may distribute the object code for the work under the terms of Section 6. Any executables containing that work also fall under Section 6, whether or not they are linked directly with the Library itself. 6. As an exception to the Sections above, you may also compile or link a "work that uses the Library" with the Library to produce a work containing portions of the Library, and distribute that work under terms of your choice, provided that the terms permit modification of the work for the customer's own use and reverse engineering for debugging such modifications. You must give prominent notice with each copy of the work that the Library is used in it and that the Library and its use are covered by this License. You must supply a copy of this License. If the work during execution displays copyright notices, you must include the copyright notice for the Library among them, as well as a reference directing the user to the copy of this License. Also, you must do one of these things: a) Accompany the work with the complete corresponding machine-readable source code for the Library including whatever changes were used in the work (which must be distributed under Sections 1 and 2 above); and, if the work is an executable linked with the Library, with the complete machine-readable "work that uses the Library", as object code and/or source code, so that the user can modify the Library and then relink to produce a modified executable containing the modified Library. (It is understood that the user who changes the contents of definitions files in the Library will not necessarily be able to recompile the application to use the modified definitions.) b) Accompany the work with a written offer, valid for at least three years, to give the same user the materials specified in Subsection 6a, above, for a charge no more than the cost of performing this distribution. c) If distribution of the work is made by offering access to copy from a designated place, offer equivalent access to copy the above specified materials from the same place. d) Verify that the user has already received a copy of these materials or that you have already sent this user a copy. For an executable, the required form of the "work that uses the Library" must include any data and utility programs needed for reproducing the executable from it. However, as a special exception, the source code distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable. It may happen that this requirement contradicts the license restrictions of other proprietary libraries that do not normally accompany the operating system. Such a contradiction means you cannot use both them and the Library together in an executable that you distribute. 7. You may place library facilities that are a work based on the Library side-by-side in a single library together with other library facilities not covered by this License, and distribute such a combined library, provided that the separate distribution of the work based on the Library and of the other library facilities is otherwise permitted, and provided that you do these two things: a) Accompany the combined library with a copy of the same work based on the Library, uncombined with any other library facilities. This must be distributed under the terms of the Sections above. b) Give prominent notice with the combined library of the fact that part of it is a work based on the Library, and explaining where to find the accompanying uncombined form of the same work. 8. You may not copy, modify, sublicense, link with, or distribute the Library except as expressly provided under this License. Any attempt otherwise to copy, modify, sublicense, link with, or distribute the Library is void, and will automatically terminate your rights under this License. However, parties who have received copies, or rights, from you under this License will not have their licenses terminated so long as such parties remain in full compliance. 9. You are not required to accept this License, since you have not signed it. However, nothing else grants you permission to modify or distribute the Library or its derivative works. These actions are prohibited by law if you do not accept this License. Therefore, by modifying or distributing the Library (or any work based on the Library), you indicate your acceptance of this License to do so, and all its terms and conditions for copying, distributing or modifying the Library or works based on it. 10. Each time you redistribute the Library (or any work based on the Library), the recipient automatically receives a license from the original licensor to copy, distribute, link with or modify the Library subject to these terms and conditions. You may not impose any further restrictions on the recipients' exercise of the rights granted herein. You are not responsible for enforcing compliance by third parties to this License. 11. If, as a consequence of a court judgment or allegation of patent infringement or for any other reason (not limited to patent issues), conditions are imposed on you (whether by court order, agreement or otherwise) that contradict the conditions of this License, they do not excuse you from the conditions of this License. If you cannot distribute so as to satisfy simultaneously your obligations under this License and any other pertinent obligations, then as a consequence you may not distribute the Library at all. For example, if a patent license would not permit royalty-free redistribution of the Library by all those who receive copies directly or indirectly through you, then the only way you could satisfy both it and this License would be to refrain entirely from distribution of the Library. If any portion of this section is held invalid or unenforceable under any particular circumstance, the balance of the section is intended to apply, and the section as a whole is intended to apply in other circumstances. It is not the purpose of this section to induce you to infringe any patents or other property right claims or to contest validity of any such claims; this section has the sole purpose of protecting the integrity of the free software distribution system which is implemented by public license practices. Many people have made generous contributions to the wide range of software distributed through that system in reliance on consistent application of that system; it is up to the author/donor to decide if he or she is willing to distribute software through any other system and a licensee cannot impose that choice. This section is intended to make thoroughly clear what is believed to be a consequence of the rest of this License. 12. If the distribution and/or use of the Library is restricted in certain countries either by patents or by copyrighted interfaces, the original copyright holder who places the Library under this License may add an explicit geographical distribution limitation excluding those countries, so that distribution is permitted only in or among countries not thus excluded. In such case, this License incorporates the limitation as if written in the body of this License. 13. The Free Software Foundation may publish revised and/or new versions of the Library General Public License from time to time. Such new versions will be similar in spirit to the present version, but may differ in detail to address new problems or concerns. Each version is given a distinguishing version number. If the Library specifies a version number of this License which applies to it and "any later version", you have the option of following the terms and conditions either of that version or of any later version published by the Free Software Foundation. If the Library does not specify a license version number, you may choose any version ever published by the Free Software Foundation. 14. If you wish to incorporate parts of the Library into other free programs whose distribution conditions are incompatible with these, write to the author to ask for permission. For software which is copyrighted by the Free Software Foundation, write to the Free Software Foundation; we sometimes make exceptions for this. Our decision will be guided by the two goals of preserving the free status of all derivatives of our free software and of promoting the sharing and reuse of software generally. NO WARRANTY 15. BECAUSE THE LIBRARY IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY FOR THE LIBRARY, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE LIBRARY "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE LIBRARY IS WITH YOU. SHOULD THE LIBRARY PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION. 16. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR REDISTRIBUTE THE LIBRARY AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE LIBRARY (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE LIBRARY TO OPERATE WITH ANY OTHER SOFTWARE), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. END OF TERMS AND CONDITIONS Appendix: How to Apply These Terms to Your New Libraries If you develop a new library, and you want it to be of the greatest possible use to the public, we recommend making it free software that everyone can redistribute and change. You can do so by permitting redistribution under these terms (or, alternatively, under the terms of the ordinary General Public License). To apply these terms, attach the following notices to the library. It is safest to attach them to the start of each source file to most effectively convey the exclusion of warranty; and each file should have at least the "copyright" line and a pointer to where the full notice is found. Copyright (C) This library is free software; you can redistribute it and/or modify it under the terms of the GNU Library General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Library General Public License for more details. You should have received a copy of the GNU Library General Public License along with this library; if not, write to the Free Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. Also add information on how to contact you by electronic and paper mail. You should also get your employer (if you work as a programmer) or your school, if any, to sign a "copyright disclaimer" for the library, if necessary. Here is a sample; alter the names: Yoyodyne, Inc., hereby disclaims all copyright interest in the library `Frob' (a library for tweaking knobs) written by James Random Hacker. , 1 April 1990 Ty Coon, President of Vice That's all there is to it! dist/rx/ChangeLog100660 765 144 175 6071416745 12564 0ustar lordusersSat Dec 30 21:19:31 1995 Tom Lord * *.[ch]: posixification and algorithmic improvement (thanks henry!). dist/rx/PLUGIN/ 42770 765 144 0 6071542634 11726 5ustar lordusersdist/rx/PLUGIN/guile.inits100660 765 144 33 6003763267 14135 0ustar lordusersgscm_init_rx gscm_init_rx dist/rx/PLUGIN/guile.config100660 765 144 43 6035047712 14250 0ustar lordusersxtra_cflags="-L../rx $xtra_cflags" dist/rx/PLUGIN/guile.libs100660 765 144 17 6003763412 13732 0ustar lordusers-lrxds -lguile dist/rx/PLUGIN/greet.~1~100660 765 144 1726 6071147064 13612 0ustar lordusersxxxxxxxxxxxxxxxx Rx/diSpencer xxxxxxxxxxxxxxxx xxxxxxxxxxxxxxx POSIX regexp functions. xxxxx xxx xxxxx xxx xxxxx xxx Let it fly in the breeze xxxxxxxxxxxxxxx and get caught in the trees xxxxxxxxxxxxxx make a home for the fleas in my Hair. xxxxxxxxx A home for fleas [oh yeah!]; xxxxx xxxx A hive for bees [good god!]; xxxxx xxxx A nest for birds; xxxxx xxxx rrr There ain't no words xxxxx xxxx rrr for the beauty xxxxx xxxxrrr the splender xxxxxxx rxrxr the wonder rrrxrx of my Hair! rrr rxxx rrr xxxx Blow it. Show it. Looongest I can grow it, my Hair. rrr xxxx xx rrr xxxxxx dist/rx/PLUGIN/OPT100660 765 144 25 6035047675 12351 0ustar lorduserslibguile rx rx guile dist/rx/PLUGIN/greet100660 765 144 1722 6071542634 13055 0ustar lordusersxxxxxxxxxxxxxxxx Rx/diSpencer xxxxxxxxxxxxxxxx xxxxxxxxxxxxxxx POSIX regexp functions. xxxxx xxx xxxxx xxx xxxxx xxx Let it fly in the breeze xxxxxxxxxxxxxxx and get caught in the trees xxxxxxxxxxxxxx make a home for the fleas in my Hair. xxxxxxxxx A home for fleas [oh yeah!]; xxxxx xxxx A hive for bees [good god!]; xxxxx xxxx A nest for birds; xxxxx xxxx rrr There ain't no words xxxxx xxxx rrr for the beauty xxxxx xxxxrrr the splender xxxxxxx rxrxr the wonder rrrxrx of my Hair! rrr rxxx rrr xxxx Blow it. Show it. rrr xxxx xx Looongest I can grow it, my Hair. rrr xxxxxx dist/rx/TESTS100660 765 144 10722 6070131313 11656 0ustar lordusers0:(.*)*\1:xx 0:^: 0:$: 0:^$: 0:^a$:a 0:abc:abc 1:abc:xbc 1:abc:axc 1:abc:abx 0:abc:xabcy 0:abc:ababc 0:ab*c:abc 0:ab*bc:abc 0:ab*bc:abbc 0:ab*bc:abbbbc 0:ab+bc:abbc 1:ab+bc:abc 1:ab+bc:abq 0:ab+bc:abbbbc 0:ab?bc:abbc 0:ab?bc:abc 1:ab?bc:abbbbc 0:ab?c:abc 0:^abc$:abc 1:^abc$:abcc 0:^abc:abcc 1:^abc$:aabc 0:abc$:aabc 0:^:abc 0:$:abc 0:a.c:abc 0:a.c:axc 0:a.*c:axyzc 1:a.*c:axyzd 1:a[bc]d:abc 0:a[bc]d:abd 1:a[b-d]e:abd 0:a[b-d]e:ace 0:a[b-d]:aac 0:a[-b]:a- 0:a[b-]:a- 1:a[b-a]:- 2:a[]b:- 2:a[:- 0:a]:a] 0:a[]]b:a]b 0:a[^bc]d:aed 1:a[^bc]d:abd 0:a[^-b]c:adc 1:a[^-b]c:a-c 1:a[^]b]c:a]c 0:a[^]b]c:adc 0:ab|cd:abc 0:ab|cd:abcd 0:()ef:def 0:()*:- 1:*a:- 0:^*:- 0:$*:- 1:(*)b:- 1:$b:b 2:a\:- 0:a\(b:a(b 0:a\(*b:ab 0:a\(*b:a((b 1:a\x:a\x 1:abc):- 2:(abc:- 0:((a)):abc 0:(a)b(c):abc 0:a+b+c:aabbabc 0:a**:- 0:a*?:- 0:(a*)*:- 0:(a*)+:- 0:(a|)*:- 0:(a*|b)*:- 0:(a+|b)*:ab 0:(a+|b)+:ab 0:(a+|b)?:ab 0:[^ab]*:cde 0:(^)*:- 0:(ab|)*:- 2:)(:- 1:abc: 1:abc: 0:a*: 0:([abc])*d:abbbcd 0:([abc])*bcd:abcd 0:a|b|c|d|e:e 0:(a|b|c|d|e)f:ef 0:((a*|b))*:- 0:abcd*efg:abcdefg 0:ab*:xabyabbbz 0:ab*:xayabbbz 0:(ab|cd)e:abcde 0:[abhgefdc]ij:hij 1:^(ab|cd)e:abcde 0:(abc|)ef:abcdef 0:(a|b)c*d:abcd 0:(ab|ab*)bc:abc 0:a([bc]*)c*:abc 0:a([bc]*)(c*d):abcd 0:a([bc]+)(c*d):abcd 0:a([bc]*)(c+d):abcd 0:a[bcd]*dcdcde:adcdcde 1:a[bcd]+dcdcde:adcdcde 0:(ab|a)b*c:abc 0:((a)(b)c)(d):abcd 0:[A-Za-z_][A-Za-z0-9_]*:alpha 0:^a(bc+|b[eh])g|.h$:abh 0:(bc+d$|ef*g.|h?i(j|k)):effgz 0:(bc+d$|ef*g.|h?i(j|k)):ij 1:(bc+d$|ef*g.|h?i(j|k)):effg 1:(bc+d$|ef*g.|h?i(j|k)):bcdd 0:(bc+d$|ef*g.|h?i(j|k)):reffgz 1:((((((((((a)))))))))):- 0:(((((((((a))))))))):a 1:multiple words of text:uh-uh 0:multiple words:multiple words, yeah 0:(.*)c(.*):abcde 1:\((.*),:(.*)\) 1:[k]:ab 0:abcd:abcd 0:a(bc)d:abcd 0:a[-]?c:ac 0:(....).*\1:beriberi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Muammar Qaddafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Mo'ammar Gadhafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Muammar Kaddafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Muammar Qadhafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Moammar El Kadhafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Muammar Gadafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Mu'ammar al-Qadafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Moamer El Kazzafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Moamar al-Gaddafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Mu'ammar Al Qathafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Muammar Al Qathafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Mo'ammar el-Gadhafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Moamar El Kadhafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Muammar al-Qadhafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Mu'ammar al-Qadhdhafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Mu'ammar Qadafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Moamar Gaddafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Mu'ammar Qadhdhafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Muammar Khaddafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Muammar al-Khaddafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Mu'amar al-Kadafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Muammar Ghaddafy 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Muammar Ghadafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Muammar Ghaddafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Muamar Kaddafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Muammar Quathafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Muammar Gheddafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Muamar Al-Kaddafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Moammar Khadafy 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Moammar Qudhafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Mu'ammar al-Qaddafi 0:M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]:Mulazim Awwal Mu'ammar Muhammad Abu Minyar al-Qadhafi dist/rx/TESTS2C.sed100660 765 144 70 6034016203 12530 0ustar lorduserss/\\/\\\\/g s/\(.*\):\(.*\):\(.*\)/ {\1, "\2", "\3"},/ dist/sed/COPYING100660 765 144 43112 6071414371 12216 0ustar lordusers GNU GENERAL PUBLIC LICENSE Version 2, June 1991 Copyright (C) 1989, 1991, 1992, 1993 Free Software Foundation, Inc. 675 Mass Ave, Cambridge, MA 02139, USA Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed. Preamble The licenses for most software are designed to take away your freedom to share and change it. By contrast, the GNU General Public License is intended to guarantee your freedom to share and change free software--to make sure the software is free for all its users. This General Public License applies to most of the Free Software Foundation's software and to any other program whose authors commit to using it. (Some other Free Software Foundation software is covered by the GNU Library General Public License instead.) You can apply it to your programs, too. When we speak of free software, we are referring to freedom, not price. Our General Public Licenses are designed to make sure that you have the freedom to distribute copies of free software (and charge for this service if you wish), that you receive source code or can get it if you want it, that you can change the software or use pieces of it in new free programs; and that you know you can do these things. To protect your rights, we need to make restrictions that forbid anyone to deny you these rights or to ask you to surrender the rights. These restrictions translate to certain responsibilities for you if you distribute copies of the software, or if you modify it. For example, if you distribute copies of such a program, whether gratis or for a fee, you must give the recipients all the rights that you have. You must make sure that they, too, receive or can get the source code. And you must show them these terms so they know their rights. We protect your rights with two steps: (1) copyright the software, and (2) offer you this license which gives you legal permission to copy, distribute and/or modify the software. Also, for each author's protection and ours, we want to make certain that everyone understands that there is no warranty for this free software. If the software is modified by someone else and passed on, we want its recipients to know that what they have is not the original, so that any problems introduced by others will not reflect on the original authors' reputations. Finally, any free program is threatened constantly by software patents. We wish to avoid the danger that redistributors of a free program will individually obtain patent licenses, in effect making the program proprietary. To prevent this, we have made it clear that any patent must be licensed for everyone's free use or not licensed at all. The precise terms and conditions for copying, distribution and modification follow. GNU GENERAL PUBLIC LICENSE TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION 0. This License applies to any program or other work which contains a notice placed by the copyright holder saying it may be distributed under the terms of this General Public License. The "Program", below, refers to any such program or work, and a "work based on the Program" means either the Program or any derivative work under copyright law: that is to say, a work containing the Program or a portion of it, either verbatim or with modifications and/or translated into another language. (Hereinafter, translation is included without limitation in the term "modification".) Each licensee is addressed as "you". Activities other than copying, distribution and modification are not covered by this License; they are outside its scope. The act of running the Program is not restricted, and the output from the Program is covered only if its contents constitute a work based on the Program (independent of having been made by running the Program). Whether that is true depends on what the Program does. 1. You may copy and distribute verbatim copies of the Program's source code as you receive it, in any medium, provided that you conspicuously and appropriately publish on each copy an appropriate copyright notice and disclaimer of warranty; keep intact all the notices that refer to this License and to the absence of any warranty; and give any other recipients of the Program a copy of this License along with the Program. You may charge a fee for the physical act of transferring a copy, and you may at your option offer warranty protection in exchange for a fee. 2. You may modify your copy or copies of the Program or any portion of it, thus forming a work based on the Program, and copy and distribute such modifications or work under the terms of Section 1 above, provided that you also meet all of these conditions: a) You must cause the modified files to carry prominent notices stating that you changed the files and the date of any change. b) You must cause any work that you distribute or publish, that in whole or in part contains or is derived from the Program or any part thereof, to be licensed as a whole at no charge to all third parties under the terms of this License. c) If the modified program normally reads commands interactively when run, you must cause it, when started running for such interactive use in the most ordinary way, to print or display an announcement including an appropriate copyright notice and a notice that there is no warranty (or else, saying that you provide a warranty) and that users may redistribute the program under these conditions, and telling the user how to view a copy of this License. (Exception: if the Program itself is interactive but does not normally print such an announcement, your work based on the Program is not required to print an announcement.) These requirements apply to the modified work as a whole. If identifiable sections of that work are not derived from the Program, and can be reasonably considered independent and separate works in themselves, then this License, and its terms, do not apply to those sections when you distribute them as separate works. But when you distribute the same sections as part of a whole which is a work based on the Program, the distribution of the whole must be on the terms of this License, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it. Thus, it is not the intent of this section to claim rights or contest your rights to work written entirely by you; rather, the intent is to exercise the right to control the distribution of derivative or collective works based on the Program. In addition, mere aggregation of another work not based on the Program with the Program (or with a work based on the Program) on a volume of a storage or distribution medium does not bring the other work under the scope of this License. 3. You may copy and distribute the Program (or a work based on it, under Section 2) in object code or executable form under the terms of Sections 1 and 2 above provided that you also do one of the following: a) Accompany it with the complete corresponding machine-readable source code, which must be distributed under the terms of Sections 1 and 2 above on a medium customarily used for software interchange; or, b) Accompany it with a written offer, valid for at least three years, to give any third party, for a charge no more than your cost of physically performing source distribution, a complete machine-readable copy of the corresponding source code, to be distributed under the terms of Sections 1 and 2 above on a medium customarily used for software interchange; or, c) Accompany it with the information you received as to the offer to distribute corresponding source code. (This alternative is allowed only for noncommercial distribution and only if you received the program in object code or executable form with such an offer, in accord with Subsection b above.) The source code for a work means the preferred form of the work for making modifications to it. For an executable work, complete source code means all the source code for all modules it contains, plus any associated interface definition files, plus the scripts used to control compilation and installation of the executable. However, as a special exception, the source code distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable. If distribution of executable or object code is made by offering access to copy from a designated place, then offering equivalent access to copy the source code from the same place counts as distribution of the source code, even though third parties are not compelled to copy the source along with the object code. 4. You may not copy, modify, sublicense, or distribute the Program except as expressly provided under this License. Any attempt otherwise to copy, modify, sublicense or distribute the Program is void, and will automatically terminate your rights under this License. However, parties who have received copies, or rights, from you under this License will not have their licenses terminated so long as such parties remain in full compliance. 5. You are not required to accept this License, since you have not signed it. However, nothing else grants you permission to modify or distribute the Program or its derivative works. These actions are prohibited by law if you do not accept this License. Therefore, by modifying or distributing the Program (or any work based on the Program), you indicate your acceptance of this License to do so, and all its terms and conditions for copying, distributing or modifying the Program or works based on it. 6. Each time you redistribute the Program (or any work based on the Program), the recipient automatically receives a license from the original licensor to copy, distribute or modify the Program subject to these terms and conditions. You may not impose any further restrictions on the recipients' exercise of the rights granted herein. You are not responsible for enforcing compliance by third parties to this License. 7. If, as a consequence of a court judgment or allegation of patent infringement or for any other reason (not limited to patent issues), conditions are imposed on you (whether by court order, agreement or otherwise) that contradict the conditions of this License, they do not excuse you from the conditions of this License. If you cannot distribute so as to satisfy simultaneously your obligations under this License and any other pertinent obligations, then as a consequence you may not distribute the Program at all. For example, if a patent license would not permit royalty-free redistribution of the Program by all those who receive copies directly or indirectly through you, then the only way you could satisfy both it and this License would be to refrain entirely from distribution of the Program. If any portion of this section is held invalid or unenforceable under any particular circumstance, the balance of the section is intended to apply and the section as a whole is intended to apply in other circumstances. It is not the purpose of this section to induce you to infringe any patents or other property right claims or to contest validity of any such claims; this section has the sole purpose of protecting the integrity of the free software distribution system, which is implemented by public license practices. Many people have made generous contributions to the wide range of software distributed through that system in reliance on consistent application of that system; it is up to the author/donor to decide if he or she is willing to distribute software through any other system and a licensee cannot impose that choice. This section is intended to make thoroughly clear what is believed to be a consequence of the rest of this License. 8. If the distribution and/or use of the Program is restricted in certain countries either by patents or by copyrighted interfaces, the original copyright holder who places the Program under this License may add an explicit geographical distribution limitation excluding those countries, so that distribution is permitted only in or among countries not thus excluded. In such case, this License incorporates the limitation as if written in the body of this License. 9. The Free Software Foundation may publish revised and/or new versions of the General Public License from time to time. Such new versions will be similar in spirit to the present version, but may differ in detail to address new problems or concerns. Each version is given a distinguishing version number. If the Program specifies a version number of this License which applies to it and "any later version", you have the option of following the terms and conditions either of that version or of any later version published by the Free Software Foundation. If the Program does not specify a version number of this License, you may choose any version ever published by the Free Software Foundation. 10. If you wish to incorporate parts of the Program into other free programs whose distribution conditions are different, write to the author to ask for permission. For software which is copyrighted by the Free Software Foundation, write to the Free Software Foundation; we sometimes make exceptions for this. Our decision will be guided by the two goals of preserving the free status of all derivatives of our free software and of promoting the sharing and reuse of software generally. NO WARRANTY 11. BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION. 12. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. END OF TERMS AND CONDITIONS Appendix: How to Apply These Terms to Your New Programs If you develop a new program, and you want it to be of the greatest possible use to the public, the best way to achieve this is to make it free software which everyone can redistribute and change under these terms. To do so, attach the following notices to the program. It is safest to attach them to the start of each source file to most effectively convey the exclusion of warranty; and each file should have at least the "copyright" line and a pointer to where the full notice is found. Copyright (C) 19yy This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. Also add information on how to contact you by electronic and paper mail. If the program is interactive, make it output a short notice like this when it starts in an interactive mode: Gnomovision version 69, Copyright (C) 19yy name of author Gnomovision comes with ABSOLUTELY NO WARRANTY; for details type `show w'. This is free software, and you are welcome to redistribute it under certain conditions; type `show c' for details. The hypothetical commands `show w' and `show c' should show the appropriate parts of the General Public License. Of course, the commands you use may be called something other than `show w' and `show c'; they could even be mouse-clicks or menu items--whatever suits your program. You should also get your employer (if you work as a programmer) or your school, if any, to sign a "copyright disclaimer" for the program, if necessary. Here is a sample; alter the names: Yoyodyne, Inc., hereby disclaims all copyright interest in the program `Gnomovision' (which makes passes at compilers) written by James Hacker. , 1 April 1989 Ty Coon, President of Vice This General Public License does not permit incorporating your program into proprietary programs. If your program is a subroutine library, you may consider it more useful to permit linking proprietary applications with the library. If this is what you want to do, use the GNU Library General Public License instead of this License. dist/sed/News100660 765 144 624 6071415002 11753 0ustar lordusersSed 3.0 This version of sed passes the new test-suite donated by Jason Molenda. Overall performance has been improved in the following sense: Sed 3.0 is often slightly slower than sed 2.05. On a few scripts, though, sed 2.05 was so slow as to be nearly useless or to use up unreasonable amounts of memory. These problems have been fixed and in such cases, sed 3.0 should have acceptable performance. dist/sed/ChangeLog100660 765 144 34104 6071407571 12742 0ustar lordusers------ sed 3.00 ------ Sat Dec 30 20:16:59 1995 Tom Lord * sed.c: Use posix entry points to regexp functions. Fix enough bugs to pass the test-suite. ....... Jason Molenda * testsuite/: trippy test suite. ------ sed 2.05 ------ Wed May 11 07:46:24 1994 Chip Salzenberg (chip@fin.uucp) * sed.c (compile_address): Recognize numeric addresses. Fixes typo made during installation of "~" feature. ------ sed 2.04 ------ Sat Apr 30 17:17:38 1994 Tom Lord (lord@x1.cygnus.com) * sed.c: applied a patch from From: kap1@tao.cpe.uchicago.edu (Dietrich Kappe) Dietrich writes: As my contribution to the creeping feature creature in sed, here is a new type of address. The address has form n~m, which means "the line number is equal to n modulo m." The modifications to sed are trivial, and the general usefulness of this address should be obvious. If m is 0 or missing, 1 is used in its place (could be a bug or a feature :-). Sat Apr 30 17:17:38 1994 Tom Lord (lord@x1.cygnus.com) * rx.c (solve_destination): protect `solution' more carefully. This is a cleanup of a patch from Kevin Buettner (kev@cujo.geg.mot.com). Sat Apr 30 17:17:38 1994 Tom Lord (lord@x1.cygnus.com) * rx.c: make translation tables unsigned chars * sed.c (main): Compile accumulated -e commands as soon as a -f command comes along. This ensures that the commands are executed in the right order. Mon Oct 25 14:41:47 1993 Tom Lord (lord@rtl.cygnus.com) * sed.c (execute_program): 'w' flushes the buffer after it writes -- diagnosed by doug@research.att.com. 'r' and 'w' to the same file is now supported -- hopefully even in a way that satisfies Posix (it now behaves differently from some /bin/sed's and the spec is hard to read so i'm not sure). Also, 'r' of a non-existent file is now permitted. Mon Oct 11 21:06:10 1993 Tom Lord (lord@cygnus.com) * sed.c (execute_program): remember that 'b' and 't' are more like longjmp than goto. Patch from tom@basil.icce.rug.nl (Tom R.Hageman) * rx.c: patch from From: fin!chip@rutgers.edu (Chip Salzenberg) to get rid of compiler warnings. ------ sed 2.03 ------ Sat Aug 7 01:04:59 1993 Tom Lord (lord@unix7.andrew.cmu.edu) * sed.c (compile_regex): report error messages for bogus regexps. SEE ALSO: ChangeLog.rx ------ no version information ------ Wed Jul 21 00:28:03 1993 Tom Lord (lord@unix8.andrew.cmu.edu) * alloca.c: upgraded to a more recent version * rx.c (re_search_2): prefer matches with longer subexpressions to those with shorter ones, giving precedence to low numbered subexpressions. * rx.c (re_compile): don't free `params' if its null. Fri Jul 16 01:12:08 1993 Tom Lord (lord@unix8.andrew.cmu.edu) * rx.[ch], sed.c: rx replaces regex. Thu May 27 11:13:03 1993 Tom Lord (lord@unix3.andrew.cmu.edu) * sed.c (execute_program, match_addr): caught more cases that need to be sensitive to a missing \n at EOF. Fri May 21 00:39:22 1993 Tom Lord (lord@unix8.andrew.cmu.edu) * sed.c (execute_program): apply gaumondp's patch to fix '\xabcxs/foo/bar/'. * sed.c (execute_program): If a second address is a regexp, never match it on the same line as the first address. * sed.c (compile_regexp): Numeric ranges x,y s.t. y < x are now treated as x,x. There was a bug in that they were being handled like x,x+1. * sed.c (execute_program, read_pattern_space, append_pattern_space) don't add newlines to lines that don't have them. Wed May 19 13:34:45 1993 Tom Lord (lord@unix9.andrew.cmu.edu) * sed.c (compile_program): grok \\n in comments. Mon May 17 16:34:50 1993 Tom Lord (lord@unix9.andrew.cmu.edu) * alloca.c: new (standard) file * configure.in: AC_CONSTified * sed.c (compile_program): properly diagnose the error of a missing command (e.g. sed /x/). (thanks gaumondp) * sed.c (compile_regexp): handle character classes correctly. Thanks gaumondp@ERE.UMontreal.CA and schwab@issan.informatik.uni-dortmund.de. Thu May 6 12:37:18 1993 Tom Lord (lord@unix10.andrew.cmu.edu) * sed.c (compile_filename, execute_program): don't use `access' or `/dev/null'. * sed.c (execute_program): 'N' at EOF should delete the pat buf. * sed.c (compile_filename): truncate, don't append files being openned for `w' or `s///w' * sed.c (execute_program): -n switch shouldn't effect `i' or `c'. * sed.c (compile_program): don't compile unescaped newlines into the substitution string of an `s' command (they are an error). * sed.c (compile_regex): correctly skip over character sets that contain `]'. * sed.c (execute_program): patch from gaumondp Correctly handle empty-string matches in the case of an `s' command with a repeat count. * sed.c (compile_program): patch from gaumondp@ere.UMontreal.ca. Don't consume characters after the label of a `b', `t' or `:' command. * sed.c (compile_program): unmatched open braces are an error. * sed.c (compile_file): when consuming an initial comment, count lines correctly. Wed Nov 18 02:10:58 1992 Tom Lord (lord@unix2.andrew.cmu.edu) * sed.c (execute_program): Made s///p print even if -n was specified. * sed.c (compile_string): Changed the type of this function to fix a compile warning. Wed Nov 4 17:15:34 1992 Tom Lord (lord@unix7.andrew.cmu.edu) * sed.c (main): Initialize the hold area to contain "\n" instead of "". In execute_program, all lines are expected to be newline terminated. Also, if H is the first command in the script, the result is a pattern buffer that begins with a blank line. Thanks to pinard@iro.umontreal.ca (Francois Pinard) for pointing out this and many other bugs. * sed.c (execute_program): Fixed a case of `D' command. Thanks Chris Weber * sed.c: added new tests of no_default_output to make -n work. Thanks Andrew Herbert * sed.c, configure.in,Makefile.in: autoconfed bcopy and const. Thanks "J.T. Conklin" * sed.c: made prog_cur, prog_start, and prog_end unsigned so that users could write `sed -e s/ÿ/foo/g'. Tue Oct 13 00:04:05 1992 Tom Lord (lord@unix3.andrew.cmu.edu) * sed.c (execute_program): fixed the cycling behavior of 'D' * sed.c: integrated patch that closes files * sed.c: changed regexp syntax Fri May 22 15:11:12 1992 Tom Lord (lord at moriarty.bh.andrew.cmu.edu) * regex.c: this is not my change, but a pointer to the fact that karl@gnu fixed some regexp bugs that were plaguing sed. Thu Apr 30 13:02:21 1992 Tom Lord (lord at unix3.andrew.cmu.edu) * sed.c (compile_program, execute_program) subprograms are now compiled with an explicit continuation ;) return_v and return_i in struct vector. execute_program no longer recurses to execute subprograms (case '{') and now understands a return instruction (case '{'). Tue Apr 28 17:13:04 1992 Tom Lord (lord at unix7.andrew.cmu.edu) * sed.c (compile_address) added \?regexp? syntax for addresses. * sed.c (main) added {} intervals to the obscure regexp syntax. * sed.c (compile_program) after calling compile_address, normalize numeric addresses (make a2.addr_number > a1.addr_number). This is necessary because line numbers must match exactly, but sed does not try to match a2 until after a1 has matched, yet a1,a2 where a2 <= a1 is defined to be equivelent to a1,a1+1 Sat Feb 29 10:55:54 1992 David J. MacKenzie (djm@nutrimat) * sed.c (usage): Document long options as starting with `--'. Mon Dec 9 23:56:40 1991 David J. MacKenzie (djm at wookumz.gnu.ai.mit.edu) * sed.c: Include sys/types.h, for new regex.h. Tue Nov 5 02:16:01 1991 David J. MacKenzie (djm at wookumz.gnu.ai.mit.edu) * utils.c: Change NO_VFPRINTF to VPRINTF_MISSING, for compatibility with autoconf. Mon Sep 2 22:02:40 1991 David J. MacKenzie (djm at apple-gunkies) * sed.c (compile_regex): Treat \ as a normal character when in a char class. Thu Aug 8 00:15:33 1991 David J. MacKenzie (djm at bleen) * Version 1.08. * sed.c (compile_filename): If reading a file fails, read /dev/null instead. It's what Unix and POSIX do, effectively. * sed.c (compile_regex): The 'slash' character doesn't terminate the regex if it's in a character class. * sed.c (main): If given no args, or bad option, print usage message. (usage): New function. * sed.c (execute_program): Amount written for 'P' command was wrong. From stephend@ksr.com (Stephen Davis). Wed Aug 7 16:51:14 1991 David J. MacKenzie (djm at apple-gunkies) * sed.c (append_pattern_space): Check for buffer full before instead of after writing to buffer. Don't need to test for EOF initially anymore, due to the next change. (execute_program): For 'n' and 'N' commands, if eof is reached in input, quit the script like Unix sed does. Fix memory allocation problems for 'a' and 'r' commands. (compile_program): Fix off by one error in processing comments. All of the above are from Tapani Tarvainen, tarvaine@tukki.jyu.fi. * sed.c (setup_jump): Use isblank instead of testing for ' ' or '\t', for POSIX locales. * utils.c (ck_strdup): Renamed from strdup. * sed.c: Change callers. * sed.c, utils.c: Clean up declarations and includes to get rid of compiler warnings. * sed.c (main): Add long-named options. Don't complain if -n is given twice. Fri Aug 2 12:33:16 1991 David J. MacKenzie (djm at apple-gunkies) * configure: Support +srcdir arg. Create config.status and remove it and Makefile if interrupted while creating them. * Makefile.in: Change DESTDIR to prefix. Mon Jul 15 13:07:39 1991 David J. MacKenzie (djm at wookumz.gnu.ai.mit.edu) * sed.c (main): Add -V option to print version number. (USAGE): Mention -V. Mon Jul 8 01:42:22 1991 David J. MacKenzie (djm at geech.gnu.ai.mit.edu) * sed.c: Define bcopy in terms of memcpy if STDC_HEADERS as well as if USG. (compile_filename): Don't glob filename (for 'r' and 'w' commands). Unix sed doesn't do it and it's not very useful, since it can only match 0 or 1 files. (execute_program): Change '\a' to 007 since some compilers don't recognize \a. * utils.c: New file; code moved from sed.c. * Replace Makefile with Makefile.in and configure. Update README. Tue Mar 26 13:00:48 EST 1991 Jay Fenlason (hack@gnu.ai.mit.edu) * sed.c (match_address) Added a trivial cast for portability. Mon Feb 25 13:23:29 EST 1991 Jay Fenlason (hack@ai.mit.edu) * sed.c Changed 's' command to work with latest version of regex() routines, which mysteriously changed somewhere in there. . . A one-line patch from David Eckelkamp (eckelkamp@mcc.com). Initialize the fastmap in the hopes that it'll make sed faster. Thu Feb 21 13:42:27 EST 1991 Jay Fenlason (hack@ai.mti.edu) * sed.c Change panic to compile with other __STDC__ compilers. Wed Jan 30 10:46:38 EST 1991 Jay Fenlason (hack@ai.mit.edu) * sed.c Changed version number. Made new release. Tue Nov 27 15:34:51 EST 1990 Jay Fenlason (hack@ai.mit.edu) * sed.c (setup_jump) Don't blow chunks if there isn't a label after a b or t command. (main) Don't panic if it a branch command doesn't have a label to branch to. (main) Collect all the -e arguments together and parse them all at once. This way, -e { -e mumble -e } will work. All these small patches from David Schmidt (davids@isc-br.isc-br.com) Tue Sep 11 12:51:37 EDT 1990 Jay Fenlason (hack@ai.mit.edu) * sed.c Changed some function forward declarations to use VOID * instead of char * Mon Jul 16 11:12:54 EDT 1990 Jay Fenlason (hack@ai.mit.edu) * sed.c (ck_malloc) Use malloc(1) instead of malloc(0) if given a request for zero bytes. Tue Jun 5 02:05:37 1990 David J. MacKenzie (djm at albert.ai.mit.edu) * sed.c: Remove excess newlines from calls to panic. Reformat some comments to fit in 79 columns. Base whether to use void * on __STDC__, not __GNU__. (main): Add missing arg when printing usage message. Print usage if given invalid arg. (panic) [__STDC__]: Add missing ", ...". (compile_filename): Print correct error message if glob_filename returns NULL. Thu Apr 5 21:41:12 1990 Jim Kingdon (kingdon at pogo.ai.mit.edu) * sed.c (execute_program, case 'r'): When need to realloc append.text, multiply append.alloc by 2 instead of adding cur_cmd->x.cmd_txt.text_len. Tue Mar 6 15:55:35 EST 1990 Jay Fenlason (hack@ai.mit.edu) * sed.c (compile_regex) Allocate 10 bytes extra space needed by re_compile_pattern. Sun Feb 25 16:32:10 1990 Jim Kingdon (kingdon at pogo.ai.mit.edu) * sed.c (execute_program, case 'l'): Print \00 instead of \0. Print backslash as \\ not \. Print \xx instead of /xx. Thu Feb 1 14:02:28 EST 1990 hack@wookumz * sed.c (memchr) Use () inside inner loop so it will work correctly. A two character patch from Robert A Bruce (rab@allspice.berkeley.edu) Wed Sep 27 18:47:39 EDT 1989 hack@ai.mit.edu * sed.c (compile_regex) New function. When compiling regex, turn ^ into \` and $ into \' so that they won't match on embedded newlines. UN*X pattern matching is a crock. (compile_program, compile_address) call compile_regex. Mon Sep 18 10:15:32 EDT 1989 hack@ai.mit.edu * sed.c (compile_program): define translate as unsigned char * so that y command will work on non-ascii characters. Changed version number to 1.06. Thu Sep 14 15:57:08 EDT 1989 hack@ai.mit.edu * sed.c (compile_program) Let programs use ; to terminate } as well as newline. (read_file) Print an error msg to stderr if it can't open an input file. Thu Mar 23 18:04:46 1989 Randall Smith (randy at apple-gunkies.ai.mit.edu) * Makefile, sed.c: Added new copyright notice. * Makefile: Make distributions which follow the symlinks. hack@ai.mit.edu 1.05 Fixed error in 'r' (now does things in the right order) 1.04 Fixed s/re/rep/[number] 1.03 Fixes from Mike Haertel for regexps that match the empty string, and for Ritchie stdio (non-sticky EOF) 1.02 Fixed 't', 'b', ':' to trim leading spaces and tabs Fixed \\ in replacement of 's' command Added comments 1.01 Added s/re/rep/[digits] added #n as first line of script added filename globbing added 'l' command All in the name of POSIX 1.00 Began (thinking about) distributing this file Local Variables: mode: indented-text left-margin: 8 version-control: never End: dist/sed/README100660 765 144 4134 6071415136 12024 0ustar lordusers* ABOUT BUGS Before reporting a bug, please check the list of oft-reported non-bugs (below). Bugs and comments may be sent to bug-gnu-utils@prep.ai.mit.edu. Please do not send a bug report like this: [while building frobme-1.3.4] $ configure sed: file sedscr line 1: Unknown option to 's' $ sed --version GNU sed version 2.04 If GNU sed doesn't configure your favorite package, take a few extra minutes to identify the specific problem and make a stand-alone test case. A stand-alone test case includes all the data necessary to perform the test, and the specific invocation of sed that causes the problem. The smaller a stand-alone test case is, the better. A test case should not involve something as far removed from sed as ``try to configure frobme-1.3.2''. Yes, that is in principle enough information to look for the bug, but that is not a very practical prospect. & NONBUGS `sed -n' and `s/regex/replace/p' Some versions of sed ignore the `p' (print) option of an `s' command unless the `-n' command switch has been specified. Other versions always honor the `p' option. GNU sed is the latter sort. regexp syntax clashes GNU sed uses the Posix basic regular expression syntax. According to the standard, the meaning of some escape sequences is undefined in this syntax; notably `\|' and `\+'. As in all GNU programs that use Posix basic regular expressions, sed interprets these escape sequences as meta-characters. So, `x\+' matches one or more occurences of `x'. `abc\|def' matches either `abc' or `def'. This syntax may cause problems when running scripts written for other seds. Some sed programs have been written with the assumption that `\|' and `\+' match the literal characters `|' and `+'. Such scripts must be modified by removing the spurious backslashes if they are to be used with GNU sed. [If you have need of a free sed that understands the regexp syntax of your choice, the source to GNU sed may be a good place to start. Consider changing the call to re_set_syntax in function main in `sed.c'. The file regex.h contains an explanation of the supported syntax options.] dist/sed/Makefile.in100660 765 144 6060 6071417011 13203 0ustar lordusers# Makefile for GNU SED, a batch editor. # Copyright (C) 1987, 1991 Free Software Foundation, Inc. # # This file is part of GNU SED. # # GNU SED is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by # the Free Software Foundation; either version 2, or (at your option) # any later version. # # GNU SED is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with GNU SED; see the file COPYING. If not, write to # the Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. SHELL = /bin/sh #### Start of system configuration section. #### srcdir = @srcdir@ VPATH = @srcdir@ CC = @CC@ INSTALL = @INSTALL@ INSTALL_PROGRAM = @INSTALL_PROGRAM@ # Things you might add to DEFS: # -DSTDC_HEADERS If you have ANSI C headers and libraries. # -DUSG If you have System V/ANSI C string # and memory functions and headers. # -D__CHAR_UNSIGNED__ If type `char' is unsigned. # gcc defines this automatically. # -DNO_VFPRINTF If you lack vprintf function (but have _doprnt). DEFS = @DEFS@ LIBS = @LIBS@ CFLAGS = -g LDFLAGS = -g extra_objs =@ALLOCA@ prefix = /usr/local exec_prefix = $(prefix) # Prefix for each installed program, normally empty or `g'. binprefix = # Where to install the executable. bindir = $(exec_prefix)/bin #### End of system configuration section. #### objs = sed.o utils.o getopt.o getopt1.o srcs = sed.c utils.c getopt.c getopt1.c alloca.c distfiles = COPYING News ChangeLog README Makefile.in \ configure configure.in getopt.h $(srcs) testsuite PLUGIN all_objs= $(objs) $(extra_objs) all: sed .c.o: $(CC) -c $(CPPFLAGS) $(DEFS) $(CFLAGS) -I$(srcdir) -I$(srcdir)/../rx $< sed: $(all_objs) ../rx/librx.a $(CC) -o $@ $(LDFLAGS) $(all_objs) ../rx/librx.a $(LIBS) sed.o rx.o: $(srcdir)/../rx/rx.h sed.o getopt1.o: getopt.h Makefile: $(srcdir)/Makefile.in $(SHELL) config.status SUBDIR=. manifest: for file in $(distfiles); do echo $(SUBDIR)/$$file; done install: all test -d $(bindir) || mkdir $(bindir) $(INSTALL_PROGRAM) sed $(bindir)/$(binprefix)sed TAGS: $(srcs) etags $(srcs) clean: rm -f sed *.o core mostlyclean: clean distclean: clean rm -f Makefile config.status realclean: distclean rm -f TAGS dist: $(distfiles) echo sed-`sed -e '/version_string/!d' -e 's/[^0-9.]*\([0-9.]*\).*/\1/' -e q sed.c` > .fname rm -rf `cat .fname` mkdir `cat .fname` ln $(distfiles) `cat .fname` tar chzf `cat .fname`.tar.Z `cat .fname` rm -rf `cat .fname` .fname dist.afs: $(distfiles) echo sed-`sed -e '/version_string/!d' -e 's/[^0-9.]*\([0-9.]*\).*/\1/' -e q sed.c` > .fname rm -rf `cat .fname` mkdir `cat .fname` cd `cat .fname`; \ for file in $(distfiles); do ln -s ../"$$file" .; done; \ cd .. tar chf `cat .fname`.tar `cat .fname` gzip `cat .fname`.tar rm -rf `cat .fname` .fname alloca.o : alloca.c dist/sed/configure100770 765 144 50363 6070131465 13076 0ustar lordusers#!/bin/sh # Guess values for system-dependent variables and create Makefiles. # Generated automatically using autoconf. # Copyright (C) 1991, 1992, 1993 Free Software Foundation, Inc. # This program is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by # the Free Software Foundation; either version 2, or (at your option) # any later version. # This program is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # You should have received a copy of the GNU General Public License # along with this program; if not, write to the Free Software # Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. # Usage: configure [--srcdir=DIR] [--host=HOST] [--gas] [--nfp] # [--prefix=PREFIX] [--exec-prefix=PREFIX] [--with-PACKAGE[=VALUE]] # Ignores all args except --srcdir, --prefix, --exec-prefix, and # --with-PACKAGE[=VALUE] unless this script has special code to handle it. for arg do # Handle --exec-prefix with a space before the argument. if test x$next_exec_prefix = xyes; then exec_prefix=$arg; next_exec_prefix= # Handle --host with a space before the argument. elif test x$next_host = xyes; then next_host= # Handle --prefix with a space before the argument. elif test x$next_prefix = xyes; then prefix=$arg; next_prefix= # Handle --srcdir with a space before the argument. elif test x$next_srcdir = xyes; then srcdir=$arg; next_srcdir= else case $arg in # For backward compatibility, recognize -exec-prefix and --exec_prefix. -exec-prefix=* | --exec_prefix=* | --exec-prefix=* | --exec-prefi=* | --exec-pref=* | --exec-pre=* | --exec-pr=* | --exec-p=* | --exec-=* | --exec=* | --exe=* | --ex=* | --e=*) exec_prefix=`echo $arg | sed 's/[-a-z_]*=//'` ;; -exec-prefix | --exec_prefix | --exec-prefix | --exec-prefi | --exec-pref | --exec-pre | --exec-pr | --exec-p | --exec- | --exec | --exe | --ex | --e) next_exec_prefix=yes ;; -gas | --gas | --ga | --g) ;; -host=* | --host=* | --hos=* | --ho=* | --h=*) ;; -host | --host | --hos | --ho | --h) next_host=yes ;; -nfp | --nfp | --nf) ;; -prefix=* | --prefix=* | --prefi=* | --pref=* | --pre=* | --pr=* | --p=*) prefix=`echo $arg | sed 's/[-a-z_]*=//'` ;; -prefix | --prefix | --prefi | --pref | --pre | --pr | --p) next_prefix=yes ;; -srcdir=* | --srcdir=* | --srcdi=* | --srcd=* | --src=* | --sr=* | --s=*) srcdir=`echo $arg | sed 's/[-a-z_]*=//'` ;; -srcdir | --srcdir | --srcdi | --srcd | --src | --sr | --s) next_srcdir=yes ;; -with-* | --with-*) package=`echo $arg|sed -e 's/-*with-//' -e 's/=.*//'` # Reject names that aren't valid shell variable names. if test -n "`echo $package| sed 's/[-a-zA-Z0-9_]//g'`"; then echo "configure: $package: invalid package name" >&2; exit 1 fi package=`echo $package| sed 's/-/_/g'` case "$arg" in *=*) val="`echo $arg|sed 's/[^=]*=//'`" ;; *) val=1 ;; esac eval "with_$package='$val'" ;; -v | -verbose | --verbose | --verbos | --verbo | --verb | --ver | --ve | --v) verbose=yes ;; *) ;; esac fi done trap 'rm -fr conftest* confdefs* core; exit 1' 1 3 15 trap 'rm -f confdefs*' 0 # NLS nuisances. # These must not be set unconditionally because not all systems understand # e.g. LANG=C (notably SCO). if test "${LC_ALL+set}" = 'set' ; then LC_ALL=C; export LC_ALL; fi if test "${LANG+set}" = 'set' ; then LANG=C; export LANG; fi rm -f conftest* confdefs.h echo > confdefs.h compile='${CC-cc} $CFLAGS conftest.c -o conftest $LIBS >/dev/null 2>&1' # A filename unique to this package, relative to the directory that # configure is in, which we can look for to find out if srcdir is correct. unique_file=sed.c # Find the source files, if location was not specified. if test -z "$srcdir"; then srcdirdefaulted=yes # Try the directory containing this script, then `..'. prog=$0 confdir=`echo $prog|sed 's%/[^/][^/]*$%%'` test "X$confdir" = "X$prog" && confdir=. srcdir=$confdir if test ! -r $srcdir/$unique_file; then srcdir=.. fi fi if test ! -r $srcdir/$unique_file; then if test x$srcdirdefaulted = xyes; then echo "configure: Can not find sources in \`${confdir}' or \`..'." 1>&2 else echo "configure: Can not find sources in \`${srcdir}'." 1>&2 fi exit 1 fi # Preserve a srcdir of `.' to avoid automounter screwups with pwd. # But we can't avoid them for `..', to make subdirectories work. case $srcdir in .|/*|~*) ;; *) srcdir=`cd $srcdir; pwd` ;; # Make relative path absolute. esac # Save the original args to write them into config.status later. configure_args="$*" if test -z "$CC"; then # Extract the first word of `gcc', so it can be a program name with args. set dummy gcc; word=$2 echo checking for $word IFS="${IFS= }"; saveifs="$IFS"; IFS="${IFS}:" for dir in $PATH; do test -z "$dir" && dir=. if test -f $dir/$word; then CC="gcc" break fi done IFS="$saveifs" fi test -z "$CC" && CC="cc" test -n "$CC" && test -n "$verbose" && echo " setting CC to $CC" # Find out if we are using GNU C, under whatever name. cat > conftest.c < conftest.out 2>&1 if egrep yes conftest.out >/dev/null 2>&1; then GCC=1 # For later tests. fi rm -f conftest* echo checking how to run the C preprocessor if test -z "$CPP"; then # This must be in double quotes, not single quotes, because CPP may get # substituted into the Makefile and ``${CC-cc}'' will simply confuse # make. It must be expanded now. CPP="${CC-cc} -E" cat > conftest.c < Syntax Error EOF err=`eval "($CPP conftest.c >/dev/null) 2>&1"` if test -z "$err"; then : else rm -rf conftest* CPP=/lib/cpp fi rm -f conftest* fi test ".${verbose}" != "." && echo " setting CPP to $CPP" # Make sure to not get the incompatible SysV /etc/install and # /usr/sbin/install, which might be in PATH before a BSD-like install, # or the SunOS /usr/etc/install directory, or the AIX /bin/install, # or the AFS install, which mishandles nonexistent args, or # /usr/ucb/install on SVR4, which tries to use the nonexistent group # `staff'. On most BSDish systems install is in /usr/bin, not /usr/ucb # anyway. Sigh. if test "z${INSTALL}" = "z" ; then echo checking for install IFS="${IFS= }"; saveifs="$IFS"; IFS="${IFS}:" for dir in $PATH; do test -z "$dir" && dir=. case $dir in /etc|/usr/sbin|/usr/etc|/usr/afsws/bin|/usr/ucb) ;; *) if test -f $dir/installbsd; then INSTALL="$dir/installbsd -c" # OSF1 INSTALL_PROGRAM='$(INSTALL)' INSTALL_DATA='$(INSTALL) -m 644' break fi if test -f $dir/install; then if grep dspmsg $dir/install >/dev/null 2>&1; then : # AIX else INSTALL="$dir/install -c" INSTALL_PROGRAM='$(INSTALL)' INSTALL_DATA='$(INSTALL) -m 644' break fi fi ;; esac done IFS="$saveifs" fi INSTALL=${INSTALL-cp} test -n "$verbose" && echo " setting INSTALL to $INSTALL" INSTALL_PROGRAM=${INSTALL_PROGRAM-'$(INSTALL)'} test -n "$verbose" && echo " setting INSTALL_PROGRAM to $INSTALL_PROGRAM" INSTALL_DATA=${INSTALL_DATA-'$(INSTALL)'} test -n "$verbose" && echo " setting INSTALL_DATA to $INSTALL_DATA" echo checking for minix/config.h cat > conftest.c < EOF err=`eval "($CPP conftest.c >/dev/null) 2>&1"` if test -z "$err"; then rm -rf conftest* MINIX=1 fi rm -f conftest* # The Minix shell can't assign to the same variable on the same line! if test -n "$MINIX"; then { test -n "$verbose" && \ echo " defining _POSIX_SOURCE" echo "#define" _POSIX_SOURCE 1 >> confdefs.h DEFS="$DEFS -D_POSIX_SOURCE=1" } { test -n "$verbose" && \ echo " defining" _POSIX_1_SOURCE to be 2 echo "#define" _POSIX_1_SOURCE 2 >> confdefs.h DEFS="$DEFS -D_POSIX_1_SOURCE=2" } { test -n "$verbose" && \ echo " defining _MINIX" echo "#define" _MINIX 1 >> confdefs.h DEFS="$DEFS -D_MINIX=1" } fi echo checking for POSIXized ISC if test -d /etc/conf/kconfig.d && grep _POSIX_VERSION /usr/include/sys/unistd.h >/dev/null 2>&1 then ISC=1 # If later tests want to check for ISC. { test -n "$verbose" && \ echo " defining _POSIX_SOURCE" echo "#define" _POSIX_SOURCE 1 >> confdefs.h DEFS="$DEFS -D_POSIX_SOURCE=1" } if test -n "$GCC"; then CC="$CC -posix" else CC="$CC -Xp" fi fi echo checking for ANSI C header files cat > conftest.c < #include #include #include EOF err=`eval "($CPP conftest.c >/dev/null) 2>&1"` if test -z "$err"; then rm -rf conftest* # SunOS 4.x string.h does not declare mem*, contrary to ANSI. echo '#include "confdefs.h" #include ' > conftest.c eval "$CPP conftest.c > conftest.out 2>&1" if egrep "memchr" conftest.out >/dev/null 2>&1; then rm -rf conftest* # SGI's /bin/cc from Irix-4.0.5 gets non-ANSI ctype macros unless using -ansi. cat > conftest.c < #define ISLOWER(c) ('a' <= (c) && (c) <= 'z') #define TOUPPER(c) (ISLOWER(c) ? 'A' + ((c) - 'a') : (c)) #define XOR(e,f) (((e) && !(f)) || (!(e) && (f))) int main () { int i; for (i = 0; i < 256; i++) if (XOR (islower (i), ISLOWER (i)) || toupper (i) != TOUPPER (i)) exit(2); exit (0); } EOF eval $compile if test -s conftest && (./conftest; exit) 2>/dev/null; then rm -rf conftest* { test -n "$verbose" && \ echo " defining STDC_HEADERS" echo "#define" STDC_HEADERS 1 >> confdefs.h DEFS="$DEFS -DSTDC_HEADERS=1" } fi rm -f conftest* fi rm -f conftest* fi rm -f conftest* for hdr in string.h do trhdr=HAVE_`echo $hdr | tr '[a-z]./' '[A-Z]__'` echo checking for ${hdr} cat > conftest.c < EOF err=`eval "($CPP conftest.c >/dev/null) 2>&1"` if test -z "$err"; then rm -rf conftest* { test -n "$verbose" && \ echo " defining ${trhdr}" echo "#define" ${trhdr} 1 >> confdefs.h DEFS="$DEFS -D${trhdr}=1" } fi rm -f conftest* done echo checking for vprintf cat > conftest.c <> confdefs.h DEFS="$DEFS -DHAVE_VPRINTF=1" } else rm -rf conftest* vprintf_missing=1 fi rm -f conftest* if test -n "$vprintf_missing"; then echo checking for _doprnt cat > conftest.c <> confdefs.h DEFS="$DEFS -DHAVE_DOPRNT=1" } fi rm -f conftest* fi for func in bcopy memcpy do trfunc=HAVE_`echo $func | tr '[a-z]' '[A-Z]'` echo checking for ${func} cat > conftest.c < int main() { exit(0); } int t() { /* The GNU C library defines this for functions which it implements to always fail with ENOSYS. Some functions are actually named something starting with __ and the normal name is an alias. */ #if defined (__stub_${func}) || defined (__stub___${func}) choke me #else /* Override any gcc2 internal prototype to avoid an error. */ extern char ${func}(); ${func}(); #endif } EOF if eval $compile; then rm -rf conftest* { test -n "$verbose" && \ echo " defining ${trfunc}" echo "#define" ${trfunc} 1 >> confdefs.h DEFS="$DEFS -D${trfunc}=1" } fi rm -f conftest* done prog='/* Ultrix mips cc rejects this. */ typedef int charset[2]; const charset x; /* SunOS 4.1.1 cc rejects this. */ char const *const *ccp; char **p; /* AIX XL C 1.02.0.0 rejects this. It does not let you subtract one const X* pointer from another in an arm of an if-expression whose if-part is not a constant expression */ const char *g = "string"; ccp = &g + (g ? g-g : 0); /* HPUX 7.0 cc rejects these. */ ++ccp; p = (char**) ccp; ccp = (char const *const *) p; { /* SCO 3.2v4 cc rejects this. */ char *t; char const *s = 0 ? (char *) 0 : (char const *) 0; *t++ = 0; } { /* Someone thinks the Sun supposedly-ANSI compiler will reject this. */ int x[] = {25,17}; const int *foo = &x[0]; ++foo; } { /* Sun SC1.0 ANSI compiler rejects this -- but not the above. */ typedef const int *iptr; iptr p = 0; ++p; } { /* AIX XL C 1.02.0.0 rejects this saying "k.c", line 2.27: 1506-025 (S) Operand must be a modifiable lvalue. */ struct s { int j; const int *ap[3]; }; struct s *b; b->j = 5; }' echo checking for lack of working const cat > conftest.c <> confdefs.h DEFS="$DEFS -Dconst=" } fi rm -f conftest* # The Ultrix 4.2 mips builtin alloca declared by alloca.h only works # for constant arguments. Useless! echo checking for working alloca.h cat > conftest.c < int main() { exit(0); } int t() { char *p = alloca(2 * sizeof(int)); } EOF if eval $compile; then rm -rf conftest* { test -n "$verbose" && \ echo " defining HAVE_ALLOCA_H" echo "#define" HAVE_ALLOCA_H 1 >> confdefs.h DEFS="$DEFS -DHAVE_ALLOCA_H=1" } fi rm -f conftest* decl="#ifdef __GNUC__ #define alloca __builtin_alloca #else #if HAVE_ALLOCA_H #include #else #ifdef _AIX #pragma alloca #else char *alloca (); #endif #endif #endif " echo checking for alloca cat > conftest.c < conftest.c < conftest.out 2>&1" if egrep "winnitude" conftest.out >/dev/null 2>&1; then rm -rf conftest* echo checking for _getb67 cat > conftest.c < int main() { exit(0); } int t() { /* The GNU C library defines this for functions which it implements to always fail with ENOSYS. Some functions are actually named something starting with __ and the normal name is an alias. */ #if defined (__stub__getb67) || defined (__stub____getb67) choke me #else /* Override any gcc2 internal prototype to avoid an error. */ extern char _getb67(); _getb67(); #endif } EOF if eval $compile; then rm -rf conftest* { test -n "$verbose" && \ echo " defining" CRAY_STACKSEG_END to be _getb67 echo "#define" CRAY_STACKSEG_END _getb67 >> confdefs.h DEFS="$DEFS -DCRAY_STACKSEG_END=_getb67" } else rm -rf conftest* echo checking for GETB67 cat > conftest.c < int main() { exit(0); } int t() { /* The GNU C library defines this for functions which it implements to always fail with ENOSYS. Some functions are actually named something starting with __ and the normal name is an alias. */ #if defined (__stub_GETB67) || defined (__stub___GETB67) choke me #else /* Override any gcc2 internal prototype to avoid an error. */ extern char GETB67(); GETB67(); #endif } EOF if eval $compile; then rm -rf conftest* { test -n "$verbose" && \ echo " defining" CRAY_STACKSEG_END to be GETB67 echo "#define" CRAY_STACKSEG_END GETB67 >> confdefs.h DEFS="$DEFS -DCRAY_STACKSEG_END=GETB67" } else rm -rf conftest* echo checking for getb67 cat > conftest.c < int main() { exit(0); } int t() { /* The GNU C library defines this for functions which it implements to always fail with ENOSYS. Some functions are actually named something starting with __ and the normal name is an alias. */ #if defined (__stub_getb67) || defined (__stub___getb67) choke me #else /* Override any gcc2 internal prototype to avoid an error. */ extern char getb67(); getb67(); #endif } EOF if eval $compile; then rm -rf conftest* { test -n "$verbose" && \ echo " defining" CRAY_STACKSEG_END to be getb67 echo "#define" CRAY_STACKSEG_END getb67 >> confdefs.h DEFS="$DEFS -DCRAY_STACKSEG_END=getb67" } fi rm -f conftest* fi rm -f conftest* fi rm -f conftest* fi rm -f conftest* fi rm -f conftest* if test -n "$alloca_missing"; then # The SVR3 libPW and SVR4 libucb both contain incompatible functions # that cause trouble. Some versions do not even contain alloca or # contain a buggy version. If you still want to use their alloca, # use ar to extract alloca.o from them instead of compiling alloca.c. ALLOCA=alloca.o { test -n "$verbose" && \ echo " defining C_ALLOCA" echo "#define" C_ALLOCA 1 >> confdefs.h DEFS="$DEFS -DC_ALLOCA=1" } echo 'checking stack direction for C alloca' echo checking whether cross-compiling # If we cannot run a trivial program, we must be cross compiling. cat > conftest.c </dev/null; then : else rm -rf conftest* cross_compiling=1 fi rm -f conftest* if test -n "$cross_compiling" then { test -n "$verbose" && \ echo " defining" STACK_DIRECTION to be 0 echo "#define" STACK_DIRECTION 0 >> confdefs.h DEFS="$DEFS -DSTACK_DIRECTION=0" } else cat > conftest.c < addr) ? 1 : -1; } main () { exit (find_stack_direction() < 0); } EOF eval $compile if test -s conftest && (./conftest; exit) 2>/dev/null; then rm -rf conftest* { test -n "$verbose" && \ echo " defining" STACK_DIRECTION to be 1 echo "#define" STACK_DIRECTION 1 >> confdefs.h DEFS="$DEFS -DSTACK_DIRECTION=1" } else rm -rf conftest* { test -n "$verbose" && \ echo " defining" STACK_DIRECTION to be -1 echo "#define" STACK_DIRECTION -1 >> confdefs.h DEFS="$DEFS -DSTACK_DIRECTION=-1" } fi fi rm -f conftest* fi # Set default prefixes. if test -n "$prefix"; then test -z "$exec_prefix" && exec_prefix='${prefix}' prsub="s%^prefix\\([ ]*\\)=\\([ ]*\\).*$%prefix\\1=\\2$prefix%" fi if test -n "$exec_prefix"; then prsub="$prsub s%^exec_prefix\\([ ]*\\)=\\([ ]*\\).*$%exec_prefix\\1=\\2$exec_prefix%" fi # Quote sed substitution magic chars in DEFS. cat >conftest.def < config.status </dev/null | sed 1q`: # # $0 $configure_args for arg do case "\$arg" in -recheck | --recheck | --rechec | --reche | --rech | --rec | --re | --r) echo running \${CONFIG_SHELL-/bin/sh} $0 $configure_args exec \${CONFIG_SHELL-/bin/sh} $0 $configure_args ;; *) echo "Usage: config.status --recheck" 2>&1; exit 1 ;; esac done trap 'rm -f Makefile; exit 1' 1 3 15 CC='$CC' CPP='$CPP' INSTALL='$INSTALL' INSTALL_PROGRAM='$INSTALL_PROGRAM' INSTALL_DATA='$INSTALL_DATA' ALLOCA='$ALLOCA' LIBS='$LIBS' srcdir='$srcdir' DEFS='$DEFS' prefix='$prefix' exec_prefix='$exec_prefix' prsub='$prsub' extrasub='$extrasub' EOF cat >> config.status <<\EOF top_srcdir=$srcdir CONFIG_FILES=${CONFIG_FILES-"Makefile"} for file in .. ${CONFIG_FILES}; do if test "x$file" != x..; then srcdir=$top_srcdir # Remove last slash and all that follows it. Not all systems have dirname. dir=`echo $file|sed 's%/[^/][^/]*$%%'` if test "$dir" != "$file"; then test "$top_srcdir" != . && srcdir=$top_srcdir/$dir test ! -d $dir && mkdir $dir fi echo creating $file rm -f $file echo "# Generated automatically from `echo $file|sed 's|.*/||'`.in by configure." > $file sed -e " $prsub $extrasub s%@CC@%$CC%g s%@CPP@%$CPP%g s%@INSTALL@%$INSTALL%g s%@INSTALL_PROGRAM@%$INSTALL_PROGRAM%g s%@INSTALL_DATA@%$INSTALL_DATA%g s%@ALLOCA@%$ALLOCA%g s%@LIBS@%$LIBS%g s%@srcdir@%$srcdir%g s%@DEFS@%$DEFS% " $top_srcdir/${file}.in >> $file fi; done exit 0 EOF chmod +x config.status ${CONFIG_SHELL-/bin/sh} config.status dist/sed/configure.in100660 765 144 407 6070131465 13433 0ustar lordusersdnl Process this file with autoconf to produce a configure script. AC_INIT(sed.c) AC_PROG_CC AC_PROG_CPP AC_PROG_INSTALL AC_MINIX AC_ISC_POSIX AC_STDC_HEADERS AC_HAVE_HEADERS(string.h) AC_VPRINTF AC_HAVE_FUNCS(bcopy memcpy) AC_CONST AC_ALLOCA AC_OUTPUT(Makefile) dist/sed/getopt.h100660 765 144 10474 6070131465 12642 0ustar lordusers/* Declarations for getopt. Copyright (C) 1989, 1990, 1991, 1992, 1993 Free Software Foundation, Inc. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */ #ifndef _GETOPT_H #define _GETOPT_H 1 #ifdef __cplusplus extern "C" { #endif /* For communication from `getopt' to the caller. When `getopt' finds an option that takes an argument, the argument value is returned here. Also, when `ordering' is RETURN_IN_ORDER, each non-option ARGV-element is returned here. */ extern char *optarg; /* Index in ARGV of the next element to be scanned. This is used for communication to and from the caller and for communication between successive calls to `getopt'. On entry to `getopt', zero means this is the first call; initialize. When `getopt' returns EOF, this is the index of the first of the non-option elements that the caller should itself scan. Otherwise, `optind' communicates from one call to the next how much of ARGV has been scanned so far. */ extern int optind; /* Callers store zero here to inhibit the error message `getopt' prints for unrecognized options. */ extern int opterr; /* Set to an option character which was unrecognized. */ extern int optopt; /* Describe the long-named options requested by the application. The LONG_OPTIONS argument to getopt_long or getopt_long_only is a vector of `struct option' terminated by an element containing a name which is zero. The field `has_arg' is: no_argument (or 0) if the option does not take an argument, required_argument (or 1) if the option requires an argument, optional_argument (or 2) if the option takes an optional argument. If the field `flag' is not NULL, it points to a variable that is set to the value given in the field `val' when the option is found, but left unchanged if the option is not found. To have a long-named option do something other than set an `int' to a compiled-in constant, such as set a value from `optarg', set the option's `flag' field to zero and its `val' field to a nonzero value (the equivalent single-letter option character, if there is one). For long options that have a zero `flag' field, `getopt' returns the contents of the `val' field. */ struct option { #if __STDC__ const char *name; #else char *name; #endif /* has_arg can't be an enum because some compilers complain about type mismatches in all the code that assumes it is an int. */ int has_arg; int *flag; int val; }; /* Names for the values of the `has_arg' field of `struct option'. */ #define no_argument 0 #define required_argument 1 #define optional_argument 2 #if __STDC__ #if defined(__GNU_LIBRARY__) /* Many other libraries have conflicting prototypes for getopt, with differences in the consts, in stdlib.h. To avoid compilation errors, only prototype getopt for the GNU C library. */ extern int getopt (int argc, char *const *argv, const char *shortopts); #else /* not __GNU_LIBRARY__ */ extern int getopt (); #endif /* not __GNU_LIBRARY__ */ extern int getopt_long (int argc, char *const *argv, const char *shortopts, const struct option *longopts, int *longind); extern int getopt_long_only (int argc, char *const *argv, const char *shortopts, const struct option *longopts, int *longind); /* Internal only. Users should not call this directly. */ extern int _getopt_internal (int argc, char *const *argv, const char *shortopts, const struct option *longopts, int *longind, int long_only); #else /* not __STDC__ */ extern int getopt (); extern int getopt_long (); extern int getopt_long_only (); extern int _getopt_internal (); #endif /* not __STDC__ */ #ifdef __cplusplus } #endif #endif /* _GETOPT_H */ dist/sed/sed.c100660 765 144 131332 6071414226 12123 0ustar lordusers/* GNU SED, a batch stream editor. Copyright (C) 1989, 1990, 1991, 1992, 1993, 1994, 1995 Free Software Foundation, Inc. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ #ifdef __STDC__ #define VOID void #else #define VOID char #endif #define _GNU_SOURCE #include #ifndef isblank #define isblank(c) ((c) == ' ' || (c) == '\t') #endif #include #include #include "rxposix.h" #include "getopt.h" #if defined(STDC_HEADERS) #include #endif #if HAVE_STRING_H || defined(STDC_HEADERS) #include #ifndef bzero #define bzero(s, n) memset ((s), 0, (n)) #endif #if !defined(STDC_HEADERS) #include #endif #else #include #endif #ifdef RX_MEMDBUG #include #endif extern char * rx_error_msg[]; static int rx_testing = 0; #ifndef HAVE_BCOPY #ifdef HAVE_MEMCPY #define bcopy(FROM,TO,LEN) memcpy(TO,FROM,LEN) #else void bcopy (from, to, len) char *from; char *to; int len; { if (from < to) { from += len - 1; to += len - 1; while (len--) *to-- = *from--; } else while (len--) *to++ = *from++; } #endif #endif char *version_string = "GNU sed version 3.00"; /* Struct vector is used to describe a chunk of a compiled sed program. * There is one vector for the main program, and one for each { } pair, * and one for the entire program. For {} blocks, RETURN_[VI] tells where * to continue execution after this VECTOR. */ struct vector { struct sed_cmd *v; int v_length; int v_allocated; struct vector *return_v; int return_i; }; /* Goto structure is used to hold both GOTO's and labels. There are two * separate lists, one of goto's, called 'jumps', and one of labels, called * 'labels'. * the V element points to the descriptor for the program-chunk in which the * goto was encountered. * the v_index element counts which element of the vector actually IS the * goto/label. The first element of the vector is zero. * the NAME element is the null-terminated name of the label. * next is the next goto/label in the list. */ struct sed_label { struct vector *v; int v_index; char *name; struct sed_label *next; }; /* ADDR_TYPE is zero for a null address, * one if addr_number is valid, or * two if addr_regex is valid, * three, if the address is '$' * Other values are undefined. */ enum addr_types { addr_is_null = 0, addr_is_num = 1, addr_is_regex = 2, addr_is_last = 3, addr_is_mod = 4 }; struct addr { int addr_type; struct rx_posix_regex * addr_regex; int addr_number; int modulo, offset; }; /* Aflags: If the low order bit is set, a1 has been * matched; apply this command until a2 matches. * If the next bit is set, apply this command to all * lines that DON'T match the address(es). */ #define A1_MATCHED_BIT 01 #define ADDR_BANG_BIT 02 struct sed_cmd { struct addr a1, a2; int aflags; char cmd; union { /* This structure is used for a, i, and c commands */ struct { char *text; int text_len; } cmd_txt; /* This is used for b and t commands */ struct sed_cmd *label; /* This for r and w commands */ FILE *io_file; /* This for the hairy s command */ /* For the flags var: low order bit means the 'g' option was given, next bit means the 'p' option was given, and the next bit means a 'w' option was given, and wio_file contains the file to write to. */ #define S_GLOBAL_BIT 01 #define S_PRINT_BIT 02 #define S_WRITE_BIT 04 #define S_NUM_BIT 010 struct { struct rx_posix_regex *regx; char *replacement; int replace_length; int flags; int numb; FILE *wio_file; } cmd_regex; /* This for the y command */ unsigned char *translate; /* For { */ struct vector *sub; /* for t and b */ struct sed_label *jump; } x; }; /* Sed operates a line at a time. */ struct line { char *text; /* Pointer to line allocated by malloc. */ int length; /* Length of text. */ int alloc; /* Allocated space for text. */ }; /* This structure holds information about files opend by the 'r', 'w', and 's///w' commands. In paticular, it holds the FILE pointers to use, the file's name. */ #define NUM_FPS 32 struct { FILE *for_read; FILE *for_write; char *name; } file_ptrs[NUM_FPS]; #if defined(__STDC__) # define P_(s) s #else # define P_(s) () #endif void close_files (); void panic P_ ((char *str,...)); char *__fp_name P_ ((FILE * fp)); FILE *ck_fopen P_ ((char *name, char *mode)); void ck_fwrite P_ ((char *ptr, int size, int nmemb, FILE * stream)); void ck_fclose P_ ((FILE * stream)); VOID *ck_malloc P_ ((int size)); VOID *ck_realloc P_ ((VOID * ptr, int size)); char *ck_strdup P_ ((char *str)); VOID *init_buffer P_ ((void)); void flush_buffer P_ ((VOID * bb)); int size_buffer P_ ((VOID * b)); void add_buffer P_ ((VOID * bb, char *p, int n)); void add1_buffer P_ ((VOID * bb, int ch)); char *get_buffer P_ ((VOID * bb)); void compile_string P_ ((char *str)); void compile_file P_ ((char *str)); struct vector *compile_program P_ ((struct vector * vector, int)); void bad_prog P_ ((char *why)); int inchar P_ ((void)); void savchar P_ ((int ch)); int compile_address P_ ((struct addr * addr)); char * last_regex_string = 0; void buffer_regex P_ ((int slash)); void compile_regex P_ ((void)); struct sed_label *setup_jump P_ ((struct sed_label * list, struct sed_cmd * cmd, struct vector * vec)); FILE *compile_filename P_ ((int readit)); void read_file P_ ((char *name)); void execute_program P_ ((struct vector * vec)); int match_address P_ ((struct addr * addr)); int read_pattern_space P_ ((void)); void append_pattern_space P_ ((void)); void line_copy P_ ((struct line * from, struct line * to)); void line_append P_ ((struct line * from, struct line * to)); void str_append P_ ((struct line * to, char *string, int length)); void usage P_ ((int)); extern char *myname; /* If set, don't write out the line unless explictly told to */ int no_default_output = 0; /* Current input line # */ int input_line_number = 0; /* Are we on the last input file? */ int last_input_file = 0; /* Have we hit EOF on the last input file? This is used to decide if we have hit the '$' address yet. */ int input_EOF = 0; /* non-zero if a quit command has been executed. */ int quit_cmd = 0; /* Have we done any replacements lately? This is used by the 't' command. */ int replaced = 0; /* How many '{'s are we executing at the moment */ int program_depth = 0; /* The complete compiled SED program that we are going to run */ struct vector *the_program = 0; /* information about labels and jumps-to-labels. This is used to do the required backpatching after we have compiled all the scripts. */ struct sed_label *jumps = 0; struct sed_label *labels = 0; /* The 'current' input line. */ struct line line; /* An input line that's been stored by later use by the program */ struct line hold; /* A 'line' to append to the current line when it comes time to write it out */ struct line append; /* When we're reading a script command from a string, 'prog_start' and 'prog_end' point to the beginning and end of the string. This would allow us to compile script strings that contain nulls, except that script strings are only read from the command line, which is null-terminated */ unsigned char *prog_start; unsigned char *prog_end; /* When we're reading a script command from a string, 'prog_cur' points to the current character in the string */ unsigned char *prog_cur; /* This is the name of the current script file. It is used for error messages. */ char *prog_name; /* This is the current script file. If it is zero, we are reading from a string stored in 'prog_start' instead. If both 'prog_file' and 'prog_start' are zero, we're in trouble! */ FILE *prog_file; /* this is the number of the current script line that we're compiling. It is used to give out useful and informative error messages. */ int prog_line = 1; /* This is the file pointer that we're currently reading data from. It may be stdin */ FILE *input_file; /* If this variable is non-zero at exit, one or more of the input files couldn't be opened. */ int bad_input = 0; /* 'an empty regular expression is equivalent to the last regular expression read' so we have to keep track of the last regex used. Here's where we store a pointer to it (it is only malloc()'d once) */ struct rx_posix_regex *last_regex; /* Various error messages we may want to print */ static char ONE_ADDR[] = "Command only uses one address"; static char NO_ADDR[] = "Command doesn't take any addresses"; static char LINE_JUNK[] = "Extra characters after command"; static char BAD_EOF[] = "Unexpected End-of-file"; static char NO_REGEX[] = "No previous regular expression"; static char NO_COMMAND[] = "Missing command"; static struct option longopts[] = { {"rxtest", 0, NULL, 'r'}, {"expression", 1, NULL, 'e'}, {"file", 1, NULL, 'f'}, {"quiet", 0, NULL, 'n'}, {"silent", 0, NULL, 'n'}, {"version", 0, NULL, 'V'}, {"help", 0, NULL, 'h'}, {NULL, 0, NULL, 0} }; static int use_extended_syntax_p = 0; extern void (*print_rexp)(); extern int rx_basic_unfaniverse_delay; void main (argc, argv) int argc; char **argv; { int opt; char *e_strings = NULL; int compiled = 0; struct sed_label *go, *lbl; if (!rx_default_cache) print_rexp (); /* Allow roughly a megabyte of cache space. */ rx_default_cache->bytes_allowed = 1048576 * 4; rx_basic_unfaniverse_delay = 512 * 4; myname = argv[0]; while ((opt = getopt_long (argc, argv, "hne:f:rV", longopts, (int *) 0)) != EOF) { switch (opt) { case 'r': use_extended_syntax_p = 1; rx_testing = 1; break; case 'n': no_default_output = 1; break; case 'e': if (e_strings == NULL) { e_strings = ck_malloc (strlen (optarg) + 2); strcpy (e_strings, optarg); } else { e_strings = ck_realloc (e_strings, strlen (e_strings) + strlen (optarg) + 2); strcat (e_strings, optarg); } strcat (e_strings, "\n"); compiled = 1; break; case 'f': if (e_strings) { compile_string (e_strings); free (e_strings); e_strings = 0; } compile_file (optarg); compiled = 1; break; case 'V': fprintf (stderr, "%s\n", version_string); exit (0); break; case 'h': usage (0); break; default: usage (4); break; } } if (e_strings) { compile_string (e_strings); free (e_strings); } if (!compiled) { if (optind == argc) usage (4); compile_string (argv[optind++]); } for (go = jumps; go; go = go->next) { for (lbl = labels; lbl; lbl = lbl->next) if (!strcmp (lbl->name, go->name)) break; if (*go->name && !lbl) panic ("Can't find label for jump to '%s'", go->name); go->v->v[go->v_index].x.jump = lbl; } line.length = 0; line.alloc = 50; line.text = ck_malloc (50); append.length = 0; append.alloc = 50; append.text = ck_malloc (50); hold.length = 1; hold.alloc = 50; hold.text = ck_malloc (50); hold.text[0] = '\n'; if (argc <= optind) { last_input_file++; read_file ("-"); } else while (optind < argc) { if (optind == argc - 1) last_input_file++; read_file (argv[optind]); optind++; if (quit_cmd) break; } close_files (); if (bad_input) exit (2); exit (0); } void close_files () { int nf; for (nf = 0; nf < NUM_FPS; nf++) { if (file_ptrs[nf].for_write) fclose (file_ptrs[nf].for_write); if (file_ptrs[nf].for_read) fclose (file_ptrs[nf].for_read); } } /* 'str' is a string (from the command line) that contains a sed command. Compile the command, and add it to the end of 'the_program' */ void compile_string (str) char *str; { prog_file = 0; prog_line = 0; prog_start = prog_cur = (unsigned char *)str; prog_end = (unsigned char *)str + strlen (str); the_program = compile_program (the_program, prog_line); } /* 'str' is the name of a file containing sed commands. Read them in and add them to the end of 'the_program' */ void compile_file (str) char *str; { int ch; prog_start = prog_cur = prog_end = 0; prog_name = str; prog_line = 1; if (str[0] == '-' && str[1] == '\0') prog_file = stdin; else prog_file = ck_fopen (str, "r"); ch = getc (prog_file); if (ch == '#') { ch = getc (prog_file); if (ch == 'n') no_default_output++; while (ch != EOF && ch != '\n') { ch = getc (prog_file); if (ch == '\\') ch = getc (prog_file); } ++prog_line; } else if (ch != EOF) ungetc (ch, prog_file); the_program = compile_program (the_program, prog_line); } #define MORE_CMDS 40 /* Read a program (or a subprogram within '{' '}' pairs) in and store the compiled form in *'vector' Return a pointer to the new vector. */ struct vector * compile_program (vector, open_line) struct vector *vector; int open_line; { struct sed_cmd *cur_cmd; int ch = 0; int pch; int slash; VOID *b; unsigned char *string; int num; if (!vector) { vector = (struct vector *) ck_malloc (sizeof (struct vector)); vector->v = (struct sed_cmd *) ck_malloc (MORE_CMDS * sizeof (struct sed_cmd)); vector->v_allocated = MORE_CMDS; vector->v_length = 0; vector->return_v = 0; vector->return_i = 0; } for (;;) { skip_comment: do { pch = ch; ch = inchar (); if ((pch == '\\') && (ch == '\n')) ch = inchar (); } while (ch != EOF && (isblank (ch) || ch == '\n' || ch == ';')); if (ch == EOF) break; savchar (ch); if (vector->v_length == vector->v_allocated) { vector->v = ((struct sed_cmd *) ck_realloc ((VOID *) vector->v, ((vector->v_length + MORE_CMDS) * sizeof (struct sed_cmd)))); vector->v_allocated += MORE_CMDS; } cur_cmd = vector->v + vector->v_length; vector->v_length++; cur_cmd->a1.addr_type = 0; cur_cmd->a2.addr_type = 0; cur_cmd->aflags = 0; cur_cmd->cmd = 0; if (compile_address (&(cur_cmd->a1))) { ch = inchar (); if (ch == ',') { do ch = inchar (); while (ch != EOF && isblank (ch)); savchar (ch); if (compile_address (&(cur_cmd->a2))) ; else bad_prog ("Unexpected ','"); } else savchar (ch); } if (cur_cmd->a1.addr_type == addr_is_num && cur_cmd->a2.addr_type == addr_is_num && cur_cmd->a2.addr_number < cur_cmd->a1.addr_number) cur_cmd->a2.addr_number = cur_cmd->a1.addr_number; ch = inchar (); if (ch == EOF) bad_prog (NO_COMMAND); new_cmd: switch (ch) { case '#': if (cur_cmd->a1.addr_type != 0) bad_prog (NO_ADDR); do { ch = inchar (); if (ch == '\\') ch = inchar (); } while (ch != EOF && ch != '\n'); vector->v_length--; goto skip_comment; case '!': if (cur_cmd->aflags & ADDR_BANG_BIT) bad_prog ("Multiple '!'s"); cur_cmd->aflags |= ADDR_BANG_BIT; do ch = inchar (); while (ch != EOF && isblank (ch)); if (ch == EOF) bad_prog (NO_COMMAND); #if 0 savchar (ch); #endif goto new_cmd; case 'a': case 'i': if (cur_cmd->a2.addr_type != 0) bad_prog (ONE_ADDR); /* Fall Through */ case 'c': cur_cmd->cmd = ch; if (inchar () != '\\' || inchar () != '\n') bad_prog (LINE_JUNK); b = init_buffer (); while ((ch = inchar ()) != EOF && ch != '\n') { if (ch == '\\') ch = inchar (); add1_buffer (b, ch); } if (ch != EOF) add1_buffer (b, ch); num = size_buffer (b); string = (unsigned char *) ck_malloc (num); bcopy (get_buffer (b), string, num); flush_buffer (b); cur_cmd->x.cmd_txt.text_len = num; cur_cmd->x.cmd_txt.text = (char *) string; break; case '{': cur_cmd->cmd = ch; program_depth++; #if 0 while ((ch = inchar ()) != EOF && ch != '\n') if (!isblank (ch)) bad_prog (LINE_JUNK); #endif cur_cmd->x.sub = compile_program ((struct vector *) 0, prog_line); /* FOO JF is this the right thing to do? almost. don't forget a return addr. -t */ cur_cmd->x.sub->return_v = vector; cur_cmd->x.sub->return_i = vector->v_length - 1; break; case '}': if (!program_depth) bad_prog ("Unexpected '}'"); --program_depth; /* a return insn for subprograms -t */ cur_cmd->cmd = ch; if (cur_cmd->a1.addr_type != 0) bad_prog ("} doesn't want any addresses"); while ((ch = inchar ()) != EOF && ch != '\n' && ch != ';') if (!isblank (ch)) bad_prog (LINE_JUNK); return vector; case ':': cur_cmd->cmd = ch; if (cur_cmd->a1.addr_type != 0) bad_prog (": doesn't want any addresses"); labels = setup_jump (labels, cur_cmd, vector); break; case 'b': case 't': cur_cmd->cmd = ch; jumps = setup_jump (jumps, cur_cmd, vector); break; case 'q': case '=': if (cur_cmd->a2.addr_type) bad_prog (ONE_ADDR); /* Fall Through */ case 'd': case 'D': case 'g': case 'G': case 'h': case 'H': case 'l': case 'n': case 'N': case 'p': case 'P': case 'x': cur_cmd->cmd = ch; do ch = inchar (); while (ch != EOF && isblank (ch) && ch != '\n' && ch != ';'); if (ch != '\n' && ch != ';' && ch != EOF) bad_prog (LINE_JUNK); break; case 'r': if (cur_cmd->a2.addr_type != 0) bad_prog (ONE_ADDR); /* FALL THROUGH */ case 'w': cur_cmd->cmd = ch; cur_cmd->x.io_file = compile_filename (ch == 'r'); break; case 's': cur_cmd->cmd = ch; slash = inchar (); buffer_regex (slash); compile_regex (); cur_cmd->x.cmd_regex.regx = last_regex; b = init_buffer (); while (((ch = inchar ()) != EOF) && (ch != slash) && (ch != '\n')) { if (ch == '\\') { int ci; ci = inchar (); if (ci != EOF) { if (ci != '\n') add1_buffer (b, ch); add1_buffer (b, ci); } } else add1_buffer (b, ch); } if (ch != slash) { if (ch == '\n' && prog_line > 1) --prog_line; bad_prog ("Unterminated `s' command"); } cur_cmd->x.cmd_regex.replace_length = size_buffer (b); cur_cmd->x.cmd_regex.replacement = ck_malloc (cur_cmd->x.cmd_regex.replace_length); bcopy (get_buffer (b), cur_cmd->x.cmd_regex.replacement, cur_cmd->x.cmd_regex.replace_length); flush_buffer (b); cur_cmd->x.cmd_regex.flags = 0; cur_cmd->x.cmd_regex.numb = 0; if (ch == EOF) break; do { ch = inchar (); switch (ch) { case 'p': if (cur_cmd->x.cmd_regex.flags & S_PRINT_BIT) bad_prog ("multiple 'p' options to 's' command"); cur_cmd->x.cmd_regex.flags |= S_PRINT_BIT; break; case 'g': if (cur_cmd->x.cmd_regex.flags & S_NUM_BIT) cur_cmd->x.cmd_regex.flags &= ~S_NUM_BIT; if (cur_cmd->x.cmd_regex.flags & S_GLOBAL_BIT) bad_prog ("multiple 'g' options to 's' command"); cur_cmd->x.cmd_regex.flags |= S_GLOBAL_BIT; break; case 'w': cur_cmd->x.cmd_regex.flags |= S_WRITE_BIT; cur_cmd->x.cmd_regex.wio_file = compile_filename (0); ch = '\n'; break; case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9': if (cur_cmd->x.cmd_regex.flags & S_NUM_BIT) bad_prog ("multiple number options to 's' command"); if ((cur_cmd->x.cmd_regex.flags & S_GLOBAL_BIT) == 0) cur_cmd->x.cmd_regex.flags |= S_NUM_BIT; num = 0; while (isdigit (ch)) { num = num * 10 + ch - '0'; ch = inchar (); } savchar (ch); cur_cmd->x.cmd_regex.numb = num; break; case '\n': case ';': case EOF: break; default: bad_prog ("Unknown option to 's'"); break; } } while (ch != EOF && ch != '\n' && ch != ';'); if (ch == EOF) break; break; case 'y': cur_cmd->cmd = ch; string = (unsigned char *) ck_malloc (256); for (num = 0; num < 256; num++) string[num] = num; b = init_buffer (); slash = inchar (); while ((ch = inchar ()) != EOF && ch != slash) add1_buffer (b, ch); cur_cmd->x.translate = string; string = (unsigned char *) get_buffer (b); for (num = size_buffer (b); num; --num) { ch = inchar (); if (ch == EOF) bad_prog (BAD_EOF); if (ch == slash) bad_prog ("strings for y command are different lengths"); cur_cmd->x.translate[*string++] = ch; } flush_buffer (b); if (inchar () != slash || ((ch = inchar ()) != EOF && ch != '\n' && ch != ';')) bad_prog (LINE_JUNK); break; default: bad_prog ("Unknown command"); } } if (program_depth) { prog_line = open_line; bad_prog ("Unmatched `{'"); } return vector; } /* Complain about a programming error and exit. */ void bad_prog (why) char *why; { if (prog_line > 0) fprintf (stderr, "%s: file %s line %d: %s\n", myname, prog_name, prog_line, why); else fprintf (stderr, "%s: %s\n", myname, why); exit (1); } /* Read the next character from the program. Return EOF if there isn't anything to read. Keep prog_line up to date, so error messages can be meaningful. */ int inchar () { int ch; if (prog_file) { if (feof (prog_file)) return EOF; else ch = getc (prog_file); } else { if (!prog_cur) return EOF; else if (prog_cur == prog_end) { ch = EOF; prog_cur = 0; } else ch = *prog_cur++; } if ((ch == '\n') && prog_line) prog_line++; return ch; } /* unget 'ch' so the next call to inchar will return it. 'ch' must not be EOF or anything nasty like that. */ void savchar (ch) int ch; { if (ch == EOF) return; if (ch == '\n' && prog_line > 1) --prog_line; if (prog_file) ungetc (ch, prog_file); else *--prog_cur = ch; } /* Try to read an address for a sed command. If it succeeeds, return non-zero and store the resulting address in *'addr'. If the input doesn't look like an address read nothing and return zero. */ int compile_address (addr) struct addr *addr; { int ch; int num; ch = inchar (); if (isdigit (ch)) { num = ch - '0'; while ((ch = inchar ()) != EOF && isdigit (ch)) num = num * 10 + ch - '0'; if (ch == '~') { addr->addr_type = addr_is_mod; addr->offset = num; ch = inchar(); num=0; if (isdigit(ch)) { num = ch - '0'; while ((ch = inchar ()) != EOF && isdigit (ch)) num = num * 10 + ch - '0'; addr->modulo = num; } addr->modulo += (addr->modulo==0); } else { addr->addr_type = addr_is_num; addr->addr_number = num; } while (ch != EOF && isblank (ch)) ch = inchar (); savchar (ch); return 1; } else if (ch == '/' || ch == '\\') { addr->addr_type = addr_is_regex; if (ch == '\\') ch = inchar (); buffer_regex (ch); compile_regex (); addr->addr_regex = last_regex; do ch = inchar (); while (ch != EOF && isblank (ch)); savchar (ch); return 1; } else if (ch == '$') { addr->addr_type = addr_is_last; do ch = inchar (); while (ch != EOF && isblank (ch)); savchar (ch); return 1; } else savchar (ch); return 0; } void buffer_regex (slash) int slash; { VOID *b; int ch; int char_class_pos = -1; b = init_buffer (); while ((ch = inchar ()) != EOF && (ch != slash || (char_class_pos >= 0))) { if (ch == '^') { #if 0 if (size_buffer (b) == 0) { add1_buffer (b, '\\'); add1_buffer (b, '`'); } else add1_buffer (b, ch); #else add1_buffer (b, ch); #endif continue; } else if (ch == '$') { #if 0 ch = inchar (); savchar (ch); if (ch == slash) { add1_buffer (b, '\\'); add1_buffer (b, '\''); } else add1_buffer (b, '$'); #else add1_buffer (b, '$'); #endif continue; } else if (ch == '[') { if (char_class_pos < 0) char_class_pos = size_buffer (b); add1_buffer (b, ch); continue; } else if (ch == ']') { add1_buffer (b, ch); { char * regexp = get_buffer (b); int pos = size_buffer (b) - 1; if (!( (char_class_pos >= 0) && ( (pos == char_class_pos + 1) || ( (pos == char_class_pos + 2) && (regexp[char_class_pos + 1] == '^'))))) char_class_pos = -1; continue; } } else if (ch != '\\' || (char_class_pos >= 0)) { add1_buffer (b, ch); continue; } ch = inchar (); switch (ch) { case 'n': add1_buffer (b, '\n'); break; #if 0 case 'b': add1_buffer (b, '\b'); break; case 'f': add1_buffer (b, '\f'); break; case 'r': add1_buffer (b, '\r'); break; case 't': add1_buffer (b, '\t'); break; #endif /* 0 */ case EOF: break; default: add1_buffer (b, '\\'); add1_buffer (b, ch); break; } } if ((ch == EOF) && rx_testing) { /* This is slightly bogus. Sed is noticing * an illformed regexp, but rx is getting * the credit. Fortunately, this only effects * a few spencer tests. */ add1_buffer (b, 0); printf ("bad regexp: %s\n", b); exit (0); } if (ch == EOF) bad_prog (BAD_EOF); if (size_buffer (b)) { if (last_regex_string) last_regex_string = (char *)ck_realloc (last_regex_string, size_buffer (b) + 1); else last_regex_string = (char *)ck_malloc (size_buffer (b) + 1); bcopy (get_buffer (b), last_regex_string, size_buffer (b)); last_regex_string [size_buffer (b)] = 0; } else if (!last_regex) bad_prog (NO_REGEX); flush_buffer (b); } void compile_regex () { int cflags; int error; last_regex = ((struct rx_posix_regex *) ck_malloc (sizeof (struct rx_posix_regex))); bzero (last_regex, sizeof (*last_regex)); cflags = (use_extended_syntax_p ? REG_EXTENDED : 0); #if 0 last_regex->fastmap = ck_malloc (256); #endif error = regcomp (last_regex, last_regex_string, cflags); if (error && rx_testing) { printf ("bad regexp: %s\n", last_regex_string); exit(0); } if (error) bad_prog ((char *)rx_error_msg[error]); } /* Store a label (or label reference) created by a ':', 'b', or 't' comand so that the jump to/from the lable can be backpatched after compilation is complete */ struct sed_label * setup_jump (list, cmd, vec) struct sed_label *list; struct sed_cmd *cmd; struct vector *vec; { struct sed_label *tmp; VOID *b; int ch; b = init_buffer (); while ((ch = inchar ()) != EOF && isblank (ch)) ; /* Possible non posixicity. */ while (ch != EOF && ch != '\n' && (!isblank (ch)) && ch != ';' && ch != '}') { add1_buffer (b, ch); ch = inchar (); } savchar (ch); add1_buffer (b, '\0'); tmp = (struct sed_label *) ck_malloc (sizeof (struct sed_label)); tmp->v = vec; tmp->v_index = cmd - vec->v; tmp->name = ck_strdup (get_buffer (b)); tmp->next = list; flush_buffer (b); return tmp; } /* read in a filename for a 'r', 'w', or 's///w' command, and update the internal structure about files. The file is opened if it isn't already open. */ FILE * compile_filename (readit) int readit; { char *file_name; int n; VOID *b; int ch; if (inchar () != ' ') bad_prog ("missing ' ' before filename"); b = init_buffer (); while ((ch = inchar ()) != EOF && ch != '\n') add1_buffer (b, ch); add1_buffer (b, '\0'); file_name = get_buffer (b); for (n = 0; n < NUM_FPS; n++) { if (!file_ptrs[n].name) break; } if (n < NUM_FPS) { file_ptrs[n].name = ck_strdup (file_name); if (!readit) { if (!file_ptrs[n].for_write) file_ptrs[n].for_write = ck_fopen (file_name, "w"); } else { if (!file_ptrs[n].for_read) file_ptrs[n].for_read = fopen (file_name, "r"); } flush_buffer (b); return readit ? file_ptrs[n].for_read : file_ptrs[n].for_write; } else { bad_prog ("Hopelessely evil compiled in limit on number of open files. re-compile sed"); return 0; } } /* Read a file and apply the compiled script to it. */ void read_file (name) char *name; { if (*name == '-' && name[1] == '\0') input_file = stdin; else { input_file = fopen (name, "r"); if (input_file == 0) { extern int errno; extern char *sys_errlist[]; extern int sys_nerr; char *ptr; ptr = ((errno >= 0 && errno < sys_nerr) ? sys_errlist[errno] : "Unknown error code"); bad_input++; fprintf (stderr, "%s: can't read %s: %s\n", myname, name, ptr); return; } } while (read_pattern_space ()) { execute_program (the_program); if (!no_default_output) ck_fwrite (line.text, 1, line.length, stdout); if (append.length) { ck_fwrite (append.text, 1, append.length, stdout); append.length = 0; } if (quit_cmd) break; } ck_fclose (input_file); } static char * eol_pos (str, len) char *str; int len; { while (len--) if (*str++ == '\n') return --str; return --str; } static void chr_copy (dest, src, len) char *dest; char *src; int len; { while (len--) *dest++ = *src++; } /* Execute the program 'vec' on the current input line. */ /* fixme */ static struct rx_registers regs[256]; void execute_program (vec) struct vector *vec; { struct sed_cmd *cur_cmd; int n; int addr_matched; static int end_cycle; static struct line tmp; struct line t; struct vector *restart_vec = vec; restart: vec = restart_vec; end_cycle = 0; for (cur_cmd = vec->v, n = vec->v_length; n; cur_cmd++, n--) { exe_loop: addr_matched = 0; if (cur_cmd->aflags & A1_MATCHED_BIT) { addr_matched = 1; if (match_address (&(cur_cmd->a2))) cur_cmd->aflags &= ~A1_MATCHED_BIT; } else if (match_address (&(cur_cmd->a1))) { addr_matched = 1; if (cur_cmd->a2.addr_type != addr_is_null) if ( (cur_cmd->a2.addr_type == addr_is_regex) || !match_address (&(cur_cmd->a2))) cur_cmd->aflags |= A1_MATCHED_BIT; } if (cur_cmd->aflags & ADDR_BANG_BIT) addr_matched = !addr_matched; if (!addr_matched) continue; switch (cur_cmd->cmd) { case '{': /* Execute sub-program */ if (cur_cmd->x.sub->v_length) { vec = cur_cmd->x.sub; cur_cmd = vec->v; n = vec->v_length; goto exe_loop; } break; case '}': cur_cmd = vec->return_v->v + vec->return_i; n = vec->return_v->v_length - vec->return_i; vec = vec->return_v; break; case ':': /* Executing labels is easy. */ break; case '=': printf ("%d\n", input_line_number); break; case 'a': while (append.alloc - append.length < cur_cmd->x.cmd_txt.text_len) { append.alloc *= 2; append.text = ck_realloc (append.text, append.alloc); } bcopy (cur_cmd->x.cmd_txt.text, append.text + append.length, cur_cmd->x.cmd_txt.text_len); append.length += cur_cmd->x.cmd_txt.text_len; break; case 'b': if (!cur_cmd->x.jump) end_cycle++; else { struct sed_label *j = cur_cmd->x.jump; n = j->v->v_length - j->v_index; cur_cmd = j->v->v + j->v_index; vec = j->v; goto exe_loop; } break; case 'c': line.length = 0; if (!((cur_cmd->aflags & A1_MATCHED_BIT))) ck_fwrite (cur_cmd->x.cmd_txt.text, 1, cur_cmd->x.cmd_txt.text_len, stdout); end_cycle++; break; case 'd': line.length = 0; end_cycle++; break; case 'D': { char *tmp; int newlength; tmp = eol_pos (line.text, line.length); newlength = line.length - (tmp - line.text) - 1; if (newlength) { chr_copy (line.text, tmp + 1, newlength); line.length = newlength; goto restart; } line.length = 0; end_cycle++; } break; case 'g': line_copy (&hold, &line); break; case 'G': line_append (&hold, &line); break; case 'h': line_copy (&line, &hold); break; case 'H': line_append (&line, &hold); break; case 'i': ck_fwrite (cur_cmd->x.cmd_txt.text, 1, cur_cmd->x.cmd_txt.text_len, stdout); break; case 'l': { char *tmp; int n; int width = 0; n = line.length; tmp = line.text; while (n--) { /* Skip the trailing newline, if there is one */ if (!n && (*tmp == '\n')) break; if (width > 77) { width = 0; putchar ('\n'); } if (*tmp == '\\') { printf ("\\\\"); width += 2; } else if (isprint (*tmp)) { putchar (*tmp); width++; } else switch (*tmp) { #if 0 /* Should print \00 instead of \0 because (a) POSIX */ /* requires it, and (b) this way \01 is unambiguous. */ case '\0': printf ("\\0"); width += 2; break; #endif case 007: printf ("\\a"); width += 2; break; case '\b': printf ("\\b"); width += 2; break; case '\f': printf ("\\f"); width += 2; break; case '\n': printf ("\\n"); width += 2; break; case '\r': printf ("\\r"); width += 2; break; case '\t': printf ("\\t"); width += 2; break; case '\v': printf ("\\v"); width += 2; break; default: printf ("\\%02x", (*tmp) & 0xFF); width += 2; break; } tmp++; } putchar ('\n'); } break; case 'n': if (feof (input_file)) goto quit; if (!no_default_output) ck_fwrite (line.text, 1, line.length, stdout); read_pattern_space (); break; case 'N': if (feof (input_file)) { line.length = 0; goto quit; } append_pattern_space (); break; case 'p': ck_fwrite (line.text, 1, line.length, stdout); break; case 'P': { char *tmp; tmp = eol_pos (line.text, line.length); ck_fwrite (line.text, 1, tmp ? tmp - line.text + 1 : line.length, stdout); } break; case 'q': quit: quit_cmd++; end_cycle++; break; case 'r': { int n = 0; if (cur_cmd->x.io_file) { rewind (cur_cmd->x.io_file); do { append.length += n; if (append.length == append.alloc) { append.alloc *= 2; append.text = ck_realloc (append.text, append.alloc); } n = fread (append.text + append.length, sizeof (char), append.alloc - append.length, cur_cmd->x.io_file); } while (n > 0); if (ferror (cur_cmd->x.io_file)) panic ("Read error on input file to 'r' command"); } } break; case 's': { int start; /* where to start scan for match in LINE */ int remain; /* length after START, sans trailing newline */ int offset; /* where in LINE a match was found */ int trail_nl_p; /* does line have a trailing newline? */ char *rep; /* the replacement string */ char *rep_end; /* end of the replacement string */ int count; /* number of matches found */ trail_nl_p = (line.text [line.length - 1] == '\n'); /* tmp will accumulate the result of the s command. */ if (!tmp.alloc) { tmp.alloc = 50; tmp.text = ck_malloc (50); } tmp.length = 0; count = 0; start = 0; remain = line.length - trail_nl_p; rep = cur_cmd->x.cmd_regex.replacement; rep_end = rep + cur_cmd->x.cmd_regex.replace_length; while (!regnexec (cur_cmd->x.cmd_regex.regx, remain, line.text + start, sizeof (regs) / sizeof (regs[0]), regs, 0)) { count++; offset = regs[0].rm_so + start; /* * +- line.text +-[offset] * V V * "blah blah blah xyzzy blah blah blah blah" * ^ ^ ^ * +-[start] +-[end] +-[start + remain] * * * regs[0].rm_so == offset - start * regs[0].rm_eo == end - start */ /* Copy stuff to the left of the next match into the output * string. */ if (offset - start) str_append (&tmp, line.text + start, offset - start); /* If we're counting up to the Nth match, are we there yet? */ if ( (cur_cmd->x.cmd_regex.flags & S_NUM_BIT) && (count != cur_cmd->x.cmd_regex.numb)) { /* Not there yet...so skip this match. */ int matched; /* n characters matched */ matched = regs[0].rm_eo - regs[0].rm_so; /* If the match was vacuous, skip ahead one character * anyway. It isn't at all obvious to me that this is * the right behavior for this case. -t */ if (!matched) matched = 1; str_append (&tmp, line.text + offset, matched); start = offset + matched; remain = (line.length - trail_nl_p) - start; continue; } /* Expand the replacement string into the output string. */ { char *rep_cur; /* next burst being processed in replacement */ char *rep_next; /* end of burst in replacement */ /* Replacement strings can be viewed as a sequence of variable * length commands. A command can be a literal string or a * backreference (either numeric or "&"). * * This loop measures off the next command between * REP_CUR and REP_NEXT, handles that command, and loops. */ for (rep_next = rep_cur = rep; rep_next < rep_end; rep_next++) { if (*rep_next == '\\') { /* Preceding the backslash may be some literal * text: */ if (rep_next - rep_cur) str_append (&tmp, rep_cur, rep_next - rep_cur); /* Skip the backslash itself and look for * a numeric back-reference: */ rep_next++; if ( (rep_next != rep_end) && (*rep_next >= '0' && *rep_next <= '9')) { int n; n = *rep_next - '0'; str_append (&tmp, line.text + start + regs[n].rm_so, regs[n].rm_eo - regs[n].rm_so); } else str_append (&tmp, rep_next, 1); rep_cur = rep_next + 1; } else if (*rep_next == '&') { /* Preceding the ampersand may be some literal * text: */ if (rep_next - rep_cur) str_append (&tmp, rep_cur, rep_next - rep_cur); str_append (&tmp, line.text + start + regs[0].rm_so, regs[0].rm_eo - regs[0].rm_so); rep_cur = rep_next + 1; } } /* There may be a trailing literal in the replacement string. */ if (rep_next - rep_cur) str_append (&tmp, rep_cur, rep_next - rep_cur); } /* If the match was vacuous, skip over one character * and add that character to the output. */ if (regs[0].rm_so == regs[0].rm_eo) { str_append (&tmp, line.text + offset, 1); ++(regs[0].rm_eo); } start = start + regs[0].rm_eo; remain = (line.length - trail_nl_p) - start; if (remain < 0) break; if (!(cur_cmd->x.cmd_regex.flags & S_GLOBAL_BIT)) break; } if (!count) break; replaced = 1; str_append (&tmp, line.text + start, remain + trail_nl_p); t.text = line.text; t.length = line.length; t.alloc = line.alloc; line.text = tmp.text; line.length = tmp.length; line.alloc = tmp.alloc; tmp.text = t.text; tmp.length = t.length; tmp.alloc = t.alloc; if ((cur_cmd->x.cmd_regex.flags & S_WRITE_BIT) && cur_cmd->x.cmd_regex.wio_file) ck_fwrite (line.text, 1, line.length, cur_cmd->x.cmd_regex.wio_file); if (cur_cmd->x.cmd_regex.flags & S_PRINT_BIT) ck_fwrite (line.text, 1, line.length, stdout); break; } case 't': if (replaced) { replaced = 0; if (!cur_cmd->x.jump) end_cycle++; else { struct sed_label *j = cur_cmd->x.jump; n = j->v->v_length - j->v_index; cur_cmd = j->v->v + j->v_index; vec = j->v; goto exe_loop; } } break; case 'w': if (cur_cmd->x.io_file) { ck_fwrite (line.text, 1, line.length, cur_cmd->x.io_file); fflush (cur_cmd->x.io_file); } break; case 'x': { struct line tmp; tmp = line; line = hold; hold = tmp; } break; case 'y': { unsigned char *p, *e; for (p = (unsigned char *) (line.text), e = p + line.length; p < e; p++) *p = cur_cmd->x.translate[*p]; } break; default: panic ("INTERNAL ERROR: Bad cmd %c", cur_cmd->cmd); } if (end_cycle) break; } } /* Return non-zero if the current line matches the address pointed to by 'addr'. */ int match_address (addr) struct addr *addr; { switch (addr->addr_type) { case addr_is_null: return 1; case addr_is_num: return (input_line_number == addr->addr_number); case addr_is_mod: return ((input_line_number%addr->modulo) == addr->offset); case addr_is_regex: { int trail_nl_p; int match; trail_nl_p = line.text [line.length - 1] == '\n'; match = !regnexec (addr->addr_regex, line.length - trail_nl_p, line.text, 0, 0, 0); return match; } case addr_is_last: return (input_EOF) ? 1 : 0; default: panic ("INTERNAL ERROR: bad address type"); break; } return -1; } /* Read in the next line of input, and store it in the pattern space. Return non-zero if this is the last line of input */ int read_pattern_space () { int n; char *p; int ch; p = line.text; n = line.alloc; if (feof (input_file)) return 0; input_line_number++; replaced = 0; for (;;) { if (n == 0) { line.text = ck_realloc (line.text, line.alloc * 2); p = line.text + line.alloc; n = line.alloc; line.alloc *= 2; } ch = getc (input_file); if (ch == EOF) { if (n == line.alloc) return 0; /* *p++ = '\n'; */ /* --n; */ line.length = line.alloc - n; if (last_input_file) input_EOF++; return 1; } *p++ = ch; --n; if (ch == '\n') { line.length = line.alloc - n; break; } } ch = getc (input_file); if (ch != EOF) ungetc (ch, input_file); else if (last_input_file) input_EOF++; return 1; } /* Inplement the 'N' command, which appends the next line of input to the pattern space. */ void append_pattern_space () { char *p; int n; int ch; p = line.text + line.length; n = line.alloc - line.length; input_line_number++; replaced = 0; for (;;) { ch = getc (input_file); if (ch == EOF) { if (n == line.alloc) return; /* *p++ = '\n'; */ /* --n; */ line.length = line.alloc - n; if (last_input_file) input_EOF++; return; } if (n == 0) { line.text = ck_realloc (line.text, line.alloc * 2); p = line.text + line.alloc; n = line.alloc; line.alloc *= 2; } *p++ = ch; --n; if (ch == '\n') { line.length = line.alloc - n; break; } } ch = getc (input_file); if (ch != EOF) ungetc (ch, input_file); else if (last_input_file) input_EOF++; } /* Copy the contents of the line 'from' into the line 'to'. This destroys the old contents of 'to'. It will still work if the line 'from' contains nulls. */ void line_copy (from, to) struct line *from, *to; { if (from->length > to->alloc) { to->alloc = from->length; to->text = ck_realloc (to->text, to->alloc); } bcopy (from->text, to->text, from->length); to->length = from->length; } /* Append the contents of the line 'from' to the line 'to'. This routine will work even if the line 'from' contains nulls */ void line_append (from, to) struct line *from, *to; { if (from->length > (to->alloc - to->length)) { to->alloc += from->length; to->text = ck_realloc (to->text, to->alloc); } bcopy (from->text, to->text + to->length, from->length); to->length += from->length; } /* Append 'length' bytes from 'string' to the line 'to' This routine *will* append bytes with nulls in them, without failing. */ void str_append (to, string, length) struct line *to; char *string; int length; { if (length > to->alloc - to->length) { to->alloc += length; to->text = ck_realloc (to->text, to->alloc); } bcopy (string, to->text + to->length, length); to->length += length; } void usage (status) int status; { fprintf (status ? stderr : stdout, "\ Usage: %s [-nV] [--quiet] [--silent] [--version] [-e script]\n\ [-f script-file] [--expression=script] [--file=script-file] [file...]\n", myname); exit (status); } dist/sed/utils.c100660 765 144 12511 6071414316 12465 0ustar lordusers/* Functions from hack's utils library. Copyright (C) 1989, 1990, 1991 Free Software Foundation, Inc. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */ /* These routines were written as part of a library (by hack), but since most people don't have the library, here they are. */ #ifdef __STDC__ #define VOID void #else #define VOID char #endif #include #if HAVE_STRING_H || defined(STDC_HEADERS) #include #else #include #endif #if defined(STDC_HEADERS) #include #else #ifdef RX_MEMDBUG #include #include #else VOID *malloc(); VOID *realloc(); #endif /* ndef RX_MEMDBUG */ #endif VOID *ck_malloc(); char *myname; #ifdef __STDC__ #include /* Print an error message and exit */ void panic(char *str, ...) { va_list iggy; fprintf(stderr,"%s: ",myname); va_start(iggy,str); #ifdef HAVE_VPRINTF vfprintf(stderr,str,iggy); #else #ifdef HAVE_DOPRNT _doprnt(str,&iggy,stderr); #endif #endif va_end(iggy); putc('\n',stderr); exit(4); } #else #include void panic(str,va_alist) char *str; va_dcl { va_list iggy; fprintf(stderr,"%s: ",myname); va_start(iggy); #ifdef HAVE_VPRINTF vfprintf(stderr,str,iggy); #else #ifdef HAVE_DOPRNT _doprnt(str,&iggy,stderr); #endif #endif va_end(iggy); putc('\n',stderr); exit(4); } #endif /* Store information about files opened with ck_fopen so that error messages from ck_fread, etc can print the name of the file that had the error */ #define N_FILE 32 struct id { FILE *fp; char *name; }; static struct id __id_s[N_FILE]; /* Internal routine to get a filename from __id_s */ char * __fp_name(fp) FILE *fp; { int n; for(n=0;nallocated=MIN_ALLOCATE; b->b=(char *)ck_malloc(MIN_ALLOCATE); b->length=0; return (VOID *)b; } void flush_buffer(bb) VOID *bb; { struct buffer *b; b=(struct buffer *)bb; free(b->b); b->b=0; b->allocated=0; b->length=0; free(b); } int size_buffer(b) VOID *b; { struct buffer *bb; bb=(struct buffer *)b; return bb->length; } void add_buffer(bb,p,n) VOID *bb; char *p; int n; { struct buffer *b; int x; char * cp; b=(struct buffer *)bb; if(b->length+n>b->allocated) { b->allocated = (b->length + n) * 2; b->b=(char *)ck_realloc(b->b,b->allocated); } x = n; cp = b->b + b->length; while (x--) *cp++ = *p++; b->length+=n; } void add1_buffer(bb,ch) VOID *bb; int ch; { struct buffer *b; b=(struct buffer *)bb; if(b->length+1>b->allocated) { b->allocated*=2; b->b=(char *)ck_realloc(b->b,b->allocated); } b->b[b->length]=ch; b->length++; } char * get_buffer(bb) VOID *bb; { struct buffer *b; b=(struct buffer *)bb; return b->b; } dist/sed/getopt.c100660 765 144 50626 6070131465 12640 0ustar lordusers/* Getopt for GNU. NOTE: getopt is now part of the C library, so if you don't know what "Keep this file name-space clean" means, talk to roland@gnu.ai.mit.edu before changing it! Copyright (C) 1987, 1988, 1989, 1990, 1991, 1992, 1993 Free Software Foundation, Inc. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */ /* NOTE!!! AIX requires this to be the first thing in the file. Do not put ANYTHING before it! */ #if !defined (__GNUC__) && defined (_AIX) #pragma alloca #endif #ifdef HAVE_CONFIG_H #include "config.h" #endif #ifdef __GNUC__ #define alloca __builtin_alloca #else /* not __GNUC__ */ #if defined (HAVE_ALLOCA_H) || (defined(sparc) && (defined(sun) || (!defined(USG) && !defined(SVR4) && !defined(__svr4__)))) #include #else #ifndef _AIX char *alloca (); #endif #endif /* alloca.h */ #endif /* not __GNUC__ */ #if !__STDC__ && !defined(const) && IN_GCC #define const #endif /* This tells Alpha OSF/1 not to define a getopt prototype in . */ #ifndef _NO_PROTO #define _NO_PROTO #endif #include /* Comment out all this code if we are using the GNU C Library, and are not actually compiling the library itself. This code is part of the GNU C Library, but also included in many other GNU distributions. Compiling and linking in this code is a waste when using the GNU C library (especially if it is a shared library). Rather than having every GNU program understand `configure --with-gnu-libc' and omit the object files, it is simpler to just do this in the source for each such file. */ #if defined (_LIBC) || !defined (__GNU_LIBRARY__) /* This needs to come after some library #include to get __GNU_LIBRARY__ defined. */ #ifdef __GNU_LIBRARY__ #undef alloca /* Don't include stdlib.h for non-GNU C libraries because some of them contain conflicting prototypes for getopt. */ #include #else /* Not GNU C library. */ #define __alloca alloca #endif /* GNU C library. */ /* If GETOPT_COMPAT is defined, `+' as well as `--' can introduce a long-named option. Because this is not POSIX.2 compliant, it is being phased out. */ /* #define GETOPT_COMPAT */ /* This version of `getopt' appears to the caller like standard Unix `getopt' but it behaves differently for the user, since it allows the user to intersperse the options with the other arguments. As `getopt' works, it permutes the elements of ARGV so that, when it is done, all the options precede everything else. Thus all application programs are extended to handle flexible argument order. Setting the environment variable POSIXLY_CORRECT disables permutation. Then the behavior is completely standard. GNU application programs can use a third alternative mode in which they can distinguish the relative order of options and other arguments. */ #include "getopt.h" /* For communication from `getopt' to the caller. When `getopt' finds an option that takes an argument, the argument value is returned here. Also, when `ordering' is RETURN_IN_ORDER, each non-option ARGV-element is returned here. */ char *optarg = 0; /* Index in ARGV of the next element to be scanned. This is used for communication to and from the caller and for communication between successive calls to `getopt'. On entry to `getopt', zero means this is the first call; initialize. When `getopt' returns EOF, this is the index of the first of the non-option elements that the caller should itself scan. Otherwise, `optind' communicates from one call to the next how much of ARGV has been scanned so far. */ /* XXX 1003.2 says this must be 1 before any call. */ int optind = 0; /* The next char to be scanned in the option-element in which the last option character we returned was found. This allows us to pick up the scan where we left off. If this is zero, or a null string, it means resume the scan by advancing to the next ARGV-element. */ static char *nextchar; /* Callers store zero here to inhibit the error message for unrecognized options. */ int opterr = 1; /* Set to an option character which was unrecognized. This must be initialized on some systems to avoid linking in the system's own getopt implementation. */ int optopt = '?'; /* Describe how to deal with options that follow non-option ARGV-elements. If the caller did not specify anything, the default is REQUIRE_ORDER if the environment variable POSIXLY_CORRECT is defined, PERMUTE otherwise. REQUIRE_ORDER means don't recognize them as options; stop option processing when the first non-option is seen. This is what Unix does. This mode of operation is selected by either setting the environment variable POSIXLY_CORRECT, or using `+' as the first character of the list of option characters. PERMUTE is the default. We permute the contents of ARGV as we scan, so that eventually all the non-options are at the end. This allows options to be given in any order, even with programs that were not written to expect this. RETURN_IN_ORDER is an option available to programs that were written to expect options and other ARGV-elements in any order and that care about the ordering of the two. We describe each non-option ARGV-element as if it were the argument of an option with character code 1. Using `-' as the first character of the list of option characters selects this mode of operation. The special argument `--' forces an end of option-scanning regardless of the value of `ordering'. In the case of RETURN_IN_ORDER, only `--' can cause `getopt' to return EOF with `optind' != ARGC. */ static enum { REQUIRE_ORDER, PERMUTE, RETURN_IN_ORDER } ordering; #ifdef __GNU_LIBRARY__ /* We want to avoid inclusion of string.h with non-GNU libraries because there are many ways it can cause trouble. On some systems, it contains special magic macros that don't work in GCC. */ #include #define my_index strchr #define my_bcopy(src, dst, n) memcpy ((dst), (src), (n)) #else /* Avoid depending on library functions or files whose names are inconsistent. */ char *getenv (); static char * my_index (str, chr) const char *str; int chr; { while (*str) { if (*str == chr) return (char *) str; str++; } return 0; } static void my_bcopy (from, to, size) const char *from; char *to; int size; { int i; for (i = 0; i < size; i++) to[i] = from[i]; } #endif /* GNU C library. */ /* Handle permutation of arguments. */ /* Describe the part of ARGV that contains non-options that have been skipped. `first_nonopt' is the index in ARGV of the first of them; `last_nonopt' is the index after the last of them. */ static int first_nonopt; static int last_nonopt; /* Exchange two adjacent subsequences of ARGV. One subsequence is elements [first_nonopt,last_nonopt) which contains all the non-options that have been skipped so far. The other is elements [last_nonopt,optind), which contains all the options processed since those non-options were skipped. `first_nonopt' and `last_nonopt' are relocated so that they describe the new indices of the non-options in ARGV after they are moved. */ static void exchange (argv) char **argv; { int nonopts_size = (last_nonopt - first_nonopt) * sizeof (char *); char **temp = (char **) __alloca (nonopts_size); /* Interchange the two blocks of data in ARGV. */ my_bcopy ((char *) &argv[first_nonopt], (char *) temp, nonopts_size); my_bcopy ((char *) &argv[last_nonopt], (char *) &argv[first_nonopt], (optind - last_nonopt) * sizeof (char *)); my_bcopy ((char *) temp, (char *) &argv[first_nonopt + optind - last_nonopt], nonopts_size); /* Update records for the slots the non-options now occupy. */ first_nonopt += (optind - last_nonopt); last_nonopt = optind; } /* Scan elements of ARGV (whose length is ARGC) for option characters given in OPTSTRING. If an element of ARGV starts with '-', and is not exactly "-" or "--", then it is an option element. The characters of this element (aside from the initial '-') are option characters. If `getopt' is called repeatedly, it returns successively each of the option characters from each of the option elements. If `getopt' finds another option character, it returns that character, updating `optind' and `nextchar' so that the next call to `getopt' can resume the scan with the following option character or ARGV-element. If there are no more option characters, `getopt' returns `EOF'. Then `optind' is the index in ARGV of the first ARGV-element that is not an option. (The ARGV-elements have been permuted so that those that are not options now come last.) OPTSTRING is a string containing the legitimate option characters. If an option character is seen that is not listed in OPTSTRING, return '?' after printing an error message. If you set `opterr' to zero, the error message is suppressed but we still return '?'. If a char in OPTSTRING is followed by a colon, that means it wants an arg, so the following text in the same ARGV-element, or the text of the following ARGV-element, is returned in `optarg'. Two colons mean an option that wants an optional arg; if there is text in the current ARGV-element, it is returned in `optarg', otherwise `optarg' is set to zero. If OPTSTRING starts with `-' or `+', it requests different methods of handling the non-option ARGV-elements. See the comments about RETURN_IN_ORDER and REQUIRE_ORDER, above. Long-named options begin with `--' instead of `-'. Their names may be abbreviated as long as the abbreviation is unique or is an exact match for some defined option. If they have an argument, it follows the option name in the same ARGV-element, separated from the option name by a `=', or else the in next ARGV-element. When `getopt' finds a long-named option, it returns 0 if that option's `flag' field is nonzero, the value of the option's `val' field if the `flag' field is zero. The elements of ARGV aren't really const, because we permute them. But we pretend they're const in the prototype to be compatible with other systems. LONGOPTS is a vector of `struct option' terminated by an element containing a name which is zero. LONGIND returns the index in LONGOPT of the long-named option found. It is only valid when a long-named option has been found by the most recent call. If LONG_ONLY is nonzero, '-' as well as '--' can introduce long-named options. */ int _getopt_internal (argc, argv, optstring, longopts, longind, long_only) int argc; char *const *argv; const char *optstring; const struct option *longopts; int *longind; int long_only; { int option_index; optarg = 0; /* Initialize the internal data when the first call is made. Start processing options with ARGV-element 1 (since ARGV-element 0 is the program name); the sequence of previously skipped non-option ARGV-elements is empty. */ if (optind == 0) { first_nonopt = last_nonopt = optind = 1; nextchar = NULL; /* Determine how to handle the ordering of options and nonoptions. */ if (optstring[0] == '-') { ordering = RETURN_IN_ORDER; ++optstring; } else if (optstring[0] == '+') { ordering = REQUIRE_ORDER; ++optstring; } else if (getenv ("POSIXLY_CORRECT") != NULL) ordering = REQUIRE_ORDER; else ordering = PERMUTE; } if (nextchar == NULL || *nextchar == '\0') { if (ordering == PERMUTE) { /* If we have just processed some options following some non-options, exchange them so that the options come first. */ if (first_nonopt != last_nonopt && last_nonopt != optind) exchange ((char **) argv); else if (last_nonopt != optind) first_nonopt = optind; /* Now skip any additional non-options and extend the range of non-options previously skipped. */ while (optind < argc && (argv[optind][0] != '-' || argv[optind][1] == '\0') #ifdef GETOPT_COMPAT && (longopts == NULL || argv[optind][0] != '+' || argv[optind][1] == '\0') #endif /* GETOPT_COMPAT */ ) optind++; last_nonopt = optind; } /* Special ARGV-element `--' means premature end of options. Skip it like a null option, then exchange with previous non-options as if it were an option, then skip everything else like a non-option. */ if (optind != argc && !strcmp (argv[optind], "--")) { optind++; if (first_nonopt != last_nonopt && last_nonopt != optind) exchange ((char **) argv); else if (first_nonopt == last_nonopt) first_nonopt = optind; last_nonopt = argc; optind = argc; } /* If we have done all the ARGV-elements, stop the scan and back over any non-options that we skipped and permuted. */ if (optind == argc) { /* Set the next-arg-index to point at the non-options that we previously skipped, so the caller will digest them. */ if (first_nonopt != last_nonopt) optind = first_nonopt; return EOF; } /* If we have come to a non-option and did not permute it, either stop the scan or describe it to the caller and pass it by. */ if ((argv[optind][0] != '-' || argv[optind][1] == '\0') #ifdef GETOPT_COMPAT && (longopts == NULL || argv[optind][0] != '+' || argv[optind][1] == '\0') #endif /* GETOPT_COMPAT */ ) { if (ordering == REQUIRE_ORDER) return EOF; optarg = argv[optind++]; return 1; } /* We have found another option-ARGV-element. Start decoding its characters. */ nextchar = (argv[optind] + 1 + (longopts != NULL && argv[optind][1] == '-')); } if (longopts != NULL && ((argv[optind][0] == '-' && (argv[optind][1] == '-' || long_only)) #ifdef GETOPT_COMPAT || argv[optind][0] == '+' #endif /* GETOPT_COMPAT */ )) { const struct option *p; char *s = nextchar; int exact = 0; int ambig = 0; const struct option *pfound = NULL; int indfound; while (*s && *s != '=') s++; /* Test all options for either exact match or abbreviated matches. */ for (p = longopts, option_index = 0; p->name; p++, option_index++) if (!strncmp (p->name, nextchar, s - nextchar)) { if (s - nextchar == strlen (p->name)) { /* Exact match found. */ pfound = p; indfound = option_index; exact = 1; break; } else if (pfound == NULL) { /* First nonexact match found. */ pfound = p; indfound = option_index; } else /* Second nonexact match found. */ ambig = 1; } if (ambig && !exact) { if (opterr) fprintf (stderr, "%s: option `%s' is ambiguous\n", argv[0], argv[optind]); nextchar += strlen (nextchar); optind++; return '?'; } if (pfound != NULL) { option_index = indfound; optind++; if (*s) { /* Don't test has_arg with >, because some C compilers don't allow it to be used on enums. */ if (pfound->has_arg) optarg = s + 1; else { if (opterr) { if (argv[optind - 1][1] == '-') /* --option */ fprintf (stderr, "%s: option `--%s' doesn't allow an argument\n", argv[0], pfound->name); else /* +option or -option */ fprintf (stderr, "%s: option `%c%s' doesn't allow an argument\n", argv[0], argv[optind - 1][0], pfound->name); } nextchar += strlen (nextchar); return '?'; } } else if (pfound->has_arg == 1) { if (optind < argc) optarg = argv[optind++]; else { if (opterr) fprintf (stderr, "%s: option `%s' requires an argument\n", argv[0], argv[optind - 1]); nextchar += strlen (nextchar); return optstring[0] == ':' ? ':' : '?'; } } nextchar += strlen (nextchar); if (longind != NULL) *longind = option_index; if (pfound->flag) { *(pfound->flag) = pfound->val; return 0; } return pfound->val; } /* Can't find it as a long option. If this is not getopt_long_only, or the option starts with '--' or is not a valid short option, then it's an error. Otherwise interpret it as a short option. */ if (!long_only || argv[optind][1] == '-' #ifdef GETOPT_COMPAT || argv[optind][0] == '+' #endif /* GETOPT_COMPAT */ || my_index (optstring, *nextchar) == NULL) { if (opterr) { if (argv[optind][1] == '-') /* --option */ fprintf (stderr, "%s: unrecognized option `--%s'\n", argv[0], nextchar); else /* +option or -option */ fprintf (stderr, "%s: unrecognized option `%c%s'\n", argv[0], argv[optind][0], nextchar); } nextchar = (char *) ""; optind++; return '?'; } } /* Look at and handle the next option-character. */ { char c = *nextchar++; char *temp = my_index (optstring, c); /* Increment `optind' when we start to process its last character. */ if (*nextchar == '\0') ++optind; if (temp == NULL || c == ':') { if (opterr) { #if 0 if (c < 040 || c >= 0177) fprintf (stderr, "%s: unrecognized option, character code 0%o\n", argv[0], c); else fprintf (stderr, "%s: unrecognized option `-%c'\n", argv[0], c); #else /* 1003.2 specifies the format of this message. */ fprintf (stderr, "%s: illegal option -- %c\n", argv[0], c); #endif } optopt = c; return '?'; } if (temp[1] == ':') { if (temp[2] == ':') { /* This is an option that accepts an argument optionally. */ if (*nextchar != '\0') { optarg = nextchar; optind++; } else optarg = 0; nextchar = NULL; } else { /* This is an option that requires an argument. */ if (*nextchar != '\0') { optarg = nextchar; /* If we end this ARGV-element by taking the rest as an arg, we must advance to the next element now. */ optind++; } else if (optind == argc) { if (opterr) { #if 0 fprintf (stderr, "%s: option `-%c' requires an argument\n", argv[0], c); #else /* 1003.2 specifies the format of this message. */ fprintf (stderr, "%s: option requires an argument -- %c\n", argv[0], c); #endif } optopt = c; if (optstring[0] == ':') c = ':'; else c = '?'; } else /* We already incremented `optind' once; increment it again when taking next ARGV-elt as argument. */ optarg = argv[optind++]; nextchar = NULL; } } return c; } } int getopt (argc, argv, optstring) int argc; char *const *argv; const char *optstring; { return _getopt_internal (argc, argv, optstring, (const struct option *) 0, (int *) 0, 0); } #endif /* _LIBC or not __GNU_LIBRARY__. */ #ifdef TEST /* Compile with -DTEST to make an executable for use in testing the above definition of `getopt'. */ int main (argc, argv) int argc; char **argv; { int c; int digit_optind = 0; while (1) { int this_option_optind = optind ? optind : 1; c = getopt (argc, argv, "abc:d:0123456789"); if (c == EOF) break; switch (c) { case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9': if (digit_optind != 0 && digit_optind != this_option_optind) printf ("digits occur in two different argv-elements.\n"); digit_optind = this_option_optind; printf ("option %c\n", c); break; case 'a': printf ("option a\n"); break; case 'b': printf ("option b\n"); break; case 'c': printf ("option c with value `%s'\n", optarg); break; case '?': break; default: printf ("?? getopt returned character code 0%o ??\n", c); } } if (optind < argc) { printf ("non-option ARGV-elements: "); while (optind < argc) printf ("%s ", argv[optind++]); printf ("\n"); } exit (0); } #endif /* TEST */ dist/sed/getopt1.c100660 765 144 10056 6070131465 12712 0ustar lordusers/* getopt_long and getopt_long_only entry points for GNU getopt. Copyright (C) 1987, 88, 89, 90, 91, 92, 1993 Free Software Foundation, Inc. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */ #ifdef HAVE_CONFIG_H #include "config.h" #endif #include "getopt.h" #if !__STDC__ && !defined(const) && IN_GCC #define const #endif #include /* Comment out all this code if we are using the GNU C Library, and are not actually compiling the library itself. This code is part of the GNU C Library, but also included in many other GNU distributions. Compiling and linking in this code is a waste when using the GNU C library (especially if it is a shared library). Rather than having every GNU program understand `configure --with-gnu-libc' and omit the object files, it is simpler to just do this in the source for each such file. */ #if defined (_LIBC) || !defined (__GNU_LIBRARY__) /* This needs to come after some library #include to get __GNU_LIBRARY__ defined. */ #ifdef __GNU_LIBRARY__ #include #else char *getenv (); #endif #ifndef NULL #define NULL 0 #endif int getopt_long (argc, argv, options, long_options, opt_index) int argc; char *const *argv; const char *options; const struct option *long_options; int *opt_index; { return _getopt_internal (argc, argv, options, long_options, opt_index, 0); } /* Like getopt_long, but '-' as well as '--' can indicate a long option. If an option that starts with '-' (not '--') doesn't match a long option, but does match a short option, it is parsed as a short option instead. */ int getopt_long_only (argc, argv, options, long_options, opt_index) int argc; char *const *argv; const char *options; const struct option *long_options; int *opt_index; { return _getopt_internal (argc, argv, options, long_options, opt_index, 1); } #endif /* _LIBC or not __GNU_LIBRARY__. */ #ifdef TEST #include int main (argc, argv) int argc; char **argv; { int c; int digit_optind = 0; while (1) { int this_option_optind = optind ? optind : 1; int option_index = 0; static struct option long_options[] = { {"add", 1, 0, 0}, {"append", 0, 0, 0}, {"delete", 1, 0, 0}, {"verbose", 0, 0, 0}, {"create", 0, 0, 0}, {"file", 1, 0, 0}, {0, 0, 0, 0} }; c = getopt_long (argc, argv, "abc:d:0123456789", long_options, &option_index); if (c == EOF) break; switch (c) { case 0: printf ("option %s", long_options[option_index].name); if (optarg) printf (" with arg %s", optarg); printf ("\n"); break; case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9': if (digit_optind != 0 && digit_optind != this_option_optind) printf ("digits occur in two different argv-elements.\n"); digit_optind = this_option_optind; printf ("option %c\n", c); break; case 'a': printf ("option a\n"); break; case 'b': printf ("option b\n"); break; case 'c': printf ("option c with value `%s'\n", optarg); break; case 'd': printf ("option d with value `%s'\n", optarg); break; case '?': break; default: printf ("?? getopt returned character code 0%o ??\n", c); } } if (optind < argc) { printf ("non-option ARGV-elements: "); while (optind < argc) printf ("%s ", argv[optind++]); printf ("\n"); } exit (0); } #endif /* TEST */ dist/sed/alloca.c100660 765 144 32724 6070131465 12570 0ustar lordusers/* alloca.c -- allocate automatically reclaimed memory (Mostly) portable public-domain implementation -- D A Gwyn This implementation of the PWB library alloca function, which is used to allocate space off the run-time stack so that it is automatically reclaimed upon procedure exit, was inspired by discussions with J. Q. Johnson of Cornell. J.Otto Tennant contributed the Cray support. There are some preprocessor constants that can be defined when compiling for your specific system, for improved efficiency; however, the defaults should be okay. The general concept of this implementation is to keep track of all alloca-allocated blocks, and reclaim any that are found to be deeper in the stack than the current invocation. This heuristic does not reclaim storage as soon as it becomes invalid, but it will do so eventually. As a special case, alloca(0) reclaims storage without allocating any. It is a good idea to use alloca(0) in your main control loop, etc. to force garbage collection. */ #ifdef HAVE_CONFIG_H #include "config.h" #endif /* If compiling with GCC, this file's not needed. */ #ifndef alloca #ifdef emacs #ifdef static /* actually, only want this if static is defined as "" -- this is for usg, in which emacs must undefine static in order to make unexec workable */ #ifndef STACK_DIRECTION you lose -- must know STACK_DIRECTION at compile-time #endif /* STACK_DIRECTION undefined */ #endif /* static */ #endif /* emacs */ /* If your stack is a linked list of frames, you have to provide an "address metric" ADDRESS_FUNCTION macro. */ #if defined (CRAY) && defined (CRAY_STACKSEG_END) long i00afunc (); #define ADDRESS_FUNCTION(arg) (char *) i00afunc (&(arg)) #else #define ADDRESS_FUNCTION(arg) &(arg) #endif #if __STDC__ typedef void *pointer; #else typedef char *pointer; #endif #define NULL 0 /* Different portions of Emacs need to call different versions of malloc. The Emacs executable needs alloca to call xmalloc, because ordinary malloc isn't protected from input signals. On the other hand, the utilities in lib-src need alloca to call malloc; some of them are very simple, and don't have an xmalloc routine. Non-Emacs programs expect this to call use xmalloc. Callers below should use malloc. */ #ifndef emacs #define malloc xmalloc extern pointer xmalloc (); #endif /* Define STACK_DIRECTION if you know the direction of stack growth for your system; otherwise it will be automatically deduced at run-time. STACK_DIRECTION > 0 => grows toward higher addresses STACK_DIRECTION < 0 => grows toward lower addresses STACK_DIRECTION = 0 => direction of growth unknown */ #ifndef STACK_DIRECTION #define STACK_DIRECTION 0 /* Direction unknown. */ #endif #if STACK_DIRECTION != 0 #define STACK_DIR STACK_DIRECTION /* Known at compile-time. */ #else /* STACK_DIRECTION == 0; need run-time code. */ static int stack_dir; /* 1 or -1 once known. */ #define STACK_DIR stack_dir static void find_stack_direction () { static char *addr = NULL; /* Address of first `dummy', once known. */ auto char dummy; /* To get stack address. */ if (addr == NULL) { /* Initial entry. */ addr = ADDRESS_FUNCTION (dummy); find_stack_direction (); /* Recurse once. */ } else { /* Second entry. */ if (ADDRESS_FUNCTION (dummy) > addr) stack_dir = 1; /* Stack grew upward. */ else stack_dir = -1; /* Stack grew downward. */ } } #endif /* STACK_DIRECTION == 0 */ /* An "alloca header" is used to: (a) chain together all alloca'ed blocks; (b) keep track of stack depth. It is very important that sizeof(header) agree with malloc alignment chunk size. The following default should work okay. */ #ifndef ALIGN_SIZE #define ALIGN_SIZE sizeof(double) #endif typedef union hdr { char align[ALIGN_SIZE]; /* To force sizeof(header). */ struct { union hdr *next; /* For chaining headers. */ char *deep; /* For stack depth measure. */ } h; } header; static header *last_alloca_header = NULL; /* -> last alloca header. */ /* Return a pointer to at least SIZE bytes of storage, which will be automatically reclaimed upon exit from the procedure that called alloca. Originally, this space was supposed to be taken from the current stack frame of the caller, but that method cannot be made to work for some implementations of C, for example under Gould's UTX/32. */ pointer alloca (size) unsigned size; { auto char probe; /* Probes stack depth: */ register char *depth = ADDRESS_FUNCTION (probe); #if STACK_DIRECTION == 0 if (STACK_DIR == 0) /* Unknown growth direction. */ find_stack_direction (); #endif /* Reclaim garbage, defined as all alloca'd storage that was allocated from deeper in the stack than currently. */ { register header *hp; /* Traverses linked list. */ for (hp = last_alloca_header; hp != NULL;) if ((STACK_DIR > 0 && hp->h.deep > depth) || (STACK_DIR < 0 && hp->h.deep < depth)) { register header *np = hp->h.next; free ((pointer) hp); /* Collect garbage. */ hp = np; /* -> next header. */ } else break; /* Rest are not deeper. */ last_alloca_header = hp; /* -> last valid storage. */ } if (size == 0) return NULL; /* No allocation required. */ /* Allocate combined header + user data storage. */ { register pointer new = malloc (sizeof (header) + size); /* Address of header. */ ((header *) new)->h.next = last_alloca_header; ((header *) new)->h.deep = depth; last_alloca_header = (header *) new; /* User storage begins just after header. */ return (pointer) ((char *) new + sizeof (header)); } } #if defined (CRAY) && defined (CRAY_STACKSEG_END) #ifdef DEBUG_I00AFUNC #include #endif #ifndef CRAY_STACK #define CRAY_STACK #ifndef CRAY2 /* Stack structures for CRAY-1, CRAY X-MP, and CRAY Y-MP */ struct stack_control_header { long shgrow:32; /* Number of times stack has grown. */ long shaseg:32; /* Size of increments to stack. */ long shhwm:32; /* High water mark of stack. */ long shsize:32; /* Current size of stack (all segments). */ }; /* The stack segment linkage control information occurs at the high-address end of a stack segment. (The stack grows from low addresses to high addresses.) The initial part of the stack segment linkage control information is 0200 (octal) words. This provides for register storage for the routine which overflows the stack. */ struct stack_segment_linkage { long ss[0200]; /* 0200 overflow words. */ long sssize:32; /* Number of words in this segment. */ long ssbase:32; /* Offset to stack base. */ long:32; long sspseg:32; /* Offset to linkage control of previous segment of stack. */ long:32; long sstcpt:32; /* Pointer to task common address block. */ long sscsnm; /* Private control structure number for microtasking. */ long ssusr1; /* Reserved for user. */ long ssusr2; /* Reserved for user. */ long sstpid; /* Process ID for pid based multi-tasking. */ long ssgvup; /* Pointer to multitasking thread giveup. */ long sscray[7]; /* Reserved for Cray Research. */ long ssa0; long ssa1; long ssa2; long ssa3; long ssa4; long ssa5; long ssa6; long ssa7; long sss0; long sss1; long sss2; long sss3; long sss4; long sss5; long sss6; long sss7; }; #else /* CRAY2 */ /* The following structure defines the vector of words returned by the STKSTAT library routine. */ struct stk_stat { long now; /* Current total stack size. */ long maxc; /* Amount of contiguous space which would be required to satisfy the maximum stack demand to date. */ long high_water; /* Stack high-water mark. */ long overflows; /* Number of stack overflow ($STKOFEN) calls. */ long hits; /* Number of internal buffer hits. */ long extends; /* Number of block extensions. */ long stko_mallocs; /* Block allocations by $STKOFEN. */ long underflows; /* Number of stack underflow calls ($STKRETN). */ long stko_free; /* Number of deallocations by $STKRETN. */ long stkm_free; /* Number of deallocations by $STKMRET. */ long segments; /* Current number of stack segments. */ long maxs; /* Maximum number of stack segments so far. */ long pad_size; /* Stack pad size. */ long current_address; /* Current stack segment address. */ long current_size; /* Current stack segment size. This number is actually corrupted by STKSTAT to include the fifteen word trailer area. */ long initial_address; /* Address of initial segment. */ long initial_size; /* Size of initial segment. */ }; /* The following structure describes the data structure which trails any stack segment. I think that the description in 'asdef' is out of date. I only describe the parts that I am sure about. */ struct stk_trailer { long this_address; /* Address of this block. */ long this_size; /* Size of this block (does not include this trailer). */ long unknown2; long unknown3; long link; /* Address of trailer block of previous segment. */ long unknown5; long unknown6; long unknown7; long unknown8; long unknown9; long unknown10; long unknown11; long unknown12; long unknown13; long unknown14; }; #endif /* CRAY2 */ #endif /* not CRAY_STACK */ #ifdef CRAY2 /* Determine a "stack measure" for an arbitrary ADDRESS. I doubt that "lint" will like this much. */ static long i00afunc (long *address) { struct stk_stat status; struct stk_trailer *trailer; long *block, size; long result = 0; /* We want to iterate through all of the segments. The first step is to get the stack status structure. We could do this more quickly and more directly, perhaps, by referencing the $LM00 common block, but I know that this works. */ STKSTAT (&status); /* Set up the iteration. */ trailer = (struct stk_trailer *) (status.current_address + status.current_size - 15); /* There must be at least one stack segment. Therefore it is a fatal error if "trailer" is null. */ if (trailer == 0) abort (); /* Discard segments that do not contain our argument address. */ while (trailer != 0) { block = (long *) trailer->this_address; size = trailer->this_size; if (block == 0 || size == 0) abort (); trailer = (struct stk_trailer *) trailer->link; if ((block <= address) && (address < (block + size))) break; } /* Set the result to the offset in this segment and add the sizes of all predecessor segments. */ result = address - block; if (trailer == 0) { return result; } do { if (trailer->this_size <= 0) abort (); result += trailer->this_size; trailer = (struct stk_trailer *) trailer->link; } while (trailer != 0); /* We are done. Note that if you present a bogus address (one not in any segment), you will get a different number back, formed from subtracting the address of the first block. This is probably not what you want. */ return (result); } #else /* not CRAY2 */ /* Stack address function for a CRAY-1, CRAY X-MP, or CRAY Y-MP. Determine the number of the cell within the stack, given the address of the cell. The purpose of this routine is to linearize, in some sense, stack addresses for alloca. */ static long i00afunc (long address) { long stkl = 0; long size, pseg, this_segment, stack; long result = 0; struct stack_segment_linkage *ssptr; /* Register B67 contains the address of the end of the current stack segment. If you (as a subprogram) store your registers on the stack and find that you are past the contents of B67, you have overflowed the segment. B67 also points to the stack segment linkage control area, which is what we are really interested in. */ stkl = CRAY_STACKSEG_END (); ssptr = (struct stack_segment_linkage *) stkl; /* If one subtracts 'size' from the end of the segment, one has the address of the first word of the segment. If this is not the first segment, 'pseg' will be nonzero. */ pseg = ssptr->sspseg; size = ssptr->sssize; this_segment = stkl - size; /* It is possible that calling this routine itself caused a stack overflow. Discard stack segments which do not contain the target address. */ while (!(this_segment <= address && address <= stkl)) { #ifdef DEBUG_I00AFUNC fprintf (stderr, "%011o %011o %011o\n", this_segment, address, stkl); #endif if (pseg == 0) break; stkl = stkl - pseg; ssptr = (struct stack_segment_linkage *) stkl; size = ssptr->sssize; pseg = ssptr->sspseg; this_segment = stkl - size; } result = address - this_segment; /* If you subtract pseg from the current end of the stack, you get the address of the previous stack segment's end. This seems a little convoluted to me, but I'll bet you save a cycle somewhere. */ while (pseg != 0) { #ifdef DEBUG_I00AFUNC fprintf (stderr, "%011o %011o\n", pseg, size); #endif stkl = stkl - pseg; ssptr = (struct stack_segment_linkage *) stkl; size = ssptr->sssize; pseg = ssptr->sspseg; result += size; } return (result); } #endif /* not CRAY2 */ #endif /* CRAY */ #endif /* no alloca */ dist/sed/testsuite/ 42770 765 144 0 6071416407 13101 5ustar lordusersdist/sed/testsuite/8bit.good100660 765 144 260 6070131553 14666 0ustar lordusersäƤâ¤è ¤ßäÆ»ý¤Á ·¡¶ú¤â¤è ¤ß·¡¶ú»ý¤Á ¤³¤ÎµÖ¤Ë ºÚŦ¤Þ¤¹»ù ²È´Ö¤«¤Ê ¹ð¤é¤µ¤Í ¤½¤é¤ß¤Ä ÆüËܤιñ¤Ï ¤ª¤·¤ã¤Ê¤Ù¤Æ ¤ï¤ì¤³¤½µï¤ì ¤·¤­¤Ê¤Ù¤Æ ¤ï¤ì¤³¤½ ºÂ¤» ¤ï¤Ë¤³¤½¤Ï ¹ð¤é¤á ²È¤ò¤â̾¤ò¤â dist/sed/testsuite/8bit.inp100660 765 144 260 6070131553 14524 0ustar lordusersäƤâ¤è ¤ßäÆ»ý¤Á ·¡¶ú¤â¤è ¤ß·¡¶ú»ý¤Á ¤³¤ÎµÖ¤Ë ºÚŦ¤Þ¤¹»ù ²È´Ö¤«¤Ê ¹ð¤é¤µ¤Í ¤½¤é¤ß¤Ä ÂçϤιñ¤Ï ¤ª¤·¤ã¤Ê¤Ù¤Æ ¤ï¤ì¤³¤½µï¤ì ¤·¤­¤Ê¤Ù¤Æ ¤ï¤ì¤³¤½ ºÂ¤» ¤ï¤Ë¤³¤½¤Ï ¹ð¤é¤á ²È¤ò¤â̾¤ò¤â dist/sed/testsuite/8bit.sed100660 765 144 1255 6070131553 14536 0ustar lordusers# The first poem from the Man'yoshu. I like Hitomaro's poems better # but I couldn't find a copy of any of them in Japanese. This version # of this poem is from $BNc2r8E8l<-E5(B($BBh;0HG(B)$B;0>JF2(B. # # Speaking of Hitomaro, here is the english translation of one of my # favorites. I just know that everyone reading these test cases wants # to see this. # # In the autumn mountains # The yellow leaves are so thick. # Alas, how shall I seek my love # Who has wandered away? # # I see the messenger come # As the yellow leaves are falling. # Oh, well I remember # How on such a day we used to meet-- # My lover and I! # -- Kakinomoto Hitomaro # s/ÂçÏÂ/ÆüËÜ/ dist/sed/testsuite/8to7.good100660 765 144 1235 6070131553 14644 0ustar lordusers\344\306\244\342\244\350 \244\337\344\306\273\375\244\301$ \267\241\266\372\244\342\244\350 \244\337\267\241\266\372\273\375\244\ \301$ \244\263\244\316\265\326\244\313 \272\332\305\246\244\336\244\271\273\ \371$ \262\310\264\326\244\253\244\312 \271\360\244\351\244\265\244\315$ \244\275\244\351\244\337\244\304 \302\347\317\302\244\316\271\361\244\ \317$ \244\252\244\267\244\343\244\312\244\331\244\306 \244\357\244\354\244\ \263\244\275\265\357\244\354$ \244\267\244\255\244\312\244\331\244\306 \244\357\244\354\244\263\244\ \275 \272\302\244\273$ \244\357\244\313\244\263\244\275\244\317 \271\360\244\351\244\341$ \262\310\244\362\244\342\314\276\244\362\244\342$ dist/sed/testsuite/8to7.sed100660 765 144 224 6070131553 14444 0ustar lordusers# it isn't entirely clear how lines should be wrapped. Plus I am # not sure whether sed is required to append a $ at the end of line or # not. l;d dist/sed/testsuite/Makefile100660 765 144 10342 6071053340 14646 0ustar lordusersSHELL = /bin/sh SED=time ./sed USED=./sed SSED=time /usr/bin/sed USSED=/usr/bin/sed all: subs othersedcmds cmdlines clean fail subs: enable sep inclib 8bit newjis cv-vars manis xabcx dollar noeol sub-num sub-all subwrite head middle maddingcrowd mac-mf othersedcmds: khadafy spencer linecnt distrib writeout readin cmdlines: help version file quiet fail: 8to7 enable:: $(SED) -f enable.sed < enable.inp > tmp cmp enable.good tmp $(SSED) -f enable.sed < enable.inp > tmp2 cmp enable.good tmp2 sep:: $(SED) -f sep.sed < sep.inp > tmp cmp sep.good tmp $(SSED) -f sep.sed < sep.inp > tmp2 cmp sep.good tmp2 inclib:: $(SED) -f inclib.sed < inclib.inp > tmp cmp inclib.good tmp $(SSED) -f inclib.sed < inclib.inp > tmp2 cmp inclib.good tmp2 8bit:: $(SED) -f 8bit.sed < 8bit.inp > tmp cmp 8bit.good tmp $(SSED) -f 8bit.sed < 8bit.inp > tmp2 cmp 8bit.good tmp2 newjis:: $(SED) -f newjis.sed < newjis.inp > tmp cmp newjis.good tmp $(SSED) -f newjis.sed < newjis.inp > tmp2 cmp newjis.good tmp2 cv-vars:: $(SED) -n -f cv-vars.sed < cv-vars.inp > tmp cmp cv-vars.good tmp $(SSED) -n -f cv-vars.sed < cv-vars.inp > tmp2 cmp cv-vars.good tmp2 head:: $(SED) -f head.sed < head.inp > tmp cmp head.good tmp $(SSED) -f head.sed < head.inp > tmp2 cmp head.good tmp2 manis:: $(SED) -f manis.sed < manis.inp > tmp cmp manis.inp tmp $(SSED) -f manis.sed < manis.inp > tmp2 cmp manis.inp tmp2 xabcx:: $(SED) -f xabcx.sed < xabcx.inp > tmp cmp xabcx.good tmp $(SSED) -f xabcx.sed < xabcx.inp > tmp2 cmp xabcx.good tmp2 middle:: $(SED) -n -f middle.sed < head.inp > tmp cmp middle.good tmp $(SSED) -n -f middle.sed < head.inp > tmp2 cmp middle.good tmp2 dollar:: $(SED) -f dollar.sed < dollar.inp > tmp cmp dollar.good tmp $(SSED) -f dollar.sed < dollar.inp > tmp2 cmp dollar.good tmp2 noeol:: $(SED) -f noeol.sed < noeol.inp > tmp cmp noeol.good tmp $(SSED) -f noeol.sed < noeol.inp > tmp2 cmp noeol.good tmp2 sub-num:: $(SED) -f sub-num.sed < sub-num.inp > tmp cmp sub-num.good tmp $(SSED) -f sub-num.sed < sub-num.inp > tmp2 cmp sub-num.good tmp2 sub-all:: $(SED) -f sub-all.sed < sub-num.inp > tmp cmp sub-all.good tmp $(SSED) -f sub-all.sed < sub-num.inp > tmp2 cmp sub-all.good tmp2 subwrite:: $(SED) -f subwrite.sed < subwrite.inp > tmp1 cmp subwrite.good1 tmp1 cmp subwrite.good2 tmp2 $(SSED) -f subwrite.sed < subwrite.inp > tmp1 cmp subwrite.good1 tmp1 cmp subwrite.good2 tmp2 khadafy:: $(SED) -f khadafy.sed < khadafy.inp > tmp cmp tmp khadafy.inp $(SSED) -f khadafy.sed < khadafy.inp > tmp2 cmp tmp2 khadafy.inp spencer:: sh spencer.sh $(USED) < spencer.inp > spencer.tmp sh spencer.tmp # this 1003.2 feature (l) hasn't been implemented yet, hence the # -cmp instead of just cmp--we know it's going to fail. 8to7:: $(SED) -f 8to7.sed < 8bit.inp > tmp -cmp tmp 8to7.good $(SSED) -f 8to7.sed < 8bit.inp > tmp2 -cmp tmp2 8to7.good linecnt:: $(SED) -f linecnt.sed < linecnt.inp > tmp cmp tmp linecnt.good $(SSED) -f linecnt.sed < linecnt.inp > tmp2 cmp tmp2 linecnt.good distrib:: sh distrib.sh $(USED) < distrib.inp > tmp cmp tmp distrib.good writeout:: $(SED) -f writeout.sed < writeout.inp >tmp1 cmp tmp1 writeout.good1 cmp tmp2 writeout.good2 readin:: $(SED) -f readin.sed < readin.inp >tmp cmp tmp readin.good $(SSED) -f readin.sed < readin.inp >tmp2 cmp tmp2 readin.good maddingcrowd:: $(SED) -f maddingcrowd.sed < maddingcrowd.inp > tmp cmp tmp maddingcrowd.good echo DO NOT $(SSED) -f maddingcrowd.sed < maddingcrowd.inp > tmp2 echo sed 2.05.... t->+inf mac-mf:: $(SED) -f mac-mf.sed < mac-mf.inp > tmp cmp tmp mac-mf.good $(SSED) -f mac-mf.sed < mac-mf.inp > tmp2 cmp tmp2 mac-mf.good # # cmdlines targets # help:: $(USED) --help > tmp1 $(USED) -h > tmp2 -$(USED) > tmp3 2>&1 cmp tmp1 tmp2 cmp tmp2 tmp3 -cmp tmp1 help.good version:: $(USED) --version > tmp1 2>&1 $(USED) -V > tmp2 2>&1 cmp tmp1 tmp2 -cmp tmp1 version.good file:: $(SED) --file=newjis.sed < newjis.inp > tmp cmp newjis.good tmp $(SSED) --file=newjis.sed < newjis.inp > tmp2 cmp newjis.good tmp2 quiet:: $(SED) --quiet -f cv-vars.sed < cv-vars.inp > tmp cmp cv-vars.good tmp $(SSED) --quiet -f cv-vars.sed < cv-vars.inp > tmp2 cmp cv-vars.good tmp2 # # # clean: rm -f tmp tmp1 tmp2 tmp3 core junk spencer.tmp dist/sed/testsuite/cv-vars.good100660 765 144 230 6070131553 15376 0ustar lordusers: ${_cv_='emptyvar'} : ${ac_cv_prog_RANLIB='/usr/bin/ranlib'} : ${ac_cv_prog_CC='/usr/unsupported/\ \ /lib/_cv_/cc'} : ${ac_cv_prog_CPP='/usr/bin/cpp'} dist/sed/testsuite/cv-vars.inp100660 765 144 224 6070131553 15237 0ustar lordusers_cv_=emptyvar ac_cv_prog_RANLIB=/usr/bin/ranlib ac_cv_prog_CC=/usr/unsupported/\ \ /lib/_cv_/cc ac_cv_prog_CPP=/usr/bin/cpp SHELL=bash GNU=GNU!UNIX dist/sed/testsuite/cv-vars.sed100660 765 144 163 6070131553 15226 0ustar lordusers# inspired by an autoconf generated configure script. s/^\([a-zA-Z0-9_]*_cv_[a-zA-Z0-9_]*\)=\(.*\)/: \${\1='\2'}/p dist/sed/testsuite/distrib.good100660 765 144 2043 6070131553 15501 0ustar lordusersPath: mailnewsgateway From crash@cygnus.com Wed Mar 8 18: 02:42 1995 From: crash@cygnus.com (Jason Molenda) Message-ID: <9503090202.AA06931.alt.buddha.fat.short.guy@phydeaux.cygnus.com> Subject: Note for sed testsuite Original-To: molenda@msi.umn.edu Date: Wed, 8 Mar 1995 18:02:24 -0800 (PST) X-Mailer: ELM [version 2.4 PL23] Newsgroups: alt.buddha.short.fat.guy Distribution: world Sender: news@cygnus.com Approved: alt.buddha.short.fat.guy@cygnus.com _Summum Bonum_ All the breath and the bloom of the year in the bag of one bee: All the wonder and wealth of the mine in the heart of one gem: In the core of one pearl all the shade and the shine of the sea: Breath and bloom, shade and shine, -- wonder, wealth, and -- how far above them -- Truth, thats brighter than gem, Trust, that's purer than pearl, -- Brightest truth, purest trust in the universe -- all were for me In the kiss of one girl. -- Robert Browning dist/sed/testsuite/distrib.inp100660 765 144 2723 6070131553 15344 0ustar lordusersFrom crash@cygnus.com Wed Mar 8 18:02:42 1995 Received: from s1.msi.umn.edu (s1.msi.umn.edu [128.101.24.1]) by cygnus.com (8.6.9/8.6.9) with ESMTP id SAA21692 for ; Wed, 8 Mar 1995 18:02:41 -0800 Received: from cygint.cygnus.com (cygint.cygnus.com [140.174.1.1]) by s1.msi.umn.edu (8.6.10/8.6.9) with ESMTP id TAA13398 for ; Wed, 8 Mar 1995 19:59:18 -0600 Received: from phydeaux.cygnus.com (phydeaux.cygnus.com [140.174.1.85]) by cygnus.com (8.6.9/8.6.9) with SMTP id SAA21688 for ; Wed, 8 Mar 1995 18:02:33 -0800 From: Jason Molenda Received: by phydeaux.cygnus.com (5.65/4.7) id AA06931; Wed, 8 Mar 1995 18:02:28 -0800 Message-Id: <9503090202.AA06931@phydeaux.cygnus.com> Subject: Note for sed testsuite To: molenda@msi.umn.edu Date: Wed, 8 Mar 1995 18:02:24 -0800 (PST) X-Mailer: ELM [version 2.4 PL23] _Summum Bonum_ All the breath and the bloom of the year in the bag of one bee: All the wonder and wealth of the mine in the heart of one gem: In the core of one pearl all the shade and the shine of the sea: Breath and bloom, shade and shine, -- wonder, wealth, and -- how far above them -- Truth, thats brighter than gem, Trust, that's purer than pearl, -- Brightest truth, purest trust in the universe -- all were for me In the kiss of one girl. -- Robert Browning dist/sed/testsuite/distrib.sed100660 765 144 2626 6070131553 15333 0ustar lordusers# This is straight out of C News # # # All this does is massage the headers so they look like what news # software expects. To:, Cc: and Resent-*: headers are masked. # Reply-To: is turned into references, which is questionable (could # just as well be dropped. # # The From: line is rewritten to use the "address (comments)" form # instead of "phrase " form our mailer uses. Also, addresses # with no "@domainname" are assumed to originate locally, and so are # given a domain. # # The Sender: field below reflects the address of the person who # maintains our mailing lists. The Approved: field is in a special # form, so that we can do bidirectional gatewaying. Any message # in a newsgroup that bears this stamp will not be fed into the # matching mailing list. 1{i\\ Path: $path } :a /^[Rr]eceived:/b r /^[Nn]ewsgroups:/b r /^[Pp]ath:/b r /^[Tt][Oo]:/s/^/Original-/ /^[Cc][Cc]:/s/^/Original-/ /^[Rr][Ee][Ss][Ee][Nn][Tt]-.*/s/^/Original-/ /^[Mm][Ee][Ss][Ss][Aa][Gg][Ee]-[Ii][Dd]:/s/@/.$newsgroup@/ s/^[Ii]n-[Rr]eply-[Tt]o:/References:/ /^From:/{ s/<\([^@]*\)>\$/<\1@$thissite>/ s/^From:[ ][ ]*\(.*\) *<\(.*\)>\$/From: \2 (\1)/ } s/-[Ii]d:/-ID:/ s/^[Ss][Uu][Bb][Jj][Ee][Cc][Tt]:[ ]*$/Subject: (none)/ s/^\([^:]*:\)[ ]*/\1 / /^\$/{i\\ Newsgroups: $newsgroup\\ Distribution: $distribution\\ Sender: $admin@$thissite\\ Approved: $newsgroup@cygnus.com b e } p n b a :r s/.*//g n /^[ ]/b r b a :e p n b e dist/sed/testsuite/dollar.good100660 765 144 164 6070131554 15301 0ustar lordusersI can't quite remember where I heard it, but I can't seem to get out of my head the phrase space the final frontier dist/sed/testsuite/dollar.inp100660 765 144 156 6070131554 15140 0ustar lordusersI can't quite remember where I heard it, but I can't seem to get out of my head the phrase the final frontier dist/sed/testsuite/dollar.sed100660 765 144 15 6070131554 15077 0ustar lordusers$s/^/space / dist/sed/testsuite/enable.good100660 765 144 54 6070131554 15230 0ustar lorduserstargets x11-testing wollybears-in-minnesota dist/sed/testsuite/enable.inp100660 765 144 165 6070131554 15111 0ustar lordusers--enable-targets=sparc-sun-sunos4.1.3,srec --enable-x11-testing=on --enable-wollybears-in-minnesota=yes-id-like-that dist/sed/testsuite/enable.sed100660 765 144 114 6070131554 15070 0ustar lordusers# inspired by an autoconf generated configure script. s/-*enable-//;s/=.*// dist/sed/testsuite/head.good100660 765 144 310 6070131554 14716 0ustar lordusers "...by imposing a tiny bit of order in a communication you are translating, you are carving out a little bit of order in the universe. You will never succeed. Everything will fail and come dist/sed/testsuite/head.inp100660 765 144 722 6070131554 14563 0ustar lordusers "...by imposing a tiny bit of order in a communication you are translating, you are carving out a little bit of order in the universe. You will never succeed. Everything will fail and come to an end finally. But you have a chance to carve a little bit of order and maybe even beauty out of the raw materials that surround you everywhere, and I think there is no greater meaning in life." Donald L. Philippi, Oct 1930 - Jan 1993 dist/sed/testsuite/head.sed100660 765 144 3 6070131555 14501 0ustar lordusers3q dist/sed/testsuite/help.good100660 765 144 217 6070131555 14754 0ustar lordusersUsage: ../sed [-nV] [--quiet] [--silent] [--version] [-e script] [-f script-file] [--expression=script] [--file=script-file] [file...] dist/sed/testsuite/inclib.good100660 765 144 1403 6070131555 15302 0ustar lordusers /usr/X11R6/include /usr/X11R5/include /usr/X11R4/include /usr/include/X11R6 /usr/include/X11R5 /usr/include/X11R4 /usr/local/X11R6/include /usr/local/X11R5/include /usr/local/X11R4/include /usr/local/include/X11R6 /usr/local/include/X11R5 /usr/local/include/X11R4 /usr/X11/include /usr/include/X11 /usr/local/X11/include /usr/local/include/X11 /usr/X386/include /usr/x386/include /usr/XFree86/include/X11 /usr/include /usr/local/include /usr/unsupported/include /usr/athena/include /usr/local/x11r5/include /usr/lpp/Xamples/include /usr/openwin/include /usr/openwin/share/include dist/sed/testsuite/inclib.inp100660 765 144 1227 6070131555 15144 0ustar lordusers /usr/X11R6/lib /usr/X11R5/lib /usr/X11R4/lib /usr/lib/X11R6 /usr/lib/X11R5 /usr/lib/X11R4 /usr/local/X11R6/lib /usr/local/X11R5/lib /usr/local/X11R4/lib /usr/local/lib/X11R6 /usr/local/lib/X11R5 /usr/local/lib/X11R4 /usr/X11/lib /usr/lib/X11 /usr/local/X11/lib /usr/local/lib/X11 /usr/X386/lib /usr/x386/lib /usr/XFree86/lib/X11 /usr/lib /usr/local/lib /usr/unsupported/lib /usr/athena/lib /usr/local/x11r5/lib /usr/lpp/Xamples/lib /usr/openwin/lib /usr/openwin/share/lib dist/sed/testsuite/inclib.sed100660 765 144 105 6070131555 15103 0ustar lordusers# inspired by an autoconf generated configure script. s;lib;include; dist/sed/testsuite/khadafy.inp100640 765 144 1331 6070131555 15305 0ustar lordusers1) Muammar Qaddafi 2) Mo'ammar Gadhafi 3) Muammar Kaddafi 4) Muammar Qadhafi 5) Moammar El Kadhafi 6) Muammar Gadafi 7) Mu'ammar al-Qadafi 8) Moamer El Kazzafi 9) Moamar al-Gaddafi 10) Mu'ammar Al Qathafi 11) Muammar Al Qathafi 12) Mo'ammar el-Gadhafi 13) Moamar El Kadhafi 14) Muammar al-Qadhafi 15) Mu'ammar al-Qadhdhafi 16) Mu'ammar Qadafi 17) Moamar Gaddafi 18) Mu'ammar Qadhdhafi 19) Muammar Khaddafi 20) Muammar al-Khaddafi 21) Mu'amar al-Kadafi 22) Muammar Ghaddafy 23) Muammar Ghadafi 24) Muammar Ghaddafi 25) Muamar Kaddafi 26) Muammar Quathafi 27) Muammar Gheddafi 28) Muamar Al-Kaddafi 29) Moammar Khadafy 30) Moammar Qudhafi 31) Mu'ammar al-Qaddafi 32) Mulazim Awwal Mu'ammar Muhammad Abu Minyar al-Qadhafi dist/sed/testsuite/khadafy.sed100640 765 144 202 6070131555 15246 0ustar lordusers# The Khadafy test is brought to you by Scott Anderson . . . /M[ou]'?am+[ae]r .*([AEae]l[- ])?[GKQ]h?[aeu]+([dtz][dhz]?)+af[iy]/p dist/sed/testsuite/linecnt.good100660 765 144 4140 6070131555 15477 0ustar lordusers1 A dialogue on poverty 2 3 On the night when the rain beats, 4 Driven by the wind, 5 On the night when the snowflakes mingle 6 With a sleety rain, 7 I feel so helplessly cold. 8 I nibble at a lump of salt, 9 Sip the hot, oft-diluted dregs of _sake_; 10 And coughing, snuffling, 11 And stroking my scanty beard, 12 I say in my pride, 13 "There's none worthy, save I!" 14 But I shiver still with cold. 15 I pull up my hempen bedclothes, 16 Wear what few sleeveless clothes I have, 17 But cold and bitter is the night! 18 As for those poorer than myself, 19 Their parents must be cold and hungry, 20 Their wives and children beg and cry. 21 Then, how do you struggle through life? 22 23 Wide as they call the heaven and earth, 24 For me they have shrunk quite small; 25 Bright though they call the sun and moon, 26 They never shine for me. 27 Is it the same with all men, 28 Or for me alone? 29 By rare chance I was born a man 30 And no meaner than my fellows, 31 But, wearing unwadded sleeveless clothes 32 In tatters, like weeds waving in the sea, 33 Hanging from my shoulders, 34 And under the sunken roof, 35 Within the leaning walls, 36 Here I lie on straw 37 Spread on bare earth, 38 With my parents at my pillow, 39 And my wife and children at my feet, 40 All huddled in grief and tears. 41 No fire sends up smoke 42 At the cooking-place, 43 And in the cauldron 44 A spider spins its web. 45 With not a grain to cook, 46 We moan like the night thrush. 47 Then, "to cut," as the saying is, 48 "The ends of what is already too short," 49 The village headman comes, 50 With rod in hand, to our sleeping place, 51 Growling for his dues. 52 Must it be so hopeless -- 53 The way of this world? 54 55 -- Yamanoue Okura dist/sed/testsuite/linecnt.inp100660 765 144 3704 6070131555 15342 0ustar lordusersA dialogue on poverty On the night when the rain beats, Driven by the wind, On the night when the snowflakes mingle With a sleety rain, I feel so helplessly cold. I nibble at a lump of salt, Sip the hot, oft-diluted dregs of _sake_; And coughing, snuffling, And stroking my scanty beard, I say in my pride, "There's none worthy, save I!" But I shiver still with cold. I pull up my hempen bedclothes, Wear what few sleeveless clothes I have, But cold and bitter is the night! As for those poorer than myself, Their parents must be cold and hungry, Their wives and children beg and cry. Then, how do you struggle through life? Wide as they call the heaven and earth, For me they have shrunk quite small; Bright though they call the sun and moon, They never shine for me. Is it the same with all men, Or for me alone? By rare chance I was born a man And no meaner than my fellows, But, wearing unwadded sleeveless clothes In tatters, like weeds waving in the sea, Hanging from my shoulders, And under the sunken roof, Within the leaning walls, Here I lie on straw Spread on bare earth, With my parents at my pillow, And my wife and children at my feet, All huddled in grief and tears. No fire sends up smoke At the cooking-place, And in the cauldron A spider spins its web. With not a grain to cook, We moan like the night thrush. Then, "to cut," as the saying is, "The ends of what is already too short," The village headman comes, With rod in hand, to our sleeping place, Growling for his dues. Must it be so hopeless -- The way of this world? -- Yamanoue Okura dist/sed/testsuite/linecnt.sed100660 765 144 2 6070131555 15233 0ustar lordusers= dist/sed/testsuite/mac-mf.good100660 765 144 14132 6070131555 15225 0ustar lordusers## config:mac-pre.in ## common Macintosh prefix for all Makefile.in in the Kerberos V5 tree # # MPW-style lines for the MakeFile # # This first part is long enough that NFS:Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make # # This first part is long enough that NFS:Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make # # This first part is long enough that NFS:Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make # # This first part is long enough that NFS:Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make # # This first part is long enough that NFS:Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make # # This first part is long enough that NFS:Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make # # This first part is long enough that NFS:Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make # # This first part is long enough that NFS:Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make # # This first part is long enough that NFS:Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make # # This first part is long enough that NFS:Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make # # This first part is long enough that NFS:Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make # # This first part is long enough that NFS:Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make # # This first part is long enough that NFS:Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make # # This first part is long enough that NFS:Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make # # End of MPW-style lines for MakeFile # WHAT = mac # Directory syntax Ä R= C= S=: U=: BUILDTOP = ::: srcdir = # FIXME Ä This doesn't translate to MPW yet, srcdir must be same as objdir # File in object dir can come from either the current dir or srcdir # # . Ä . "{srcdir}" # Default rule that puts each file into separate segment .c.o Ä .c {CC} {DepDir}{Default}.c {CFLAGS} -s {Default} -o {TargDir}{Default}.c.o CPPFLAGS = -i {SRCTOP}:include -i {BUILDTOP}:include -i {SRCTOP}:include:krb5 -i {BUILDTOP}:include:krb5 -i {CIncludes} DEFS = {CPPFLAGS} CC = c LD = link # The funny quoting in the LDFLAGS is to avoid xxx.c.o being mangled by # mac-mf.sed into xxx.c.o LDFLAGS=-t MPST -c "MPS " -sym on {Libraries}"Runtime."o {CLibraries}"StdClib."o {Libraries}"ToolLibs."o {Libraries}"Interface."o CCOPTS = LIBS = KRB5ROOT= @KRB5ROOT@ KRB4=@KRB4@ INSTALL=Duplicate -y INSTALL_PROGRAM=Duplicate -y INSTALL_DATA=Duplicate -y INSTALL_SETUID=Duplicate -y KRB5MANROOT = {KRB5ROOT}{S}man ADMIN_BINDIR = {KRB5ROOT}{S}admin SERVER_BINDIR = {KRB5ROOT}{S}sbin CLIENT_BINDIR = {KRB5ROOT}{S}bin ADMIN_MANDIR = {KRB5MANROOT}{S}man8 SERVER_MANDIR = {KRB5MANROOT}{S}man8 CLIENT_MANDIR = {KRB5MANROOT}{S}man1 FILE_MANDIR = {KRB5MANROOT}{S}man5 KRB5_LIBDIR = {KRB5ROOT}{S}lib KRB5_INCDIR = {KRB5ROOT}{S}include KRB5_INCSUBDIRS = ¶ {KRB5_INCDIR}{S}krb5 ¶ {KRB5_INCDIR}{S}asn.1 ¶ {KRB5_INCDIR}{S}kerberosIV RM = Delete -y -i CP = Duplicate -y MV = mv -f CHMOD=chmod RANLIB = @RANLIB@ ARCHIVE = @ARCHIVE@ ARADD = @ARADD@ LN = @LN_S@ AWK = @AWK@ LEX = @LEX@ LEXLIB = @LEXLIB@ YACC = @YACC@ # FIXME Ä This won't work for srcdir != objdir. But on the Mac, there # is no easy way to build a relative or absolute path, because Ä means # both the path separator, and the "go up a directory" indicator #SRCTOP = {srcdir}{S}{BUILDTOP} SRCTOP = {BUILDTOP} SUBDIRS = @subdirs@ TOPLIBD = {BUILDTOP}{S}lib OBJEXT = c.o LIBEXT = a EXEEXT = all ÄÄ # Generated automatically from Makefile.in by configure CFLAGS = {CCOPTS} {DEFS} -i ::des ##DOSBUILDTOP = ..\..\: ##DOSLIBNAME=..\crypto.lib ##DOS!include {BUILDTOP}\config\windows.in OBJS= md5.{OBJEXT} md5glue.{OBJEXT} md5crypto.{OBJEXT} SRCS= md5.c md5glue.c md5crypto.c all ÄÄ {OBJS} t_mddriver Ä t_mddriver.c.o md5.c.o Link {LDFLAGS} -o t_mddriver t_mddriver.c.o md5.c.o t_mddriver.exe Ä {CC} {CFLAGS2} -o t_mddriver.exe t_mddriver.c md5.c check ÄÄ t_mddriver{EXEEXT} {C}t_mddriver{EXEEXT} -x clean ÄÄ {RM} t_mddriver{EXEEXT} t_mddriver.{OBJEXT} # config:post.in # put all ÄÄ first just in case no other rules occur here # all ÄÄ check ÄÄ clean ÄÄ clean-{WHAT} {RM} config.log pre.c.out post.c.out Makefile.c.out clean-unix ÄÄ if test -n "{OBJS}" ; then {RM} {OBJS}; else Ä ; fi clean-windows ÄÄ {RM} Å.{OBJEXT} {RM} msvc.pdb Å.err dist/sed/testsuite/mac-mf.inp100660 765 144 14253 6070131555 15067 0ustar lordusers## config/mac-pre.in ## common Macintosh prefix for all Makefile.in in the Kerberos V5 tree. # # MPW-style lines for the MakeFile. # # This first part is long enough that NFS/Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make. # # This first part is long enough that NFS/Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make. # # This first part is long enough that NFS/Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make. # # This first part is long enough that NFS/Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make. # # This first part is long enough that NFS/Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make. # # This first part is long enough that NFS/Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make. # # This first part is long enough that NFS/Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make. # # This first part is long enough that NFS/Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make. # # This first part is long enough that NFS/Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make. # # This first part is long enough that NFS/Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make. # # This first part is long enough that NFS/Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make. # # This first part is long enough that NFS/Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make. # # This first part is long enough that NFS/Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make. # # This first part is long enough that NFS/Share doesn't notice the non-ASCII # characters in the rest of the file, so it claims that the file is type # TEXT, which is what we want. The non-ASCII chars are necessary for MPW # Make. # # End of MPW-style lines for MakeFile. # WHAT = mac # Directory syntax: R= C= S=: U=: BUILDTOP = ../../.. srcdir = . # FIXME: This doesn't translate to MPW yet, srcdir must be same as objdir. # File in object dir can come from either the current dir or srcdir. # # . : . "{srcdir}" # Default rule that puts each file into separate segment. .c.o: .c {CC} {DepDir}{Default}.c {CFLAGS} -s {Default} -o {TargDir}{Default}.c.o CPPFLAGS = -I$(SRCTOP)/include -I$(BUILDTOP)/include -I$(SRCTOP)/include/krb5 -I$(BUILDTOP)/include/krb5 -i {CIncludes} DEFS = $(CPPFLAGS) CC = c LD = link # The funny quoting in the LDFLAGS is to avoid xxx.o being mangled by # mac-mf.sed into xxx.c.o. LDFLAGS=-t MPST -c "MPS " -sym on {Libraries}"Runtime."o {CLibraries}"StdClib."o {Libraries}"ToolLibs."o {Libraries}"Interface."o CCOPTS = LIBS = KRB5ROOT= @KRB5ROOT@ KRB4=@KRB4@ INSTALL=Duplicate -y INSTALL_PROGRAM=Duplicate -y INSTALL_DATA=Duplicate -y INSTALL_SETUID=Duplicate -y KRB5MANROOT = $(KRB5ROOT)$(S)man ADMIN_BINDIR = $(KRB5ROOT)$(S)admin SERVER_BINDIR = $(KRB5ROOT)$(S)sbin CLIENT_BINDIR = $(KRB5ROOT)$(S)bin ADMIN_MANDIR = $(KRB5MANROOT)$(S)man8 SERVER_MANDIR = $(KRB5MANROOT)$(S)man8 CLIENT_MANDIR = $(KRB5MANROOT)$(S)man1 FILE_MANDIR = $(KRB5MANROOT)$(S)man5 KRB5_LIBDIR = $(KRB5ROOT)$(S)lib KRB5_INCDIR = $(KRB5ROOT)$(S)include KRB5_INCSUBDIRS = \ $(KRB5_INCDIR)$(S)krb5 \ $(KRB5_INCDIR)$(S)asn.1 \ $(KRB5_INCDIR)$(S)kerberosIV RM = Delete -y -i CP = Duplicate -y MV = mv -f CHMOD=chmod RANLIB = @RANLIB@ ARCHIVE = @ARCHIVE@ ARADD = @ARADD@ LN = @LN_S@ AWK = @AWK@ LEX = @LEX@ LEXLIB = @LEXLIB@ YACC = @YACC@ # FIXME: This won't work for srcdir != objdir. But on the Mac, there # is no easy way to build a relative or absolute path, because : means # both the path separator, and the "go up a directory" indicator. #SRCTOP = $(srcdir)$(S)$(BUILDTOP) SRCTOP = $(BUILDTOP) SUBDIRS = @subdirs@ TOPLIBD = $(BUILDTOP)$(S)lib OBJEXT = c.o LIBEXT = a EXEEXT = all:: # Generated automatically from Makefile.in by configure. CFLAGS = $(CCOPTS) $(DEFS) -I$(srcdir)/../des ##DOSBUILDTOP = ..\..\.. ##DOSLIBNAME=..\crypto.lib ##DOS!include $(BUILDTOP)\config\windows.in OBJS= md5.$(OBJEXT) md5glue.$(OBJEXT) md5crypto.$(OBJEXT) SRCS= $(srcdir)/md5.c $(srcdir)/md5glue.c $(srcdir)/md5crypto.c all:: $(OBJS) t_mddriver: t_mddriver.o md5.o $(CC) $(CFLAGS) $(LDFLAGS) -o t_mddriver t_mddriver.o md5.o t_mddriver.exe: $(CC) $(CFLAGS2) -o t_mddriver.exe t_mddriver.c md5.c check:: t_mddriver$(EXEEXT) $(C)t_mddriver$(EXEEXT) -x clean:: $(RM) t_mddriver$(EXEEXT) t_mddriver.$(OBJEXT) # config/post.in # put all:: first just in case no other rules occur here # all:: check:: clean:: clean-$(WHAT) $(RM) config.log pre.out post.out Makefile.out clean-unix:: if test -n "$(OBJS)" ; then $(RM) $(OBJS); else :; fi clean-windows:: $(RM) *.$(OBJEXT) $(RM) msvc.pdb *.err dist/sed/testsuite/mac-mf.sed100640 765 144 10740 6070131555 15047 0ustar lordusers# Rewrite default rules from .c.o: to .c.o: .c /^\./s/^\(\.[a-z]*\)\(\.[a-z]*\)\( *: *\)$/\1\2\3 \1/ # Change dependency char. /::/s/::/ \\Option-f\\Option-f /g /:/s/:/ \\Option-f /g /^[SU]=/s/ \\Option-f /:/g # Change syntax of Makefile vars. /\$/s/\${\([a-zA-Z0-9_]*\)}/{\1}/g /\$/s/\$(\([a-zA-Z0-9_]*\))/{\1}/g # Change $@ to {targ} /\$@/s/\$@/{targ}/g # Change pathname syntax. # # If line ends with .. then assume it sets a variable that will # be used to prefix something else -- eliminate one colon, assuming # that a slash after the ${name} will turn into the missing colon. # Mac pathname conventions are IRREGULAR and UGLY! /\./s,\.\./\.\.$,::, /\./s,\.\.$,:, # Same if it ends with . (a single dot); turn it into nothing. /\./s,\.$,,g # Rules for .. and . elsewhere in the line # Convert ../: to ::, recur to get whole paths. /\./s,\.\./:,::,g # Convert ../../ to ::: /\./s,\.\./\.\./,:::,g /\./s,\.\./,::,g /\.\//s,\./,:,g /\//s,/,:,g /=/s/ = \.$/ = :/ # Comment out any explicit srcdir setting. # /srcdir/s/^srcdir/# srcdir/ /version/s/^version=/# version=/ /BASEDIR/s/^BASEDIR =.*$/BASEDIR = "{srcroot}"/ /{BASEDIR}:/s/{BASEDIR}:/{BASEDIR}/g # The original lines screw up -I$(srcdir)/../des by eliminating a colon. # Proposed fix: Eliminate srcdir prefixes totally. #/{srcdir}:/s/{srcdir}:/"{srcdir}"/g /{srcdir}:/s/{srcdir}://g #/"{srcdir}":/s/"{srcdir}":/"{srcdir}"/g # Comment out settings of anything set by mpw host config. ##/CC/s/^CC *=/#CC =/ ##/CFLAGS/s/^CFLAGS *=/#CFLAGS =/ ##/LDFLAGS/s/^LDFLAGS *=/#LDFLAGS =/ # Change -I usage. /-I/s/-I\./-i :/g /-I/s/-I::bfd/-i ::bfd:/g /-I/s/-I::include/-i ::include:/g /-I/s/-I/-i /g # Change -D usage. /-D/s/\([ =]\)-D\([^ ]*\)/\1-d \2/g # Change continuation char. /\\$/s/\\$/\\Option-d/ # Change wildcard char. /^[^#]/s/\*/\\Option-x/g # Change path of various types of source files. #/\.[chly]/s/\([ ><=]\)\([-a-zA-Z0-9_$:"]*\)\.\([chly]\)/\1"{s}"\2.\3/g #/\.[chly]/s/^\([-a-zA-Z0-9_${}:"]*\)\.\([chly]\)/"{s}"\1.\2/g # Skip the {s} and {o} business for now... # Fix some overenthusiasms. #/{s}/s/"{s}""{srcdir}"/"{srcdir}"/g #/{s}/s/"{s}"{\([a-zA-Z0-9_]*\)dir}/"{\1dir}"/g #/{s}/s/"{s}"{\([a-zA-Z0-9_]*\)DIR}/"{\1DIR}"/g #/{s}/s/"{s}""{\([a-zA-Z0-9_]*\)dir}"/"{\1dir}"/g #/{s}/s/"{s}""{\([a-zA-Z0-9_]*\)DIR}"/"{\1DIR}"/g #/{s}/s/"{s}":/:/g #/{s}/s/^"{s}"//g #/^\./s/"{s}"\././g # Change extension and path of objects, except in the OBJEXT line. #/^OBJEXT/!s/\([ =]\)\([-a-zA-Z0-9_${}:"]*\)\.o/\1"{o}"\2.c.o/g #/\.o/s/^\([-a-zA-Z0-9_${}:"]*\)\.o/"{o}"\1.c.o/g # Skip the {o} stuff for now... /^OBJEXT/!s/\([ =]\)\([-a-zA-Z0-9_${}:"]*\)\.o/\1\2.c.o/g /\.o/s/^\([-a-zA-Z0-9_${}:"]*\)\.o/\1.c.o/g # Clean up. #/\.o/s/"{o}""{o}"/"{o}"/g #/{o}/s/^"{o}"\([a-zA-Z0-9_]*\)=/\1=/g # Change extension of libs. # /\.a/s/lib\([a-z]*\)\.a/lib\1.o/g # Remove non-echo option. /^ -/s/^ -/ / # Change cp to duplicate. # /cp/s/^\([ ]*\)cp /\1Duplicate -d -y / # Change mv to rename. # /mv/s/^\([ ]*\)mv /\1Rename -y / # /Rename/s/^\([ ]*\)Rename -y -f/\1Rename -y/ # Change rm to delete. /^RM=/s/rm -f/Delete -i -y/ # /rm/s/^\([ ]*\)rm /\1Delete -y / # /Delete/s/^\([ ]*\)Delete -y -f/\1Delete -y/ # Comment out symlinking. # /ln/s/^\([ ]*\)ln /\1# ln / # Remove -c from explicit compiler calls. # /-c/s/{CC}\(.*\) -c \(.*\)\([-a-z]*\)\.c/{CC}\1 \2\3.c -o "{o}"\3.c.o/g # Don't ask... prev subst seems to omit the second filename. # /-o/s/\([-a-z]*\)\.c -o "{o}".c.o/\1\.c -o "{o}"\1.c.o/ # Change linking cc to link. /LDFLAGS/ s/{CC} \(.*\){CFLAGS}\(.*\){LDFLAGS}/Link \1 \2 {LDFLAGS}/ /CFLAGS_LINK/s/{CC} \(.*\){CFLAGS_LINK}\(.*\){LDFLAGS}/Link \1 \2 {LDFLAGS}/ # Comment out .PHONY rules. /\.PHONY/s/^\.PHONY/# \.PHONY/ # Comment out .SUFFIXES rules. /\.SUFFIXES/s/^\.SUFFIXES/# \.SUFFIXES/ # Comment out .PRECIOUS rules. /\.PRECIOUS/s/^\.PRECIOUS/# \.PRECIOUS/ ## Comment out default rules. ##/^\./s/^\(\.[a-z]*\.[a-z]* \)/# \1/ # # End of original hack-mf.sed # # Begin original hack-mf2.sed # # Transform expressions. # Set the install program appropriate. # /INSTALL/s/^INSTALL *= *`.*`:install.sh -c/INSTALL = Duplicate -y/ # Include from the extra-include dir. # /^INCLUDES = /s/^INCLUDES = /INCLUDES = -i "{srcroot}"extra-include / # Yuck - remove unconverted autoconf things. # /@/s/@[^ ]*@//g # Hackery, pure and simple # To speed up compiles, remove duplicated -i options. /-i/s/\(-i [^ ]*\) \1 /\1 /g # Note! There are 8-bit characters in the three lines below: # 0xc4, 0xb6, 0xc5. /Option/s/\\Option-f/Ä/g /Option/s/\\Option-d/¶/g /Option/s/\\Option-x/Å/g dist/sed/testsuite/maddingcrowd.good100660 765 144 12515 6070131555 16532 0ustar lordusersThe girl on the summit of the load sat motionless, surrounded by tables and chairs with their legs upwards, backed by an oak settle, and ornamented in front by pots of geraniums, myrtles, and cactuses, together with a caged canary -- all probably from the windows of the house just vacated. There was also a cat in a willow basket, from the partly-opened lid of which she gazed with half-closed eyes, and affectionately-surveyed the small birds around. The handsome girl waited for some time idly in her place, and the only sound heard in the stillness was the hopping of the canary up and down the perches of its prison. Then she looked attentively downwards. It was not at the bird, nor at the cat; it was at an oblong package tied in paper, and lying between them. She turned her head to learn if the waggoner were coming. He was not yet in sight; and her eyes crept back to the package, her thoughts seeming to run upon what was inside it. At length she drew the article into her lap, and untied the paper covering; a small swing looking- glass was disclosed, in which she proceeded to survey herself attentively. She parted her lips and smiled. It was a fine morning, and the sun lighted up to a scarlet glow the crimson jacket she wore, and painted a soft lustre upon her bright face and dark hair. The myrtles, geraniums, and cactuses packed around her were fresh and green, and at such a leafless season they invested the whole concern of horses, waggon, furniture, and girl with a peculiar vernal charm. What possessed her to indulge in such a performance in the sight of the sparrows, blackbirds, and unperceived farmer who were alone its spectators, -- whether the smile began as a factitious one, to test her capacity in that art, -- nobody knows; it ended certainly in a real smile. She blushed at herself, and seeing her reflection blush, blushed the more. The change from the customary spot and necessary occasion of such an act -- from the dressing hour in a bedroom to a time of travelling out of doors -- lent to the idle deed a novelty it did not intrinsically possess. The picture was a delicate one. Woman's prescriptive infirmity had stalked into the sunlight, which had clothed it in the freshness of an originality. A cynical inference was irresistible by Gabriel Oak as he regarded the scene, generous though he fain would have been. There was no necessity whatever for her looking in the glass. She did not adjust her hat, or pat her hair, or press a dimple into shape, or do one thing to signify that any such intention had been her motive in taking up the glass. She simply observed herself as a fair product of Nature in the feminine kind, her thoughts seeming to glide into far-off though likely dramas in which men would play a part -- vistas of probable triumphs -- the smiles being of a phase suggesting that hearts were imagined as lost and won. Still, this was but conjecture, and the whole series of actions was so idly put forth as to make it rash to assert that intention had any part in them at all. The waggoner's steps were heard returning. She put the glass in the paper, and the whole again into its place. When the waggon had passed on, Gabriel withdrew from his point of espial, and descending into the road, followed the vehicle to the turnpike-gate some way beyond the bottom of the hill, where the object of his contemplation now halted for the payment of toll. About twenty steps still remained between him and the gate, when he heard a dispute. It was a difference concerning twopence between the persons with the waggon and the man at the toll-bar. "Mis'ess's niece is upon the top of the things, and she says that's enough that I've offered ye, you great miser, and she won't pay any more." These were the waggoner's words. "Very well; then mis'ess's niece can't pass," said the turnpike-keeper, closing the gate. Oak looked from one to the other of the disputants, and fell into a reverie. There was something in the tone of twopence remarkably insignificant. Threepence had a definite value as money -- it was an appreciable infringement on a day's wages, and, as such, a higgling matter; but twopence -- "Here," he said, stepping forward and handing twopence to the gatekeeper; "let the young woman pass." He looked up at her then; she heard his words, and looked down. Gabriel's features adhered throughout their form so exactly to the middle line between the beauty of St. John and the ugliness of Judas Iscariot, as represented in a window of the church he attended, that not a single lineament could be selected and called worthy either of distinction or notoriety. The red-jacketed and dark-haired maiden seemed to think so too, for she carelessly glanced over him, and told her man to drive on. She might have looked her thanks to Gabriel on a minute scale, but she did not speak them; more probably she felt none, for in gaining her a passage he had lost her her point, and we know how women take a favour of that kind. The gatekeeper surveyed the retreating vehicle. "That's a handsome maid," he said to Oak. "But she has her faults," said Gabriel. "True, farmer." "And the greatest of them is -- well, what it is always." "Beating people down? ay, 'tis so." "O no." "What, then?" Gabriel, perhaps a little piqued by the comely traveller's indifference, glanced back to where he had witnessed her performance over the hedge, and said, "Vanity, dude." dist/sed/testsuite/maddingcrowd.inp100660 765 144 12507 6070131555 16371 0ustar lordusersThe girl on the summit of the load sat motionless, surrounded by tables and chairs with their legs upwards, backed by an oak settle, and ornamented in front by pots of geraniums, myrtles, and cactuses, together with a caged canary -- all probably from the windows of the house just vacated. There was also a cat in a willow basket, from the partly-opened lid of which she gazed with half-closed eyes, and affectionately-surveyed the small birds around. The handsome girl waited for some time idly in her place, and the only sound heard in the stillness was the hopping of the canary up and down the perches of its prison. Then she looked attentively downwards. It was not at the bird, nor at the cat; it was at an oblong package tied in paper, and lying between them. She turned her head to learn if the waggoner were coming. He was not yet in sight; and her eyes crept back to the package, her thoughts seeming to run upon what was inside it. At length she drew the article into her lap, and untied the paper covering; a small swing looking- glass was disclosed, in which she proceeded to survey herself attentively. She parted her lips and smiled. It was a fine morning, and the sun lighted up to a scarlet glow the crimson jacket she wore, and painted a soft lustre upon her bright face and dark hair. The myrtles, geraniums, and cactuses packed around her were fresh and green, and at such a leafless season they invested the whole concern of horses, waggon, furniture, and girl with a peculiar vernal charm. What possessed her to indulge in such a performance in the sight of the sparrows, blackbirds, and unperceived farmer who were alone its spectators, -- whether the smile began as a factitious one, to test her capacity in that art, -- nobody knows; it ended certainly in a real smile. She blushed at herself, and seeing her reflection blush, blushed the more. The change from the customary spot and necessary occasion of such an act -- from the dressing hour in a bedroom to a time of travelling out of doors -- lent to the idle deed a novelty it did not intrinsically possess. The picture was a delicate one. Woman's prescriptive infirmity had stalked into the sunlight, which had clothed it in the freshness of an originality. A cynical inference was irresistible by Gabriel Oak as he regarded the scene, generous though he fain would have been. There was no necessity whatever for her looking in the glass. She did not adjust her hat, or pat her hair, or press a dimple into shape, or do one thing to signify that any such intention had been her motive in taking up the glass. She simply observed herself as a fair product of Nature in the feminine kind, her thoughts seeming to glide into far-off though likely dramas in which men would play a part -- vistas of probable triumphs -- the smiles being of a phase suggesting that hearts were imagined as lost and won. Still, this was but conjecture, and the whole series of actions was so idly put forth as to make it rash to assert that intention had any part in them at all. The waggoner's steps were heard returning. She put the glass in the paper, and the whole again into its place. When the waggon had passed on, Gabriel withdrew from his point of espial, and descending into the road, followed the vehicle to the turnpike-gate some way beyond the bottom of the hill, where the object of his contemplation now halted for the payment of toll. About twenty steps still remained between him and the gate, when he heard a dispute. It was a difference concerning twopence between the persons with the waggon and the man at the toll-bar. "Mis'ess's niece is upon the top of the things, and she says that's enough that I've offered ye, you great miser, and she won't pay any more." These were the waggoner's words. "Very well; then mis'ess's niece can't pass," said the turnpike-keeper, closing the gate. Oak looked from one to the other of the disputants, and fell into a reverie. There was something in the tone of twopence remarkably insignificant. Threepence had a definite value as money -- it was an appreciable infringement on a day's wages, and, as such, a higgling matter; but twopence -- "Here," he said, stepping forward and handing twopence to the gatekeeper; "let the young woman pass." He looked up at her then; she heard his words, and looked down. Gabriel's features adhered throughout their form so exactly to the middle line between the beauty of St. John and the ugliness of Judas Iscariot, as represented in a window of the church he attended, that not a single lineament could be selected and called worthy either of distinction or notoriety. The red-jacketed and dark-haired maiden seemed to think so too, for she carelessly glanced over him, and told her man to drive on. She might have looked her thanks to Gabriel on a minute scale, but she did not speak them; more probably she felt none, for in gaining her a passage he had lost her her point, and we know how women take a favour of that kind. The gatekeeper surveyed the retreating vehicle. "That's a handsome maid," he said to Oak. "But she has her faults," said Gabriel. "True, farmer." "And the greatest of them is -- well, what it is always." "Beating people down? ay, 'tis so." "O no." "What, then?" Gabriel, perhaps a little piqued by the comely traveller's indifference, glanced back to where he had witnessed her performance over the hedge, and said, "Vanity." dist/sed/testsuite/maddingcrowd.sed100660 765 144 25603 6070131556 16360 0ustar lordusers# this is from Thomas Hardy's _Far From the Madding Crowd_. # # cf ftp://mrcnext.cso.uiuc.edu/pub/etext/etext94/crowd10a.txt # # the point of this test, in case it isn't obvious, is to overfill fixed # buffers wherever they might be. # s/The girl on the summit of the load sat motionless, surrounded by tables and chairs with their legs upwards, backed by an oak settle, and ornamented in front by pots of geraniums, myrtles, and cactuses, together with a caged canary -- all probably from the windows of the house just vacated. There was also a cat in a willow basket, from the partly-opened lid of which she gazed with half-closed eyes, and affectionately-surveyed the small birds around. The handsome girl waited for some time idly in her place, and the only sound heard in the stillness was the hopping of the canary up and down the perches of its prison. Then she looked attentively downwards. It was not at the bird, nor at the cat; it was at an oblong package tied in paper, and lying between them. She turned her head to learn if the waggoner were coming. He was not yet in sight; and her eyes crept back to the package, her thoughts seeming to run upon what was inside it. At length she drew the article into her lap, and untied the paper covering; a small swing looking- glass was disclosed, in which she proceeded to survey herself attentively. She parted her lips and smiled. It was a fine morning, and the sun lighted up to a scarlet glow the crimson jacket she wore, and painted a soft lustre upon her bright face and dark hair. The myrtles, geraniums, and cactuses packed around her were fresh and green, and at such a leafless season they invested the whole concern of horses, waggon, furniture, and girl with a peculiar vernal charm. What possessed her to indulge in such a performance in the sight of the sparrows, blackbirds, and unperceived farmer who were alone its spectators, -- whether the smile began as a factitious one, to test her capacity in that art, -- nobody knows; it ended certainly in a real smile. She blushed at herself, and seeing her reflection blush, blushed the more. The change from the customary spot and necessary occasion of such an act -- from the dressing hour in a bedroom to a time of travelling out of doors -- lent to the idle deed a novelty it did not intrinsically possess. The picture was a delicate one. Woman's prescriptive infirmity had stalked into the sunlight, which had clothed it in the freshness of an originality. A cynical inference was irresistible by Gabriel Oak as he regarded the scene, generous though he fain would have been. There was no necessity whatever for her looking in the glass. She did not adjust her hat, or pat her hair, or press a dimple into shape, or do one thing to signify that any such intention had been her motive in taking up the glass. She simply observed herself as a fair product of Nature in the feminine kind, her thoughts seeming to glide into far-off though likely dramas in which men would play a part -- vistas of probable triumphs -- the smiles being of a phase suggesting that hearts were imagined as lost and won. Still, this was but conjecture, and the whole series of actions was so idly put forth as to make it rash to assert that intention had any part in them at all. The waggoner's steps were heard returning. She put the glass in the paper, and the whole again into its place. When the waggon had passed on, Gabriel withdrew from his point of espial, and descending into the road, followed the vehicle to the turnpike-gate some way beyond the bottom of the hill, where the object of his contemplation now halted for the payment of toll. About twenty steps still remained between him and the gate, when he heard a dispute. It was a difference concerning twopence between the persons with the waggon and the man at the toll-bar. "Mis'ess's niece is upon the top of the things, and she says that's enough that I've offered ye, you great miser, and she won't pay any more." These were the waggoner's words. "Very well; then mis'ess's niece can't pass," said the turnpike-keeper, closing the gate. Oak looked from one to the other of the disputants, and fell into a reverie. There was something in the tone of twopence remarkably insignificant. Threepence had a definite value as money -- it was an appreciable infringement on a day's wages, and, as such, a higgling matter; but twopence -- "Here," he said, stepping forward and handing twopence to the gatekeeper; "let the young woman pass." He looked up at her then; she heard his words, and looked down. Gabriel's features adhered throughout their form so exactly to the middle line between the beauty of St. John and the ugliness of Judas Iscariot, as represented in a window of the church he attended, that not a single lineament could be selected and called worthy either of distinction or notoriety. The red-jacketed and dark-haired maiden seemed to think so too, for she carelessly glanced over him, and told her man to drive on. She might have looked her thanks to Gabriel on a minute scale, but she did not speak them; more probably she felt none, for in gaining her a passage he had lost her her point, and we know how women take a favour of that kind. The gatekeeper surveyed the retreating vehicle. "That's a handsome maid," he said to Oak. "But she has her faults," said Gabriel. "True, farmer." "And the greatest of them is -- well, what it is always." "Beating people down? ay, 'tis so." "O no." "What, then?" Gabriel, perhaps a little piqued by the comely traveller's indifference, glanced back to where he had witnessed her performance over the hedge, and said, "Vanity."/The girl on the summit of the load sat motionless, surrounded by tables and chairs with their legs upwards, backed by an oak settle, and ornamented in front by pots of geraniums, myrtles, and cactuses, together with a caged canary -- all probably from the windows of the house just vacated. There was also a cat in a willow basket, from the partly-opened lid of which she gazed with half-closed eyes, and affectionately-surveyed the small birds around. The handsome girl waited for some time idly in her place, and the only sound heard in the stillness was the hopping of the canary up and down the perches of its prison. Then she looked attentively downwards. It was not at the bird, nor at the cat; it was at an oblong package tied in paper, and lying between them. She turned her head to learn if the waggoner were coming. He was not yet in sight; and her eyes crept back to the package, her thoughts seeming to run upon what was inside it. At length she drew the article into her lap, and untied the paper covering; a small swing looking- glass was disclosed, in which she proceeded to survey herself attentively. She parted her lips and smiled. It was a fine morning, and the sun lighted up to a scarlet glow the crimson jacket she wore, and painted a soft lustre upon her bright face and dark hair. The myrtles, geraniums, and cactuses packed around her were fresh and green, and at such a leafless season they invested the whole concern of horses, waggon, furniture, and girl with a peculiar vernal charm. What possessed her to indulge in such a performance in the sight of the sparrows, blackbirds, and unperceived farmer who were alone its spectators, -- whether the smile began as a factitious one, to test her capacity in that art, -- nobody knows; it ended certainly in a real smile. She blushed at herself, and seeing her reflection blush, blushed the more. The change from the customary spot and necessary occasion of such an act -- from the dressing hour in a bedroom to a time of travelling out of doors -- lent to the idle deed a novelty it did not intrinsically possess. The picture was a delicate one. Woman's prescriptive infirmity had stalked into the sunlight, which had clothed it in the freshness of an originality. A cynical inference was irresistible by Gabriel Oak as he regarded the scene, generous though he fain would have been. There was no necessity whatever for her looking in the glass. She did not adjust her hat, or pat her hair, or press a dimple into shape, or do one thing to signify that any such intention had been her motive in taking up the glass. She simply observed herself as a fair product of Nature in the feminine kind, her thoughts seeming to glide into far-off though likely dramas in which men would play a part -- vistas of probable triumphs -- the smiles being of a phase suggesting that hearts were imagined as lost and won. Still, this was but conjecture, and the whole series of actions was so idly put forth as to make it rash to assert that intention had any part in them at all. The waggoner's steps were heard returning. She put the glass in the paper, and the whole again into its place. When the waggon had passed on, Gabriel withdrew from his point of espial, and descending into the road, followed the vehicle to the turnpike-gate some way beyond the bottom of the hill, where the object of his contemplation now halted for the payment of toll. About twenty steps still remained between him and the gate, when he heard a dispute. It was a difference concerning twopence between the persons with the waggon and the man at the toll-bar. "Mis'ess's niece is upon the top of the things, and she says that's enough that I've offered ye, you great miser, and she won't pay any more." These were the waggoner's words. "Very well; then mis'ess's niece can't pass," said the turnpike-keeper, closing the gate. Oak looked from one to the other of the disputants, and fell into a reverie. There was something in the tone of twopence remarkably insignificant. Threepence had a definite value as money -- it was an appreciable infringement on a day's wages, and, as such, a higgling matter; but twopence -- "Here," he said, stepping forward and handing twopence to the gatekeeper; "let the young woman pass." He looked up at her then; she heard his words, and looked down. Gabriel's features adhered throughout their form so exactly to the middle line between the beauty of St. John and the ugliness of Judas Iscariot, as represented in a window of the church he attended, that not a single lineament could be selected and called worthy either of distinction or notoriety. The red-jacketed and dark-haired maiden seemed to think so too, for she carelessly glanced over him, and told her man to drive on. She might have looked her thanks to Gabriel on a minute scale, but she did not speak them; more probably she felt none, for in gaining her a passage he had lost her her point, and we know how women take a favour of that kind. The gatekeeper surveyed the retreating vehicle. "That's a handsome maid," he said to Oak. "But she has her faults," said Gabriel. "True, farmer." "And the greatest of them is -- well, what it is always." "Beating people down? ay, 'tis so." "O no." "What, then?" Gabriel, perhaps a little piqued by the comely traveller's indifference, glanced back to where he had witnessed her performance over the hedge, and said, "Vanity, dude."/ dist/sed/testsuite/manis.inp100660 765 144 731 6070131556 14773 0ustar lorduserss%@CFLAGS@%%g s%@CPPFLAGS@%-I/%g s%@CXXFLAGS@%-x c++%g s%@DEFS@%$DEFS%g s%@LDFLAGS@%-L/usr/lib%g s%@LIBS@%-lgnu -lbfd%g s%@exec_prefix@%%g s%@prefix@%$prefix%g s%@RANLIB@%$RANLIB%g s%@CC@%/usr/local/bin/gcc%g s%@CPP@%$CPP%g s%@XCFLAGS@%$XCFLAGS%g s%@XINCLUDES@%$XINCLUDES%g s%@XLIBS@%$XLIBS%g s%@XPROGS@%$XPROGS%g s%@TCLHDIR@%$TCLHDIR%g s%@TCLLIB@%$TCLLIB%g s%@TKHDIR@%$TKHDIR%g s%@TKLIB@%$TKLIB%g s%@PTY_TYPE@%$PTY_TYPE%g s%@EVENT_TYPE@%$EVENT_TYPE%g s%@SETUID@%$SETUID%g dist/sed/testsuite/manis.sed100660 765 144 430 6070131556 14754 0ustar lordusers# straight out of an autoconf-generated configure. # The input should look just like the input after this is run. # # Protect against being on the right side of a sed subst in config.status. s/%@/@@/; s/@%/@@/; s/%g$/@g/; /@g$/s/[\\\\&%]/\\\\&/g; s/@@/%@/; s/@@/@%/; s/@g$/%g/ dist/sed/testsuite/middle.good100660 765 144 310 6070131556 15255 0ustar lordusers universe. You will never succeed. Everything will fail and come to an end finally. But you have a chance to carve a little bit of order and maybe even beauty out of the raw materials that dist/sed/testsuite/middle.sed100660 765 144 5 6070131556 15041 0ustar lordusers3,5p dist/sed/testsuite/newjis.good100660 765 144 132 6070131556 15320 0ustar lordusers$B$H$J$j$NM9JX6I$K(B $B$?$F$+$1$?$N$O(B $B$?$F$+$1$?$+$C$?$+$i(B $B$?$F$+$1$?!#(B dist/sed/testsuite/newjis.inp100660 765 144 134 6070131556 15160 0ustar lordusers$B$H$J$j$N$?$1$,$-$K(B $B$?$F$+$1$?$N$O(B $B$?$F$+$1$?$+$C$?$+$i(B $B$?$F$+$1$?!#(B dist/sed/testsuite/newjis.sed100660 765 144 23 6070131556 15122 0ustar lorduserss/$?$1$,$-/M9JX6I/ dist/sed/testsuite/noeol.good100660 765 144 115 6070131557 15137 0ustar lordusersThis file is uniquewakuwaku in that it doeswakuwaku end in a newline.wakuwakudist/sed/testsuite/noeol.inp100660 765 144 65 6070131557 14761 0ustar lordusersThis file is unique in that it does end in a newline.dist/sed/testsuite/noeol.sed100660 765 144 17 6070131557 14743 0ustar lorduserss/$/wakuwaku/g dist/sed/testsuite/readin.good100660 765 144 1012 6070131557 15302 0ustar lordusers``Democracy will not come today, this year, nor ever through compromise and fear. MOO I have as much right as the other fellow has to stand on my two feet and own the land. MOO I tire so of hearing people say let things take their course, tomorrow is another day. MOO I do not need my freedom when I'm dead. MOO I cannot live on tomorrow's bread. MOO Freedom is a strong seed planted in a great need. MOO I live here, too. MOO I want freedom just as you.'' ``The Weary Blues'', Langston Hughes dist/sed/testsuite/readin.inp100660 765 144 756 6070131557 15136 0ustar lordusers``Democracy will not come today, this year, nor ever through compromise and fear. I have as much right as the other fellow has to stand on my two feet and own the land. I tire so of hearing people say let things take their course, tomorrow is another day. I do not need my freedom when I'm dead. I cannot live on tomorrow's bread. Freedom is a strong seed planted in a great need. I live here, too. I want freedom just as you.'' ``The Weary Blues'', Langston Hughes dist/sed/testsuite/readin.sed100660 765 144 22 6070131560 15057 0ustar lordusers/\.$/r readin.in2 dist/sed/testsuite/sep.good100660 765 144 243 6070131560 14606 0ustar lordusersmiss mary mack mack//mack/ran down slashes\aren't%used enough/in/casual-conversation/// possibly sentences would be more attractive if they ended in two slashes// dist/sed/testsuite/tmp100660 765 144 314 6071120021 13656 0ustar lordusersäƤâ¤\e8 ¤ßäÆ»ý¤Á ·¡¶ú¤â¤\e8 ¤ß·¡¶ú»ý¤Á ¤³¤Îµ\d6¤Ë º\daÅ\a6¤Þ¤¹»ù ²È´\d6¤«¤\ca ¹ð¤é¤µ¤\cd ¤½¤é¤ß¤Ä ÂçϤιñ¤Ï ¤ª¤·¤ã¤\ca¤Ù¤Æ ¤\ef¤ì¤³¤½µ\ef¤ì ¤·¤­¤\ca¤Ù¤Æ ¤\ef¤ì¤³¤½ ºÂ¤» ¤\ef¤Ë¤³¤½¤Ï ¹ð¤é¤á ²È¤ò¤â̾¤ò¤â dist/sed/testsuite/sep.inp100660 765 144 273 6070131560 14447 0ustar lordusersmiss mary mack mack//mack/ran down/the track track track slashes\aren't%used enough/in/casual-conversation/// possibly sentences would be more attractive if they ended in two slashes// dist/sed/testsuite/sep.sed100660 765 144 106 6070131560 14427 0ustar lordusers# inspired by an autoconf generated configure script. s%/[^/][^/]*$%% dist/sed/testsuite/spencer.inp100640 765 144 3324 6071054647 15350 0ustar lordusers0:abc:abc 1:abc:xbc 1:abc:axc 1:abc:abx 0:abc:xabcy 0:abc:ababc 0:ab*c:abc 0:ab*bc:abc 0:ab*bc:abbc 0:ab*bc:abbbbc 0:ab+bc:abbc 1:ab+bc:abc 1:ab+bc:abq 0:ab+bc:abbbbc 0:ab?bc:abbc 0:ab?bc:abc 1:ab?bc:abbbbc 0:ab?c:abc 0:^abc$:abc 1:^abc$:abcc 0:^abc:abcc 1:^abc$:aabc 0:abc$:aabc 0:^:abc 0:$:abc 0:a.c:abc 0:a.c:axc 0:a.*c:axyzc 1:a.*c:axyzd 1:a[bc]d:abc 0:a[bc]d:abd 1:a[b-d]e:abd 0:a[b-d]e:ace 0:a[b-d]:aac 0:a[-b]:a- 0:a[b-]:a- 2:a[b-a]:- 2:a[]b:- 2:a[:- 0:a]:a] 0:a[]]b:a]b 0:a[^bc]d:aed 1:a[^bc]d:abd 0:a[^-b]c:adc 1:a[^-b]c:a-c 1:a[^]b]c:a]c 0:a[^]b]c:adc 0:ab|cd:abc 0:ab|cd:abcd 0:()ef:def 0:()*:- 1:*a:- 0:^*:- 0:$*:- 1:(*)b:- 1:$b:b 2:a\:- 0:a\(b:a(b 0:a\(*b:ab 0:a\(*b:a((b 1:a\x:a\x 1:abc):- 2:(abc:- 0:((a)):abc 0:(a)b(c):abc 0:a+b+c:aabbabc 0:a**:- 0:a*?:- 0:(a*)*:- 0:(a*)+:- 0:(a|)*:- 0:(a*|b)*:- 0:(a+|b)*:ab 0:(a+|b)+:ab 0:(a+|b)?:ab 0:[^ab]*:cde 0:(^)*:- 0:(ab|)*:- 2:)(:- 1:abc: 1:abc: 0:a*: 0:([abc])*d:abbbcd 0:([abc])*bcd:abcd 0:a|b|c|d|e:e 0:(a|b|c|d|e)f:ef 0:((a*|b))*:- 0:abcd*efg:abcdefg 0:ab*:xabyabbbz 0:ab*:xayabbbz 0:(ab|cd)e:abcde 0:[abhgefdc]ij:hij 1:^(ab|cd)e:abcde 0:(abc|)ef:abcdef 0:(a|b)c*d:abcd 0:(ab|ab*)bc:abc 0:a([bc]*)c*:abc 0:a([bc]*)(c*d):abcd 0:a([bc]+)(c*d):abcd 0:a([bc]*)(c+d):abcd 0:a[bcd]*dcdcde:adcdcde 1:a[bcd]+dcdcde:adcdcde 0:(ab|a)b*c:abc 0:((a)(b)c)(d):abcd 0:[A-Za-z_][A-Za-z0-9_]*:alpha 0:^a(bc+|b[eh])g|.h$:abh 0:(bc+d$|ef*g.|h?i(j|k)):effgz 0:(bc+d$|ef*g.|h?i(j|k)):ij 1:(bc+d$|ef*g.|h?i(j|k)):effg 1:(bc+d$|ef*g.|h?i(j|k)):bcdd 0:(bc+d$|ef*g.|h?i(j|k)):reffgz 1:((((((((((a)))))))))):- 0:(((((((((a))))))))):a 1:multiple words of text:uh-uh 0:multiple words:multiple words, yeah 0:(.*)c(.*):abcde 1:\((.*),:(.*)\) 1:[k]:ab 0:abcd:abcd 0:a(bc)d:abcd 0:a[-]?c:ac 0:(....).*\1:beriberi dist/sed/testsuite/sub-all.good100660 765 144 107 6070131560 15355 0ustar lordusersbar bar fo oo f oo bar bar bar bar bar bar bar bar bar bar bar bar bar dist/sed/testsuite/sub-all.sed100660 765 144 14 6070131560 15155 0ustar lorduserss/foo/bar/g dist/sed/testsuite/sub-num.good100660 765 144 107 6070131560 15404 0ustar lordusersfoo foo fo oo f oo foo foo foo foo foo foo foo bar foo foo foo foo foo dist/sed/testsuite/sub-num.inp100660 765 144 107 6070131561 15243 0ustar lordusersfoo foo fo oo f oo foo foo foo foo foo foo foo foo foo foo foo foo foo dist/sed/testsuite/sub-num.sed100660 765 144 15 6070131561 15206 0ustar lorduserss/foo/bar/10 dist/sed/testsuite/subwrite.inp100660 765 144 212 6070131561 15516 0ustar lordusersNot some church, and not the state, Not some dark capricious fate. Who you are, and when you lose, Comes only from the things you choose. dist/sed/testsuite/subwrite.sed100660 765 144 21 6070131561 15461 0ustar lorduserss/you/YoU/w tmp2 dist/sed/testsuite/version.good100660 765 144 25 6070131562 15464 0ustar lordusersGNU sed version 2.05 dist/sed/testsuite/writeout.inp100660 765 144 220 6070131562 15534 0ustar lordusersFacts are simple and facts are straight Facts are lazy and facts are late Facts all come with points of view Facts don't do what I want them to dist/sed/testsuite/writeout.sed100660 765 144 22 6070131562 15501 0ustar lordusers/^Facts ar/w tmp2 dist/sed/testsuite/xabcx.good100660 765 144 102 6070131562 15120 0ustar lordusersroses are red violets are blue my feet are cold your feet are too dist/sed/testsuite/xabcx.inp100660 765 144 103 6070131562 14757 0ustar lordusersroses are red violets are blue my feet are cold your feet are blue dist/sed/testsuite/xabcx.sed100660 765 144 72 6070131563 14732 0ustar lordusers# from the ChangeLog (Fri May 21 1993) \xfeetxs/blue/too/ dist/sed/testsuite/distrib.sh100660 765 144 3015 6070131553 15163 0ustar lordusers#! /bin/sh # # This is stolen from C News # # # All this does is massage the headers so they look like what news # software expects. To:, Cc: and Resent-*: headers are masked. # Reply-To: is turned into references, which is questionable (could # just as well be dropped. # # The From: line is rewritten to use the "address (comments)" form # instead of "phrase " form our mailer uses. Also, addresses # with no "@domainname" are assumed to originate locally, and so are # given a domain. # # The Sender: field below reflects the address of the person who # maintains our mailing lists. The Approved: field is in a special # form, so that we can do bidirectional gatewaying. Any message # in a newsgroup that bears this stamp will not be fed into the # matching mailing list. sed=${1-sed} $sed -n -e "1{i\\ Path: mailnewsgateway }" \ -e ":a /^[Rr]eceived:/b r /^[Nn]ewsgroups:/b r /^[Pp]ath:/b r /^[Tt][Oo]:/s/^/Original-/ /^[Cc][Cc]:/s/^/Original-/ /^[Rr][Ee][Ss][Ee][Nn][Tt]-.*/s/^/Original-/ /^[Mm][Ee][Ss][Ss][Aa][Gg][Ee]-[Ii][Dd]:/s/@/.alt.buddha.fat.short.guy@/ s/^[Ii]n-[Rr]eply-[Tt]o:/References:/ /^From:/{ s/<\([^@]*\)>\$/<\1@$thissite>/ s/^From:[ ][ ]*\(.*\) *<\(.*\)>\$/From: \2 (\1)/ } s/-[Ii]d:/-ID:/ s/^[Ss][Uu][Bb][Jj][Ee][Cc][Tt]:[ ]*$/Subject: (none)/ s/^\([^:]*:\)[ ]*/\1 / /^\$/{i\\ Newsgroups: alt.buddha.short.fat.guy\\ Distribution: world\\ Sender: news@cygnus.com\\ Approved: alt.buddha.short.fat.guy@cygnus.com b e } p n b a :r s/.*//g n /^[ ]/b r b a :e p n b e" dist/sed/testsuite/subwrite.good1100660 765 144 212 6070131561 15741 0ustar lordusersNot some church, and not the state, Not some dark capricious fate. Who YoU are, and when you lose, Comes only from the things YoU choose. dist/sed/testsuite/subwrite.good2100660 765 144 107 6070131561 15745 0ustar lordusersWho YoU are, and when you lose, Comes only from the things YoU choose. dist/sed/testsuite/writeout.good1100660 765 144 220 6070131562 15757 0ustar lordusersFacts are simple and facts are straight Facts are lazy and facts are late Facts all come with points of view Facts don't do what I want them to dist/sed/testsuite/writeout.good2100660 765 144 112 6070131562 15760 0ustar lordusersFacts are simple and facts are straight Facts are lazy and facts are late dist/sed/testsuite/spencer.sh100660 765 144 731 6071053255 15147 0ustar lordusers#! /bin/sh # # this uses the -r command line option to sed (--rx-test) which changes # the regex syntax used to that of POSIX egrep for the purposes of these # tests. # sed=${1-sed} $sed -e "\ /^0:/s,0:\(.*\):\(.*\),echo '\2' | $sed -re '/\1/d' | $sed -e, /^1:/s,1:\(.*\):\(.*\),echo '\2' | $sed -re '/\1/!d' | $sed -e, /^2:/s,2:\(.*\):\(.*\),echo '\2' | $sed -re '/\1/p' | $sed -e '/bad regexp:/d' -e," | \ awk '{print $0 " s\"/.*/Test\ #" NR "\ failed!/\""}' dist/sed/testsuite/readin.in2100660 765 144 4 6070131557 14762 0ustar lordusersMOO dist/sed/testsuite/LOG100660 765 144 166660 6071416402 13613 0ustar lordusersThu Dec 28 too slow on mac-mf; run gprof [15] 38.81 11.52 8.00 9498974+8346392 rx_rexp_equal [15] __ added code to rx_rexp_equal to count n calls and result bias int re_abs = 0; int re_ans = 0; /**/ ++re_abs; re_ans += (ret*2 - 1); in gdb, br exit and check those Breakpoint 1, 0x1359a in exit () (gdb) p re_abs $1 = 3416821 (gdb) p re_ans $2 = -2793859 82% likely to not be equal -- notice: [16] 15.91 8.00 0.00 9847935 rx_bitset_is_equal [16] the test in rx_rexp_equal (a, b) is ret = ( (a->type == b->type) && (a->params.cset_size == b->params.cset_size) && (a->params.intval == b->params.intval) && (a->params.intval2 == b->params.intval2) && rx_bitset_is_equal (a->params.cset_size, a->params.cset, b->params.cset) && rx_rexp_equal (a->params.pair.left, b->params.pair.left) && rx_rexp_equal (a->params.pair.right, b->params.pair.right)); and gprof confirms that rx_bitset_is_equal isn't being called from anyplace else. most calls are useless since most nodes prbly aren't event cset nodes. re-order and prune... before (-pg -g) 20.74user 0.31system 0:22.02elapsed 95%CPU (0avgtext+0avgdata 0maxresident)k after 19.46user 0.15system 0:20.60elapsed 95%CPU (0avgtext+0avgdata 0maxresident)k gprof says: 37.46 7.33 7.33 mcount 17.12 3.35 10.68 4021435 rx_rexp_equal 8.33 1.63 12.31 1114696 rx_bitset_is_equal [13] 56.78 1.38 5.57 130048 rx_hash_store [13] [15] 40.69 3.35 1.63 2275803+1745632 rx_rexp_equal [15] that seems awefully high for such short input ... perhaps rx_hash_store is being called too often. These functions can get faster, but not an order of magnitude faster, which is what we're really after. i think simplified regexps aren't cached reasonably. cache cached_rexp in rexp_node 17.05user 0.06system 0:17.14elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k % time seconds cumsec calls function 35.05 5.98 5.98 mcount 15.89 2.71 8.69 3271085 rx_rexp_equal 7.74 1.32 10.01 919773 rx_bitset_is_equal Thu Dec 28 15:28:26 PST 1995 -O3 comparison to /usr/bin/sed (GNU sed version 2.05) show about an O(10) slowdown. -pg -g: *** Start "time ./sed -f sep.sed < sep.inp > tmp" in ~/vac/rxdispencer/sed/testsuite/ at 15:30:08 0.66user 0.06system 0:00.81elapsed 88%CPU (0avgtext+0avgdata 0maxresident)k % time seconds cumsec calls function 37.31 0.25 0.25 mcount 10.45 0.07 0.32 41465 rx_bitset_hash 7.46 0.05 0.37 43124 rx_rexp_equal 5.97 0.04 0.41 7003 rx_hash_store 4.48 0.03 0.44 33682 unfa_equal 4.48 0.03 0.47 12985 malloc 2.99 0.02 0.49 6899 rx_make_solutions 2.99 0.02 0.51 5727 rx_basic_make_solutions 1.49 0.01 0.52 89829 rx_rexp_hash 1.49 0.01 0.53 9491 nfa_set_enjoin 1.49 0.01 0.54 7791 hash_se_prog 1.49 0.01 0.55 7775 se_list_cmp 1.49 0.01 0.56 7760 rx_free_solutions 1.49 0.01 0.57 7204 rx_next_solution 1.49 0.01 0.58 6992 bzero 1.49 0.01 0.59 6899 canonical_unfa 1.49 0.01 0.60 6899 rx_free_unfa 1.49 0.01 0.61 6899 rx_start_superstate 1.49 0.01 0.62 6899 rx_terminate_system 1.49 0.01 0.63 6899 rx_unfa 1.49 0.01 0.64 6899 rx_unfa_rx 1.49 0.01 0.65 5727 init_basic_once 1.49 0.01 0.66 29 __sbrk 1.49 0.01 0.67 20 rx_superstate sep.sed: -------- # inspired by an autoconf generated configure script. s%/[^/][^/]*$%% sep.inp: -------- miss mary mack mack//mack/ran down/the track track track slashes\aren't%used enough/in/casual-conversation/// possibly sentences would be more attractive if they ended in two slashes// Some call graph entries are revealing: ---------------------------------------- 0.00 0.42 163/163 rx_regexec [5] [2] 99.97 0.00 0.42 163 rx_regmatch [2] 0.02 0.30 5727/5727 rx_basic_make_solutions [9] 0.01 0.07 5727/5727 rx_next_solution [15] 0.00 0.02 5727/5727 rx_basic_free_solutions [23] rx_regmatch is called 163 times -- almost once per character in the input. It calls rx_basic_make_solutions 5727 times because for each starting position, it calls rx_basic_make_solutions for each possible length of match from longest to shortest until a match is found or all matches tried. The fastmap optimization could save a lot here but would lose on a pattern starting with "." or "[a-zA-Z]". In contrast, sed 2.05 doesn't slow down noticably when the command is changed to "s%./[^/][^/]*$%%". rx_basic_make_solutions is called 5727 times more from the call graph: 0.01 0.20 6899/6899 rx_unfa [10] [11] 49.68 0.01 0.20 6899 canonical_unfa [11] there is a caching optimization in rx_unfa for the purpose of avoiding calls to canonical_unfa. apparently it is buggy. why? it depends on the caching of simplified regexps....does that work? Here is a classic profiling trick, to find out: Program exited normally. Breakpoint 1 at 0x109f1: file ../../rx/rxsimp.c, line 67. Breakpoint 2 at 0x10a18: file ../../rx/rxsimp.c, line 72. (gdb) comm 1 Type commands for when breakpoint 1 is hit, one per line. End with a line saying just "end". silent cont end (gdb) comm 2 Type commands for when breakpoint 2 is hit, one per line. End with a line saying just "end". silent cont end (gdb) run [....] (gdb) i br 1 breakpoint keep y 0x000109f1 in rx_simple_rexp at ../../rx/rxsimp.c:67 breakpoint already hit 5802 times silent cont 2 breakpoint keep y 0x00010a18 in rx_simple_rexp at ../../rx/rxsimp.c:72 breakpoint already hit 245 times silent cont which means that caching is working just fine here (245:5802 miss ratio) There's still a puzzle. The expression in "s%/[a-z][a-z]*$%%" comes out to: (gdb) p print_rexp (256, 0, preg->pattern) Node 3 type 1 (r_concat), iv=0( ), iv2=0, len=-1 obs=1 cs=nil Node -1 type 0 (r_cset), iv=0( ), iv2=0, len=1 obs=0 cs=[/] Node 2 type 1 (r_concat), iv=0( ), iv2=0, len=-1 obs=1 cs=nil Node -1 type 0 (r_cset), iv=0( ), iv2=0, len=1 obs=0 cs=[abcdefghijklmnopqrstuvwxyz] Node 1 type 1 (r_concat), iv=0( ), iv2=0, len=-1 obs=1 cs=nil Node -1 type 4 (r_star), iv=0( ), iv2=0, len=-1 obs=0 cs=nil Node -1 type 0 (r_cset), iv=0( ), iv2=0, len=1 obs=0 cs=[abcdefghijklmnopqrstuvwxyz] nil Node 0 type 8 (r_conext), iv=36($), iv2=0, len=0 obs=1 cs=nil even 245 non-cached calls to rx_simple_rexp seems way out of hand. Every node of the tree can be simplified and cached in at most: 8 ; 1 call per node 6 ; 2 recursive calls per concat 1 ; 1 recursive call per star -- 15 ; 15 calls probably fewer as some calls are likely counted twice. So why 245? pop to the main perf problem which was: 0.01 0.20 6899/6899 rx_unfa [10] [11] 49.68 0.01 0.20 6899 canonical_unfa [11] there is a caching optimization in rx_unfa for the purpose of avoiding calls to canonical_unfa. apparently it is buggy. why? What expressions is rx_unfa called for? $40 = (struct rexp_node *) 0x5adc0 $41 = (struct rexp_node *) 0x5adc0 $42 = (struct rexp_node *) 0x5adc0 $43 = (struct rexp_node *) 0x5adc0 [....] Poking around in gdb says that expression's unfa is cached properly but the cache is wrongly cleared here: => rx_basic_free_solutions (solutions); if (answer == rx_yes) .... in rxposix.c. This is only a few "finish" and "next" steps away from where the cache is set. So, it was found just by searching forward in the coarsest steps from a call to rx_unfa. (Watchpoints crash gdb and sometimes linux under linux). set a breakpoint there, ignored it 20 times, and confirmed that the cache is cleared. Before: (gdb) set $exp=solutions->exp (gdb) p $exp->simplified $3519 = (struct rexp_node *) 0x5adc0 (gdb) p $exp->simplified->cr $3520 = (struct rx_cached_rexp *) 0x57000 (gdb) p $exp->refs $3521 = 2 (gdb) p $exp->simplified->refs $3522 = 3 After: (gdb) p $exp->simplified $3526 = (struct rexp_node *) 0x5adc0 (gdb) p $exp->simplified->cr $3523 = (struct rx_cached_rexp *) 0x0 (gdb) p $exp->refs $3525 = 1 (gdb) p $exp->simplified->refs $3524 = 3 so next time, step into it. ugh.... rx_free_unfa (unfa) was simply clearing the cache in what was obviously the wrong place (when the unfa refcount dropped to 0 instead of when it was actually flushed from the cache.) should have checked here a few steps ago, but writing the LOG was a distracting source of amusment. rebuild with fix before: *** Start "time ./sed -f sep.sed < sep.inp > tmp" in ~/vac/rxdispencer/sed/testsuite/ at 15:30:08 0.66user 0.06system 0:00.81elapsed 88%CPU (0avgtext+0avgdata 0maxresident)k after *** Start "time ./sed -f sep.sed < sep.inp > tmp" in /home/lord/vac/rxdispencer/sed/testsuite/ at 16:34:52 0.35user 0.05system 0:00.45elapsed 88%CPU (0avgtext+0avgdata 0maxresident)k 2 or 3 more of those, and we're totally in business. Thu Dec 28 15:28:26 PST 1995 *** Start "time ./sed -f sep.sed < sep.inp > tmp" in /home/lord/vac/rxdispencer/sed/testsuite/ at 16:34:52 0.35user 0.05system 0:00.45elapsed 88%CPU (0avgtext+0avgdata 0maxresident)k profile has changed considerably. good! % time seconds cumsec calls function 38.89 0.14 0.14 mcount 16.67 0.06 0.20 12985 malloc 11.11 0.04 0.24 8085 eclose_node 8.33 0.03 0.27 6137 rx_soluntion_fit_p 5.56 0.02 0.29 12637 free 2.78 0.01 0.30 7760 rx_free_solutions 2.78 0.01 0.31 6899 rx_make_solutions 2.78 0.01 0.32 6899 rx_terminate_system 2.78 0.01 0.33 6899 rx_unfa 2.78 0.01 0.34 6137 rx_fit_p 2.78 0.01 0.35 163 rx_regmatch 2.78 0.01 0.36 20 rx_superstate eclose_node?!?!?! but more pressing is that some tests now segfault. ... -O3 causes the segfaults. hopefully this is a compiler bug. i think i've seen this on linux before. -O has no problem. Stepping back a bit...here is the state of some timing tests. This is -O vs. the mysterious sed 2.05 in /usr/bin. time ./sed -f enable.sed < enable.inp > tmp 0.02user 0.03system 0:00.16elapsed 31%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp enable.good tmp time /usr/bin/sed -f enable.sed < enable.inp > tmp2 0.00user 0.01system 0:00.10elapsed 10%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp enable.good tmp2 time ./sed -f sep.sed < sep.inp > tmp 0.13user 0.02system 0:00.16elapsed 93%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp sep.good tmp time /usr/bin/sed -f sep.sed < sep.inp > tmp2 0.00user 0.02system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp sep.good tmp2 time ./sed -f inclib.sed < inclib.inp > tmp 0.10user 0.01system 0:00.12elapsed 91%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp inclib.good tmp time /usr/bin/sed -f inclib.sed < inclib.inp > tmp2 0.00user 0.02system 0:00.03elapsed 66%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp inclib.good tmp2 time ./sed -f 8bit.sed < 8bit.inp > tmp 0.03user 0.02system 0:00.05elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp 8bit.good tmp time /usr/bin/sed -f 8bit.sed < 8bit.inp > tmp2 0.00user 0.03system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp 8bit.good tmp2 time ./sed -f newjis.sed < newjis.inp > tmp 0.03user 0.01system 0:00.09elapsed 44%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp newjis.good tmp time /usr/bin/sed -f newjis.sed < newjis.inp > tmp2 0.00user 0.03system 0:00.06elapsed 50%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp newjis.good tmp2 time ./sed -n -f cv-vars.sed < cv-vars.inp > tmp 0.02user 0.05system 0:00.08elapsed 87%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp cv-vars.good tmp time /usr/bin/sed -n -f cv-vars.sed < cv-vars.inp > tmp2 0.01user 0.02system 0:00.04elapsed 75%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp cv-vars.good tmp2 time ./sed -f manis.sed < manis.inp > tmp 0.44user 0.03system 0:00.48elapsed 97%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp manis.inp tmp time /usr/bin/sed -f manis.sed < manis.inp > tmp2 0.02user 0.02system 0:00.04elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp manis.inp tmp2 time ./sed -f xabcx.sed < xabcx.inp > tmp 0.00user 0.04system 0:00.04elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp xabcx.good tmp time /usr/bin/sed -f xabcx.sed < xabcx.inp > tmp2 0.01user 0.02system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp xabcx.good tmp2 time ./sed -f dollar.sed < dollar.inp > tmp 0.00user 0.03system 0:00.05elapsed 60%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp dollar.good tmp time /usr/bin/sed -f dollar.sed < dollar.inp > tmp2 0.01user 0.01system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp dollar.good tmp2 time ./sed -f noeol.sed < noeol.inp > tmp 0.01user 0.03system 0:00.04elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp noeol.good tmp time /usr/bin/sed -f noeol.sed < noeol.inp > tmp2 0.00user 0.02system 0:00.03elapsed 66%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp noeol.good tmp2 time ./sed -f sub-num.sed < sub-num.inp > tmp 0.01user 0.03system 0:00.05elapsed 80%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp sub-num.good tmp time /usr/bin/sed -f sub-num.sed < sub-num.inp > tmp2 0.00user 0.03system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp sub-num.good tmp2 time ./sed -f sub-all.sed < sub-num.inp > tmp 0.02user 0.02system 0:00.04elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp sub-all.good tmp time /usr/bin/sed -f sub-all.sed < sub-num.inp > tmp2 0.01user 0.02system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp sub-all.good tmp2 time ./sed -f subwrite.sed < subwrite.inp > tmp1 0.03user 0.02system 0:00.05elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp subwrite.good1 tmp1 cmp subwrite.good2 tmp2 time /usr/bin/sed -f subwrite.sed < subwrite.inp > tmp1 0.01user 0.01system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp subwrite.good1 tmp1 cmp subwrite.good2 tmp2 time ./sed -f head.sed < head.inp > tmp 0.01user 0.00system 0:00.01elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp head.good tmp time /usr/bin/sed -f head.sed < head.inp > tmp2 0.00user 0.02system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp head.good tmp2 time ./sed -n -f middle.sed < head.inp > tmp 0.01user 0.00system 0:00.02elapsed 50%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp middle.good tmp time /usr/bin/sed -n -f middle.sed < head.inp > tmp2 0.00user 0.01system 0:00.02elapsed 50%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp middle.good tmp2 time ./sed --file=newjis.sed < newjis.inp > tmp 0.02user 0.01system 0:00.04elapsed 75%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp newjis.good tmp time /usr/bin/sed --file=newjis.sed < newjis.inp > tmp2 0.01user 0.01system 0:00.03elapsed 66%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp newjis.good tmp2 time ./sed --quiet -f cv-vars.sed < cv-vars.inp > tmp 0.01user 0.04system 0:00.07elapsed 71%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp cv-vars.good tmp time /usr/bin/sed --quiet -f cv-vars.sed < cv-vars.inp > tmp2 0.01user 0.02system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp cv-vars.good tmp2 So far, most often ./sed loses (not always). But we're still winning: time ./sed -f maddingcrowd.sed < maddingcrowd.inp > tmp 2.17user 0.22system 0:02.40elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp maddingcrowd.good # DO NOT time /usr/bin/sed -f maddingcrowd.sed < maddingcrowd.inp > tmp2 # sed 2.05.... t->+inf but also still losing: time /usr/bin/sed -f mac-mf.sed < mac-mf.inp > tmp" in ~/vac/rxdispencer/sed/testsuite/ at 17:19:08 0.29user 0.09system 0:00.39elapsed 97%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps time ./sed -f mac-mf.sed < mac-mf.inp > tmp" in ~/vac/rxdispencer/sed/testsuite/ at 17:19:08 t->inf Fri Dec 29 12:03:28 PST 1995 ran mac-mf.sed on small input for profiling. 32.17 1.66 1.66 mcount 10.27 0.53 2.19 262506 malloc 6.98 0.36 2.55 260030 free 6.20 0.32 2.87 129914 rx_make_solutions 6.01 0.31 3.18 129914 rx_start_superstate 5.43 0.28 3.46 129978 eclose_node 3.68 0.19 3.65 129972 hash_se_prog 3.49 0.18 3.83 4205 rx_regmatch 3.10 0.16 3.99 129916 rx_next_solution 3.10 0.16 4.15 129914 rx_unfa 2.52 0.13 4.28 129728 rx_basic_make_solutions 2.33 0.12 4.40 129914 rx_free_unfa 1.94 0.10 4.50 132334 bzero 1.74 0.09 4.59 161119 rx_free_rexp 1.36 0.07 4.66 161229 rx_save_rexp 1.16 0.06 4.72 130096 rx_free_solutions 1.16 0.06 4.78 129992 nfa_set_enjoin 1.16 0.06 4.84 129914 rx_terminate_system 0.97 0.05 4.89 129958 rx_state_possible_futures 0.97 0.05 4.94 129728 rx_basic_free_solutions 0.78 0.04 4.98 129914 rx_unfa_rx 0.58 0.03 5.01 129914 rx_init_system 0.58 0.03 5.04 129914 se_list_cmp 0.58 0.03 5.07 31241 rx_simple_rexp 0.39 0.02 5.09 698 inverse_translation 0.19 0.01 5.10 23082 rx_fit_p 0.19 0.01 5.11 4820 inchar 0.19 0.01 5.12 364 rx_hash_store 0.19 0.01 5.13 125 rx_bitset_difference 0.19 0.01 5.14 94 __sbrk 0.19 0.01 5.15 1 __isatty 0.19 0.01 5.16 1 compile_program malloc is top, called many many times -- approx. half the calls to that must be from rx_make_solutions. Most of those are from rx_basic_make_solutions. All of those are from rx_regmatch, which is called far fewer times (approx 1:31). All of those (in this case) are from rx_regexec, called only 138 times. fastmapping can prune many calls from regexec to regmatch so add the fastmapping optimization copying/fixing code from an older rx. before fastmapping: time ./sed -f mac-mf.sed < ,a 5.14user 0.14system 0:08.13elapsed 64%CPU (0avgtext+0avgdata 0maxresident)k after fastmapping: ./sed -f mac-mf.sed < ,a 0.85user 0.09system 0:01.08elapsed 87%CPU (0avgtext+0avgdata 0maxresident)k rerun -O timing/bug tests some bug in the spencer test.... time ./sed -f enable.sed < enable.inp > tmp 0.00user 0.02system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp enable.good tmp time /usr/bin/sed -f enable.sed < enable.inp > tmp2 0.00user 0.03system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp enable.good tmp2 time ./sed -f sep.sed < sep.inp > tmp 0.02user 0.03system 0:00.05elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp sep.good tmp time /usr/bin/sed -f sep.sed < sep.inp > tmp2 0.01user 0.02system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp sep.good tmp2 time ./sed -f inclib.sed < inclib.inp > tmp 0.03user 0.00system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp inclib.good tmp time /usr/bin/sed -f inclib.sed < inclib.inp > tmp2 0.01user 0.02system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp inclib.good tmp2 time ./sed -f 8bit.sed < 8bit.inp > tmp 0.00user 0.02system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp 8bit.good tmp time /usr/bin/sed -f 8bit.sed < 8bit.inp > tmp2 0.00user 0.03system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp 8bit.good tmp2 time ./sed -f newjis.sed < newjis.inp > tmp 0.00user 0.03system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp newjis.good tmp time /usr/bin/sed -f newjis.sed < newjis.inp > tmp2 0.00user 0.02system 0:00.03elapsed 66%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp newjis.good tmp2 time ./sed -n -f cv-vars.sed < cv-vars.inp > tmp 0.01user 0.05system 0:00.06elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp cv-vars.good tmp time /usr/bin/sed -n -f cv-vars.sed < cv-vars.inp > tmp2 0.01user 0.02system 0:00.04elapsed 75%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp cv-vars.good tmp2 time ./sed -f manis.sed < manis.inp > tmp 0.10user 0.03system 0:00.13elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp manis.inp tmp time /usr/bin/sed -f manis.sed < manis.inp > tmp2 0.01user 0.03system 0:00.04elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp manis.inp tmp2 time ./sed -f xabcx.sed < xabcx.inp > tmp 0.01user 0.02system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp xabcx.good tmp time /usr/bin/sed -f xabcx.sed < xabcx.inp > tmp2 0.01user 0.01system 0:00.03elapsed 66%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp xabcx.good tmp2 time ./sed -f dollar.sed < dollar.inp > tmp 0.01user 0.01system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp dollar.good tmp time /usr/bin/sed -f dollar.sed < dollar.inp > tmp2 0.00user 0.02system 0:00.03elapsed 66%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp dollar.good tmp2 time ./sed -f noeol.sed < noeol.inp > tmp 0.01user 0.02system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp noeol.good tmp time /usr/bin/sed -f noeol.sed < noeol.inp > tmp2 0.01user 0.01system 0:00.03elapsed 66%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp noeol.good tmp2 time ./sed -f sub-num.sed < sub-num.inp > tmp 0.01user 0.01system 0:00.03elapsed 66%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp sub-num.good tmp time /usr/bin/sed -f sub-num.sed < sub-num.inp > tmp2 0.01user 0.01system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp sub-num.good tmp2 time ./sed -f sub-all.sed < sub-num.inp > tmp 0.01user 0.02system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp sub-all.good tmp time /usr/bin/sed -f sub-all.sed < sub-num.inp > tmp2 0.00user 0.02system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp sub-all.good tmp2 time ./sed -f subwrite.sed < subwrite.inp > tmp1 0.02user 0.00system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp subwrite.good1 tmp1 cmp subwrite.good2 tmp2 time /usr/bin/sed -f subwrite.sed < subwrite.inp > tmp1 0.00user 0.02system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp subwrite.good1 tmp1 cmp subwrite.good2 tmp2 time ./sed -f head.sed < head.inp > tmp 0.01user 0.01system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp head.good tmp time /usr/bin/sed -f head.sed < head.inp > tmp2 0.00user 0.02system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp head.good tmp2 time ./sed -n -f middle.sed < head.inp > tmp 0.00user 0.01system 0:00.01elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp middle.good tmp time /usr/bin/sed -n -f middle.sed < head.inp > tmp2 0.00user 0.02system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp middle.good tmp2 time ./sed -f maddingcrowd.sed < maddingcrowd.inp > tmp 2.13user 0.26system 0:02.76elapsed 86%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp maddingcrowd.good time /usr/bin/sed -f maddingcrowd.sed < maddingcrowd.inp > tmp2 t->inf time ./sed -f mac-mf.sed < mac-mf.inp > tmp 21.74user 0.16system 0:21.90elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp mac-mf.good time /usr/bin/sed -f mac-mf.sed < mac-mf.inp > tmp2 0.28user 0.11system 0:00.39elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp2 mac-mf.good time ./sed -f khadafy.sed < khadafy.inp > tmp 0.02user 0.01system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp khadafy.inp time /usr/bin/sed -f khadafy.sed < khadafy.inp > tmp2 0.01user 0.03system 0:00.04elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp2 khadafy.inp time ./sed -f linecnt.sed < linecnt.inp > tmp 0.00user 0.01system 0:00.01elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp linecnt.good time /usr/bin/sed -f linecnt.sed < linecnt.inp > tmp2 0.01user 0.01system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp2 linecnt.good Generally close. One very winning case (maddingcrowd). One pretty much losing case (but not as bad as before) mac-mf. All tests pass except spencer. All tests reasonable performance except "distrib" and "mac-mf". Spencer: was typo in test suite but that fixed, still bugs: Test #37 failed! Test #62 failed! #37 -- test suite was wrong -- sed, right -- pattern bogus. lines 2945-2947 of 2003.2-1992 #62 -- test suite wrong again -- sed, right -- pattern not bogus lines 3058-3060 and 3066-3077 of 2003.2-1992 spencer test passes with those fixes. Fri Dec 29 13:14:02 PST 1995 last development before release issues: All tests reasonable performance except "distrib" and "mac-mf". time ./sed -f mac-mf.sed < mac-mf.inp > tmp 22.10user 0.22system 0:22.52elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp mac-mf.good time /usr/bin/sed -f mac-mf.sed < mac-mf.inp > tmp2 0.31user 0.09system 0:00.40elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp2 mac-mf.good % time seconds cumsec calls function 34.18 19.43 19.43 mcount 6.03 3.43 22.86 2545726 free 5.96 3.39 26.25 2561762 malloc 5.44 3.09 29.34 1324635 eclose_node 3.83 2.18 31.52 41025 rx_regmatch 3.61 2.05 33.57 1276850 rx_next_solution 3.36 1.91 35.48 1276625 rx_start_superstate 3.22 1.83 37.31 1276625 rx_make_solutions 3.17 1.80 39.11 1276625 rx_free_unfa 2.94 1.67 40.78 1217560 rx_fit_p 2.41 1.37 42.15 1265155 rx_basic_make_solutions 2.27 1.29 43.44 1276625 rx_unfa 2.20 1.25 44.69 1223796 rx_simple_rexp 2.13 1.21 45.90 1280457 bzero 2.01 1.14 47.04 1287172 rx_free_solutions 1.55 0.88 47.92 1217560 rx_soluntion_fit_p 1.51 0.86 48.78 1278517 rx_state_possible_futures 1.44 0.82 49.60 1312946 hash_se_prog 1.41 0.80 50.40 1276625 rx_unfa_rx 1.39 0.79 51.19 1217560 rx_str_vmfn 1.18 0.67 51.86 2500305 rx_free_rexp 1.13 0.64 52.50 2500621 rx_save_rexp 1.11 0.63 53.13 1381011 nfa_set_enjoin 0.83 0.47 53.60 1312343 se_list_cmp 0.83 0.47 54.07 12948 rx_hash_store 0.77 0.44 54.51 1265155 init_basic_once which means, really, that no function dominates and again, avoiding the many many calls to rx_regmatch is key. looking at the script, there are many anchored patterns and i recall an old rx optimization "don't waste time trying an anchored match in the middle of a pattern." Let's put that back in. -pg before is_anchored time ./sed -f mac-mf.sed < mac-mf.inp > tmp 22.10user 0.22system 0:22.52elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp mac-mf.good -pg after time ./sed -f mac-mf.sed < mac-mf.inp > tmp 8.22user 0.31system 0:15.97elapsed 53%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp mac-mf.good also: 2.60 0.22 5.76 141219 rx_make_solutions meaning that we cut down about O(10) on calls to rx_make_solutions. *** Start "make" in /home/lord/vac/rxdispencer/sed/testsuite/ at 13:49:53 time ./sed -f enable.sed < enable.inp > tmp 0.01user 0.04system 0:01.75elapsed 2%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp enable.good tmp time /usr/bin/sed -f enable.sed < enable.inp > tmp2 0.01user 0.03system 0:00.12elapsed 33%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp enable.good tmp2 time ./sed -f sep.sed < sep.inp > tmp 0.02user 0.01system 0:00.25elapsed 12%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp sep.good tmp time /usr/bin/sed -f sep.sed < sep.inp > tmp2 0.00user 0.04system 0:00.16elapsed 25%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp sep.good tmp2 time ./sed -f inclib.sed < inclib.inp > tmp 0.02user 0.01system 0:00.05elapsed 60%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp inclib.good tmp time /usr/bin/sed -f inclib.sed < inclib.inp > tmp2 0.00user 0.02system 0:00.03elapsed 66%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp inclib.good tmp2 time ./sed -f 8bit.sed < 8bit.inp > tmp 0.00user 0.01system 0:00.03elapsed 33%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp 8bit.good tmp time /usr/bin/sed -f 8bit.sed < 8bit.inp > tmp2 0.00user 0.03system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp 8bit.good tmp2 time ./sed -f newjis.sed < newjis.inp > tmp 0.01user 0.01system 0:00.04elapsed 50%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp newjis.good tmp time /usr/bin/sed -f newjis.sed < newjis.inp > tmp2 0.00user 0.04system 0:00.10elapsed 40%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp newjis.good tmp2 time ./sed -n -f cv-vars.sed < cv-vars.inp > tmp 0.02user 0.03system 0:00.40elapsed 12%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp cv-vars.good tmp time /usr/bin/sed -n -f cv-vars.sed < cv-vars.inp > tmp2 0.00user 0.04system 0:00.04elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp cv-vars.good tmp2 time ./sed -f manis.sed < manis.inp > tmp 0.07user 0.04system 0:00.16elapsed 68%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp manis.inp tmp time /usr/bin/sed -f manis.sed < manis.inp > tmp2 0.00user 0.04system 0:00.16elapsed 25%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp manis.inp tmp2 time ./sed -f xabcx.sed < xabcx.inp > tmp 0.00user 0.02system 0:00.05elapsed 40%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp xabcx.good tmp time /usr/bin/sed -f xabcx.sed < xabcx.inp > tmp2 0.00user 0.03system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp xabcx.good tmp2 time ./sed -f dollar.sed < dollar.inp > tmp 0.00user 0.02system 0:00.04elapsed 50%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp dollar.good tmp time /usr/bin/sed -f dollar.sed < dollar.inp > tmp2 0.00user 0.03system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp dollar.good tmp2 time ./sed -f noeol.sed < noeol.inp > tmp 0.03user 0.00system 0:00.05elapsed 60%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp noeol.good tmp time /usr/bin/sed -f noeol.sed < noeol.inp > tmp2 0.01user 0.01system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp noeol.good tmp2 time ./sed -f sub-num.sed < sub-num.inp > tmp 0.01user 0.01system 0:00.04elapsed 50%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp sub-num.good tmp time /usr/bin/sed -f sub-num.sed < sub-num.inp > tmp2 0.00user 0.03system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp sub-num.good tmp2 time ./sed -f sub-all.sed < sub-num.inp > tmp 0.01user 0.02system 0:00.08elapsed 37%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp sub-all.good tmp time /usr/bin/sed -f sub-all.sed < sub-num.inp > tmp2 0.00user 0.02system 0:00.03elapsed 66%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp sub-all.good tmp2 time ./sed -f subwrite.sed < subwrite.inp > tmp1 0.01user 0.01system 0:00.04elapsed 50%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp subwrite.good1 tmp1 cmp subwrite.good2 tmp2 time /usr/bin/sed -f subwrite.sed < subwrite.inp > tmp1 0.01user 0.01system 0:00.03elapsed 66%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp subwrite.good1 tmp1 cmp subwrite.good2 tmp2 time ./sed -f head.sed < head.inp > tmp 0.00user 0.02system 0:00.04elapsed 50%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp head.good tmp time /usr/bin/sed -f head.sed < head.inp > tmp2 0.00user 0.02system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp head.good tmp2 time ./sed -n -f middle.sed < head.inp > tmp 0.00user 0.01system 0:00.01elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp middle.good tmp time /usr/bin/sed -n -f middle.sed < head.inp > tmp2 0.00user 0.01system 0:00.02elapsed 50%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp middle.good tmp2 time ./sed -f maddingcrowd.sed < maddingcrowd.inp > tmp 2.18user 0.36system 0:10.25elapsed 24%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp maddingcrowd.good echo DO NOT time /usr/bin/sed -f maddingcrowd.sed < maddingcrowd.inp > tmp2 echo sed 2.05.... t->+inf time ./sed -f mac-mf.sed < mac-mf.inp > tmp 2.52user 0.14system 0:02.92elapsed 91%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp mac-mf.good time /usr/bin/sed -f mac-mf.sed < mac-mf.inp > tmp2 0.26user 0.14system 0:00.44elapsed 90%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp2 mac-mf.good time ./sed -f khadafy.sed < khadafy.inp > tmp 0.02user 0.00system 0:00.06elapsed 33%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp khadafy.inp time /usr/bin/sed -f khadafy.sed < khadafy.inp > tmp2 0.01user 0.03system 0:00.04elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp2 khadafy.inp sh spencer.sh ./sed < spencer.inp > spencer.tmp sh spencer.tmp time ./sed -f linecnt.sed < linecnt.inp > tmp 0.00user 0.01system 0:00.05elapsed 20%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp linecnt.good time /usr/bin/sed -f linecnt.sed < linecnt.inp > tmp2 0.00user 0.01system 0:00.02elapsed 50%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp2 linecnt.good sh distrib.sh ./sed < distrib.inp > tmp cmp tmp distrib.good time ./sed -f writeout.sed < writeout.inp >tmp1 0.01user 0.03system 0:00.05elapsed 80%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp1 writeout.good1 cmp tmp2 writeout.good2 time ./sed -f readin.sed < readin.inp >tmp 0.01user 0.02system 0:00.05elapsed 60%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp readin.good time /usr/bin/sed -f readin.sed < readin.inp >tmp2 0.00user 0.03system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp2 readin.good ./sed --help > tmp1 ./sed -h > tmp2 ./sed > tmp3 2>&1 make: [help] Error 4 (ignored) cmp tmp1 tmp2 cmp tmp2 tmp3 cmp tmp1 help.good tmp1 help.good differ: char 9, line 1 make: [help] Error 1 (ignored) ./sed --version > tmp1 2>&1 ./sed -V > tmp2 2>&1 cmp tmp1 tmp2 cmp tmp1 version.good time ./sed --file=newjis.sed < newjis.inp > tmp 0.01user 0.01system 0:00.04elapsed 50%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp newjis.good tmp time /usr/bin/sed --file=newjis.sed < newjis.inp > tmp2 0.00user 0.03system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp newjis.good tmp2 time ./sed --quiet -f cv-vars.sed < cv-vars.inp > tmp 0.03user 0.03system 0:00.06elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp cv-vars.good tmp time /usr/bin/sed --quiet -f cv-vars.sed < cv-vars.inp > tmp2 0.01user 0.02system 0:00.04elapsed 75%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp cv-vars.good tmp2 rm -f tmp tmp1 tmp2 tmp3 core junk spencer.tmp time ./sed -f 8to7.sed < 8bit.inp > tmp 0.00user 0.01system 0:00.03elapsed 33%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp 8to7.good tmp 8to7.good differ: char 1, line 1 make: [8to7] Error 1 (ignored) time /usr/bin/sed -f 8to7.sed < 8bit.inp > tmp2 0.01user 0.01system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+0minor)pagefaults 0swaps cmp tmp2 8to7.good tmp2 8to7.good differ: char 1, line 1 make: [8to7] Error 1 (ignored) one (known) failure for not having an obscure posix feature in sed. everything else works. perf very comperable with 2.05 on most of these (small) tests. rxdispencer totally wins on the maddingcrowd; 2.05 uses vast memory and takes "forever" rxdispencer annoyingly 10x slower on mac-mf. prbly acceptable for a release but.... Fri Dec 29 13:54:34 PST 1995 looking at mac-mf calls 0.00 0.00 8.36 470 canonical_unfa that suggests unfa caching is doing fine. 0.00 0.00 8.36 847 rx_really_free_superstate aha... we are overrunning the superstate cache. The top functions look suspicious: 5.86 0.49 3.26 299341 malloc 5.38 0.45 3.71 4376 rx_hash_store 4.78 0.40 4.11 147917 eclose_node once again ... that is a lot of calls to eclose_node 3.35 0.28 5.08 865 rx_superstate That is *really* weird. Almost as many superstates were really freed as were constructed. br in exit and look at the "default_cache" and process size: TSIZ DSIZ RSS 9767 pp4 T 0:09 168 120 2608 1728 xx 11.6 /home/lord/vac/rxdispencer/sed/testsuite/sed -f mac-mf.sed lru_superstate = 0x18c000, semifree_superstate = 0x0, empty_superset = 0x79d80, superstates = 239, semifree_superstates = 0, hits = 4, misses = 135, bytes_allowed = 1048576, bytes_used = 1052192, the hit/miss ratio is consistant with overrunning the cache. but why are there *no* semifree_superstates? br in semifree_superstate comm silent c end hit 584 times rx_really_free_superstate was (in a convoluted way) semifreeing superstates at about the same rate they were being really freed. unlikely to fix the perf. problem on mac-mf...but that is a bug the solution is to semifree at twice the rate of really_free. fix that and retry test. before: time ./sed -f mac-mf.sed < mac-mf.inp > tmp 8.22user 0.31system 0:15.97elapsed 53%CPU (0avgtext+0avgdata 0maxresident)k after: time ./sed -f mac-mf.sed < mac-mf.inp > tmp 8.37user 0.37system 0:10.84elapsed 80%CPU (0avgtext+0avgdata 0maxresident)k and cache stats: superstates = 238, semifree_superstates = 166, hits = 5, misses = 124, bytes_allowed = 1048576, bytes_used = 1047452, all of which is good news (except the hit/miss which still suggests the cache is being over-run.) re-compile with 4x superstate cache size time ./sed -f mac-mf.sed < mac-mf.inp > tmp 8.10user 0.60system 0:15.41elapsed 56%CPU (0avgtext+0avgdata 0maxresident)k hardly any change re-compile with 4x unfa cache size. slow down! time ./sed -f mac-mf.sed < mac-mf.inp > tmp 8.16user 0.49system 0:10.63elapsed 81%CPU (0avgtext+0avgdata 0maxresident)k odd cache data, too: superstates = 596, semifree_superstates = 0, hits = 17, misses = 596, bytes_allowed = 4194304, bytes_used = 2608228, far from overflowing the superstate cache p rx_basic_uv, the unfa cache: delay = 2048, delayed = 100, so that isn't overflowing either. exploring that structure more shows that the hash table is totally screwed -- everything hashes into the 0 bucket. out of 100 regexps to rexp_hash...got these results...not the greatest hash function. 0 1 4 5 61032 61033 61036 61037 use more fields: 0 2 6 60954 4294906336 4294967290 hmm. cribbed some hashing clue from ouster on grounds that hashing csets and rexp_nodes is almost like hashing strings. much nicer: 100 distinguishable calls to rexp_hash (distinct by addr of rexp_node) giving 100 distinct hash values. great. 0 (struct rexp_node *) 0x0 17006040 (struct rexp_node *) 0x74500 19131804 (struct rexp_node *) 0x5e0c0 21257568 (struct rexp_node *) 0x62e00 21789009 (struct rexp_node *) 0x62c40 22320450 (struct rexp_node *) 0x6b3c0 24446214 (struct rexp_node *) 0x62a40 24977655 (struct rexp_node *) 0x65cc0 30823506 (struct rexp_node *) 0x5e7c0 32417829 (struct rexp_node *) 0x65c80 34543593 (struct rexp_node *) 0x70bc0 35606475 (struct rexp_node *) 0x6e140 35846082 (struct rexp_node *) 0x74300 37200798 (struct rexp_node *) 0x70cc0 37732239 (struct rexp_node *) 0x70b40 40389444 (struct rexp_node *) 0x70c40 44109531 (struct rexp_node *) 0x70ac0 48892500 (struct rexp_node *) 0x6b640 65367171 (struct rexp_node *) 0x6e0c0 66430053 (struct rexp_node *) 0x70fc0 86351200 (struct rexp_node *) 0x6b900 110001468 (struct rexp_node *) 0x5eac0 225255629 (struct rexp_node *) 0x6b480 231371560 (struct rexp_node *) 0x70800 312638818 (struct rexp_node *) 0x5ee00 390934594 (struct rexp_node *) 0x70e40 517283529 (struct rexp_node *) 0x746c0 549067790 (struct rexp_node *) 0x1839c0 572893326 (struct rexp_node *) 0x6b080 601001026 (struct rexp_node *) 0x27f580 606081253 (struct rexp_node *) 0x65dc0 607263981 (struct rexp_node *) 0x69680 621718215 (struct rexp_node *) 0x27f6c0 650824273 (struct rexp_node *) 0x6ef40 826472566 (struct rexp_node *) 0x6b2c0 868635774 (struct rexp_node *) 0x65f00 876091122 (struct rexp_node *) 0x206140 888472439 (struct rexp_node *) 0x6e400 929543565 (struct rexp_node *) 0x1e0b80 931845394 (struct rexp_node *) 0x206180 1058747076 (struct rexp_node *) 0x183900 1207849365 (struct rexp_node *) 0x67880 1278079982 (struct rexp_node *) 0x69f00 1295802040 (struct rexp_node *) 0x27f480 1310711465 (struct rexp_node *) 0x1e0b40 1352930633 (struct rexp_node *) 0x76240 1353620468 (struct rexp_node *) 0x27f700 1356456513 (struct rexp_node *) 0x79ac0 1379844369 (struct rexp_node *) 0x62400 1402493327 (struct rexp_node *) 0x6bfc0 1601102720 (struct rexp_node *) 0x5ef00 1772669690 (struct rexp_node *) 0x183940 1900599955 (struct rexp_node *) 0x27f440 1968417710 (struct rexp_node *) 0xac140 2108985499 (struct rexp_node *) 0x79b00 2137694068 (struct rexp_node *) 0x651c0 2204601669 (struct rexp_node *) 0x27f600 2218807864 (struct rexp_node *) 0x799c0 2223118861 (struct rexp_node *) 0x74780 2236502798 (struct rexp_node *) 0x79840 2278246773 (struct rexp_node *) 0x27f640 2290967666 (struct rexp_node *) 0x73980 2298725010 (struct rexp_node *) 0x1be500 2303527273 (struct rexp_node *) 0xaca80 2381124070 (struct rexp_node *) 0x27f680 2418247494 (struct rexp_node *) 0x62200 2437460954 (struct rexp_node *) 0xac000 2439348845 (struct rexp_node *) 0x79140 2455486158 (struct rexp_node *) 0x695c0 2662071862 (struct rexp_node *) 0x183980 2745345239 (struct rexp_node *) 0x74040 2832160954 (struct rexp_node *) 0x70400 2900057973 (struct rexp_node *) 0x76940 2918628823 (struct rexp_node *) 0x79d00 2956546377 (struct rexp_node *) 0x1e0b00 3017127356 (struct rexp_node *) 0x27f400 3031395249 (struct rexp_node *) 0xac1c0 3111540809 (struct rexp_node *) 0xac180 3225480365 (struct rexp_node *) 0x27f540 3277401564 (struct rexp_node *) 0xe5b40 3409840041 (struct rexp_node *) 0x27f740 3441564462 (struct rexp_node *) 0x27f3c0 3456946400 (struct rexp_node *) 0x65940 3496700879 (struct rexp_node *) 0x27f4c0 3531401660 (struct rexp_node *) 0x62cc0 3546066190 (struct rexp_node *) 0x73f80 3547538510 (struct rexp_node *) 0x790c0 3548169742 (struct rexp_node *) 0x1e0d40 3589031006 (struct rexp_node *) 0x79dc0 3782143090 (struct rexp_node *) 0x79040 3824014835 (struct rexp_node *) 0x693c0 3837791013 (struct rexp_node *) 0x6ed80 4036337975 (struct rexp_node *) 0x27f5c0 4063573689 (struct rexp_node *) 0x5e580 4107576897 (struct rexp_node *) 0x2061c0 4140954487 (struct rexp_node *) 0x73080 4168978309 (struct rexp_node *) 0x79b40 4185396955 (struct rexp_node *) 0x27f500 4200919815 (struct rexp_node *) 0x5e640 4247337862 (struct rexp_node *) 0xaca00 but they still all go into the same damn bucket. oops...H2B (hash->bucket) in rxhash.c was bogus do to a recent change away from %13. fixed for 16 buckets --- still considering going back to %13 anyway...much better...here is rx_basic_uv...notice the spreading of elements across all buckets in the hash table { delay = 2048, delayed = 100, table = { parent = 0x0, refs = 16, nested_p = 65535, children = {0x59180, 0xb0c80, 0x8e380, 0x14dd80, 0xd6380, 0x8e880, 0x59d00, 0x217d80, 0x59900, 0x59600, 0xd6000, 0x1f8380, 0xd6580, 0x59b80, 0x59e80, 0x59f00} }, free_queue = 0x103960 } and overall time: before (-pg): time ./sed -f mac-mf.sed < mac-mf.inp > tmp 8.37user 0.37system 0:10.84elapsed 80%CPU (0avgtext+0avgdata 0maxresident)k after: 6.94user 0.59system 0:13.71elapsed 54%CPU (0avgtext+0avgdata 0maxresident)k Here is the superstate cache stats superstates = 596, semifree_superstates = 0, hits = 17, misses = 596, bytes_allowed = 4194304, bytes_used = 2611648, this and the unfa cache make it clear that caching is not an issue now. gprof says: 33.33 2.50 2.50 mcount 9.47 0.71 3.21 299018 malloc 5.47 0.41 3.62 613 rx_superstate 4.53 0.34 3.96 147465 eclose_node 4.13 0.31 4.27 141219 rx_unfa 3.73 0.28 4.55 280370 free 3.33 0.25 4.80 5042 rx_regmatch 3.20 0.24 5.04 141219 rx_free_unfa 3.20 0.24 5.28 141219 rx_start_superstate 2.13 0.16 5.44 145590 bzero 2.13 0.16 5.60 141219 rx_make_solutions 2.00 0.15 5.75 82154 rx_fit_p 1.87 0.14 5.89 82154 rx_str_vmfn 1.87 0.14 6.03 4039 rx_hash_store 1.47 0.11 6.14 141444 rx_next_solution 1.47 0.11 6.25 82154 rx_soluntion_fit_p 1.33 0.10 6.35 146135 hash_se_prog 1.33 0.10 6.45 143200 rx_free_solutions 1.33 0.10 6.55 9006 rx_regexec 1.20 0.09 6.64 142747 rx_state_possible_futures 1.20 0.09 6.73 138315 rx_basic_make_solutions Ripe for the picken'.... every one of these: 2.13 0.16 5.60 141219 rx_make_solutions which calls here: 3.20 0.24 5.28 141219 rx_start_superstate which calls here 1.20 0.09 6.64 142747 rx_state_possible_futures which calls here 4.53 0.34 3.96 147465 eclose_node the starting superstate should be better cached than that. oops...this is another optimization not carried over from the old rx. looking further....grr! struct rx_possible_future * rx_state_possible_futures (rx, n) struct rx * rx; struct rx_nfa_state * n; #endif { if (n->futures_computed) return n->futures; else { struct eclose_frame frame; frame.prog_backwards = 0; if (!eclose_node (rx, n, n, &frame)) return 0; else { struct rx_possible_future * p; return n->futures; } } } that's the only occurence of "futures_computed" ...oops... fixing that... before: 6.94user 0.59system 0:13.71elapsed 54%CPU (0avgtext+0avgdata 0maxresident)k after: 5.75user 0.32system 0:07.17elapsed 84%CPU (0avgtext+0avgdata 0maxresident)k profiling % time seconds cumsec calls function 34.00 2.05 2.05 mcount 7.63 0.46 2.51 299018 malloc 5.31 0.32 2.83 141219 rx_start_superstate 4.64 0.28 3.11 280370 free 4.48 0.27 3.38 141219 rx_unfa 4.15 0.25 3.63 141219 rx_make_solutions 3.81 0.23 3.86 613 rx_superstate 3.65 0.22 4.08 5042 rx_regmatch 3.48 0.21 4.29 141219 rx_free_unfa 2.49 0.15 4.44 82154 rx_fit_p 2.16 0.13 4.57 142747 rx_state_possible_futures 1.82 0.11 4.68 145590 bzero 1.82 0.11 4.79 138315 rx_basic_make_solutions 1.82 0.11 4.90 9006 regnexec 1.66 0.10 5.00 143200 rx_free_solutions 1.49 0.09 5.09 130333 supersetcmp 1.49 0.09 5.18 82154 rx_soluntion_fit_p 1.49 0.09 5.27 4039 rx_hash_store 1.33 0.08 5.35 745 __sbrk 1.16 0.07 5.42 141444 rx_next_solution 1.00 0.06 5.48 229493 rx_free_rexp so good, eclose_node is off the map. -O time: before (last known): time ./sed -f mac-mf.sed < mac-mf.inp > tmp 21.74user 0.16system 0:21.90elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k after: 2.34user 0.36system 0:02.72elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k but 2.05 is still: time /usr/bin/sed -f mac-mf.sed < mac-mf.inp > tmp2 0.28user 0.11system 0:00.39elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k code reordering in rx_start_superstate should eliminate calls to rx_state_possible_futures after (-O): 2.28user 0.34system 0:02.64elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k hehe -O2: 2.32user 0.31system 0:02.64elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k Fri Dec 29 16:43:44 PST 1995 getting harder to speed up now. one observation is that malloc is the heaviest function, and almost all the calls there come from rx_basic_make_solutions and rx_make_solutions. put in a (thread-losing) malloc/free buffer for the two spots where malloc/free are called often. -pg mac-mf before: 5.75user 0.32system 0:07.17elapsed 84%CPU (0avgtext+0avgdata 0maxresident)k after (!): 3.98user 0.45system 0:04.74elapsed 93%CPU (0avgtext+0avgdata 0maxresident)k (after profiling) inlined the common & fastest path through rx_start_superstate into rx_make_solutions. after: 3.88user 0.41system 0:04.34elapsed 98%CPU (0avgtext+0avgdata 0maxresident)k back that one out since it is a tiny win and maintainence headache. -O time ./sed -f enable.sed < enable.inp > tmp 0.00user 0.03system 0:00.05elapsed 60%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed -f enable.sed < enable.inp > tmp2 0.01user 0.02system 0:00.05elapsed 60%CPU (0avgtext+0avgdata 0maxresident)k time ./sed -f sep.sed < sep.inp > tmp 0.02user 0.03system 0:00.06elapsed 83%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed -f sep.sed < sep.inp > tmp2 0.00user 0.02system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k time ./sed -f inclib.sed < inclib.inp > tmp 0.00user 0.03system 0:00.05elapsed 60%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed -f inclib.sed < inclib.inp > tmp2 0.00user 0.02system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k time ./sed -f 8bit.sed < 8bit.inp > tmp 0.00user 0.02system 0:00.05elapsed 40%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed -f 8bit.sed < 8bit.inp > tmp2 0.00user 0.02system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k time ./sed -f newjis.sed < newjis.inp > tmp 0.01user 0.01system 0:00.04elapsed 50%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed -f newjis.sed < newjis.inp > tmp2 0.01user 0.01system 0:00.03elapsed 66%CPU (0avgtext+0avgdata 0maxresident)k time ./sed -n -f cv-vars.sed < cv-vars.inp > tmp 0.01user 0.05system 0:00.08elapsed 75%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed -n -f cv-vars.sed < cv-vars.inp > tmp2 0.01user 0.02system 0:00.04elapsed 75%CPU (0avgtext+0avgdata 0maxresident)k time ./sed -f manis.sed < manis.inp > tmp 0.07user 0.02system 0:00.09elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed -f manis.sed < manis.inp > tmp2 0.02user 0.01system 0:00.04elapsed 75%CPU (0avgtext+0avgdata 0maxresident)k time ./sed -f xabcx.sed < xabcx.inp > tmp 0.00user 0.03system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed -f xabcx.sed < xabcx.inp > tmp2 0.01user 0.03system 0:00.12elapsed 33%CPU (0avgtext+0avgdata 0maxresident)k time ./sed -f dollar.sed < dollar.inp > tmp 0.01user 0.02system 0:00.04elapsed 75%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed -f dollar.sed < dollar.inp > tmp2 0.00user 0.02system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k time ./sed -f noeol.sed < noeol.inp > tmp 0.01user 0.01system 0:00.04elapsed 50%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed -f noeol.sed < noeol.inp > tmp2 0.00user 0.02system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k time ./sed -f sub-num.sed < sub-num.inp > tmp 0.00user 0.03system 0:00.04elapsed 75%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed -f sub-num.sed < sub-num.inp > tmp2 0.00user 0.03system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k time ./sed -f sub-all.sed < sub-num.inp > tmp 0.01user 0.00system 0:00.03elapsed 33%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed -f sub-all.sed < sub-num.inp > tmp2 0.01user 0.01system 0:00.03elapsed 66%CPU (0avgtext+0avgdata 0maxresident)k time ./sed -f subwrite.sed < subwrite.inp > tmp1 0.00user 0.02system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed -f subwrite.sed < subwrite.inp > tmp1 0.00user 0.02system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k time ./sed -f head.sed < head.inp > tmp 0.00user 0.01system 0:00.03elapsed 33%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed -f head.sed < head.inp > tmp2 0.00user 0.01system 0:00.02elapsed 50%CPU (0avgtext+0avgdata 0maxresident)k time ./sed -n -f middle.sed < head.inp > tmp 0.00user 0.02system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed -n -f middle.sed < head.inp > tmp2 0.00user 0.02system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k time ./sed -f maddingcrowd.sed < maddingcrowd.inp > tmp 2.64user 0.70system 0:05.88elapsed 56%CPU (0avgtext+0avgdata 0maxresident)k echo DO NOT time /usr/bin/sed -f maddingcrowd.sed < maddingcrowd.inp > tmp2 echo sed 2.05.... t->+inf time ./sed -f mac-mf.sed < mac-mf.inp > tmp 1.51user 0.30system 0:01.91elapsed 94%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed -f mac-mf.sed < mac-mf.inp > tmp2 0.30user 0.10system 0:00.45elapsed 88%CPU (0avgtext+0avgdata 0maxresident)k time ./sed -f khadafy.sed < khadafy.inp > tmp 0.01user 0.01system 0:00.04elapsed 50%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed -f khadafy.sed < khadafy.inp > tmp2 0.00user 0.03system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k sh spencer.sh ./sed < spencer.inp > spencer.tmp sh spencer.tmp time ./sed -f linecnt.sed < linecnt.inp > tmp 0.00user 0.02system 0:00.04elapsed 50%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed -f linecnt.sed < linecnt.inp > tmp2 0.00user 0.02system 0:00.02elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k time ./sed -f readin.sed < readin.inp >tmp 0.02user 0.02system 0:00.05elapsed 80%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed -f readin.sed < readin.inp >tmp2 0.01user 0.01system 0:00.06elapsed 33%CPU (0avgtext+0avgdata 0maxresident)k time ./sed --file=newjis.sed < newjis.inp > tmp 0.01user 0.01system 0:00.03elapsed 66%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed --file=newjis.sed < newjis.inp > tmp2 0.00user 0.03system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k time ./sed --quiet -f cv-vars.sed < cv-vars.inp > tmp 0.03user 0.02system 0:00.05elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k time /usr/bin/sed --quiet -f cv-vars.sed < cv-vars.inp > tmp2 0.01user 0.02system 0:00.03elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k dist/sed/testsuite/gmon.out100660 765 144 172450 6071117436 14737 0ustar lordusersà ðJ`GRXFX–ôm?´Å^G„G¨GÖ$'H•jÈ@"YDJc" … ˜ G¨ Gæ µ\ á fÐ µ®µLµÎÅ.[c""Áojµ\áØ'µ; î'•) (EF p(µF*:0="ˆ1%oÒ;Å. b<c"B=c"’=ÁoÊB#LfD"lªEJI¸EGòECZF´m ¾Fk*G¶BêÙ?Ü’¡"³OL"ÿd($.dd$³O´$w¢&)³O,)- Â,î!Œ6å_7œ‘Â=ÂHˆ?{6Þ?@ÚI5ŸüI5Ÿ0JÏITJH¼JÏI4LJêL#L\N³JlV7‚ŽVœ‘ˆW#LÐWC8XCTXCÞX³Oú_7‚` V>`} –`­,ž`i¬`N$Ü`- a_`$aî(Bb} Rdž$fdn!zdph ež$Lk_`Zka(l#LLlM(Xl.`nl@a6o–qpm)r³iäzƒÕ{ƒÕ${³i°‚œ‘Ä‚œ‘Ê‘ƒI\>Ÿ6¡Œ¢íhÒ¢#Lä¢N$dist/sed/testsuite/tmp2100660 765 144 320 6071120021 13735 0ustar lordusersäƤâ¤è ¤ßäÆ»ý¤Á ·¡¶\fa¤â¤è ¤ß·¡¶\fa»ý¤Á ¤³¤ÎµÖ¤\cb ºÚŦ¤Þ¤¹»\f9 \b2È´Ö¤\ab¤\ca ¹ð¤é¤µ¤Í ¤½¤é¤ß¤Ä \c2çÏ\c2¤Î¹ñ¤Ï ¤ª¤·¤ã¤\ca¤Ù¤Æ ¤ï¤ì¤³¤½µï¤ì ¤·¤­¤\ca¤Ù¤Æ ¤ï¤ì¤³¤½ º\c2¤» ¤ï¤\cb¤³¤½¤Ï ¹ð¤é¤á \b2Ȥ\f2¤â̾¤\f2¤â dist/sed/testsuite/spencer.tmp2100660 765 144 24707 6071053303 15462 0ustar lordusersecho 'abc' | /usr/bin/sed -re '/abc/d' | /usr/bin/sed -e s"/.*/Test #1 failed!/" echo 'xbc' | /usr/bin/sed -re '/abc/!d' | /usr/bin/sed -e s"/.*/Test #2 failed!/" echo 'axc' | /usr/bin/sed -re '/abc/!d' | /usr/bin/sed -e s"/.*/Test #3 failed!/" echo 'abx' | /usr/bin/sed -re '/abc/!d' | /usr/bin/sed -e s"/.*/Test #4 failed!/" echo 'xabcy' | /usr/bin/sed -re '/abc/d' | /usr/bin/sed -e s"/.*/Test #5 failed!/" echo 'ababc' | /usr/bin/sed -re '/abc/d' | /usr/bin/sed -e s"/.*/Test #6 failed!/" echo 'abc' | /usr/bin/sed -re '/ab*c/d' | /usr/bin/sed -e s"/.*/Test #7 failed!/" echo 'abc' | /usr/bin/sed -re '/ab*bc/d' | /usr/bin/sed -e s"/.*/Test #8 failed!/" echo 'abbc' | /usr/bin/sed -re '/ab*bc/d' | /usr/bin/sed -e s"/.*/Test #9 failed!/" echo 'abbbbc' | /usr/bin/sed -re '/ab*bc/d' | /usr/bin/sed -e s"/.*/Test #10 failed!/" echo 'abbc' | /usr/bin/sed -re '/ab+bc/d' | /usr/bin/sed -e s"/.*/Test #11 failed!/" echo 'abc' | /usr/bin/sed -re '/ab+bc/!d' | /usr/bin/sed -e s"/.*/Test #12 failed!/" echo 'abq' | /usr/bin/sed -re '/ab+bc/!d' | /usr/bin/sed -e s"/.*/Test #13 failed!/" echo 'abbbbc' | /usr/bin/sed -re '/ab+bc/d' | /usr/bin/sed -e s"/.*/Test #14 failed!/" echo 'abbc' | /usr/bin/sed -re '/ab?bc/d' | /usr/bin/sed -e s"/.*/Test #15 failed!/" echo 'abc' | /usr/bin/sed -re '/ab?bc/d' | /usr/bin/sed -e s"/.*/Test #16 failed!/" echo 'abbbbc' | /usr/bin/sed -re '/ab?bc/!d' | /usr/bin/sed -e s"/.*/Test #17 failed!/" echo 'abc' | /usr/bin/sed -re '/ab?c/d' | /usr/bin/sed -e s"/.*/Test #18 failed!/" echo 'abc' | /usr/bin/sed -re '/^abc$/d' | /usr/bin/sed -e s"/.*/Test #19 failed!/" echo 'abcc' | /usr/bin/sed -re '/^abc$/!d' | /usr/bin/sed -e s"/.*/Test #20 failed!/" echo 'abcc' | /usr/bin/sed -re '/^abc/d' | /usr/bin/sed -e s"/.*/Test #21 failed!/" echo 'aabc' | /usr/bin/sed -re '/^abc$/!d' | /usr/bin/sed -e s"/.*/Test #22 failed!/" echo 'aabc' | /usr/bin/sed -re '/abc$/d' | /usr/bin/sed -e s"/.*/Test #23 failed!/" echo 'abc' | /usr/bin/sed -re '/^/d' | /usr/bin/sed -e s"/.*/Test #24 failed!/" echo 'abc' | /usr/bin/sed -re '/$/d' | /usr/bin/sed -e s"/.*/Test #25 failed!/" echo 'abc' | /usr/bin/sed -re '/a.c/d' | /usr/bin/sed -e s"/.*/Test #26 failed!/" echo 'axc' | /usr/bin/sed -re '/a.c/d' | /usr/bin/sed -e s"/.*/Test #27 failed!/" echo 'axyzc' | /usr/bin/sed -re '/a.*c/d' | /usr/bin/sed -e s"/.*/Test #28 failed!/" echo 'axyzd' | /usr/bin/sed -re '/a.*c/!d' | /usr/bin/sed -e s"/.*/Test #29 failed!/" echo 'abc' | /usr/bin/sed -re '/a[bc]d/!d' | /usr/bin/sed -e s"/.*/Test #30 failed!/" echo 'abd' | /usr/bin/sed -re '/a[bc]d/d' | /usr/bin/sed -e s"/.*/Test #31 failed!/" echo 'abd' | /usr/bin/sed -re '/a[b-d]e/!d' | /usr/bin/sed -e s"/.*/Test #32 failed!/" echo 'ace' | /usr/bin/sed -re '/a[b-d]e/d' | /usr/bin/sed -e s"/.*/Test #33 failed!/" echo 'aac' | /usr/bin/sed -re '/a[b-d]/d' | /usr/bin/sed -e s"/.*/Test #34 failed!/" echo 'a-' | /usr/bin/sed -re '/a[-b]/d' | /usr/bin/sed -e s"/.*/Test #35 failed!/" echo 'a-' | /usr/bin/sed -re '/a[b-]/d' | /usr/bin/sed -e s"/.*/Test #36 failed!/" echo '-' | /usr/bin/sed -re '/a[b-a]/!d' | /usr/bin/sed -e s"/.*/Test #37 failed!/" echo '-' | /usr/bin/sed -re '/a[]b/p' | /usr/bin/sed -e '/bad regexp:/d' -e s"/.*/Test #38 failed!/" echo '-' | /usr/bin/sed -re '/a[/p' | /usr/bin/sed -e '/bad regexp:/d' -e s"/.*/Test #39 failed!/" echo 'a]' | /usr/bin/sed -re '/a]/d' | /usr/bin/sed -e s"/.*/Test #40 failed!/" echo 'a]b' | /usr/bin/sed -re '/a[]]b/d' | /usr/bin/sed -e s"/.*/Test #41 failed!/" echo 'aed' | /usr/bin/sed -re '/a[^bc]d/d' | /usr/bin/sed -e s"/.*/Test #42 failed!/" echo 'abd' | /usr/bin/sed -re '/a[^bc]d/!d' | /usr/bin/sed -e s"/.*/Test #43 failed!/" echo 'adc' | /usr/bin/sed -re '/a[^-b]c/d' | /usr/bin/sed -e s"/.*/Test #44 failed!/" echo 'a-c' | /usr/bin/sed -re '/a[^-b]c/!d' | /usr/bin/sed -e s"/.*/Test #45 failed!/" echo 'a]c' | /usr/bin/sed -re '/a[^]b]c/!d' | /usr/bin/sed -e s"/.*/Test #46 failed!/" echo 'adc' | /usr/bin/sed -re '/a[^]b]c/d' | /usr/bin/sed -e s"/.*/Test #47 failed!/" echo 'abc' | /usr/bin/sed -re '/ab|cd/d' | /usr/bin/sed -e s"/.*/Test #48 failed!/" echo 'abcd' | /usr/bin/sed -re '/ab|cd/d' | /usr/bin/sed -e s"/.*/Test #49 failed!/" echo 'def' | /usr/bin/sed -re '/()ef/d' | /usr/bin/sed -e s"/.*/Test #50 failed!/" echo '-' | /usr/bin/sed -re '/()*/d' | /usr/bin/sed -e s"/.*/Test #51 failed!/" echo '-' | /usr/bin/sed -re '/*a/!d' | /usr/bin/sed -e s"/.*/Test #52 failed!/" echo '-' | /usr/bin/sed -re '/^*/d' | /usr/bin/sed -e s"/.*/Test #53 failed!/" echo '-' | /usr/bin/sed -re '/$*/d' | /usr/bin/sed -e s"/.*/Test #54 failed!/" echo '-' | /usr/bin/sed -re '/(*)b/!d' | /usr/bin/sed -e s"/.*/Test #55 failed!/" echo 'b' | /usr/bin/sed -re '/$b/!d' | /usr/bin/sed -e s"/.*/Test #56 failed!/" echo '-' | /usr/bin/sed -re '/a\/p' | /usr/bin/sed -e '/bad regexp:/d' -e s"/.*/Test #57 failed!/" echo 'a(b' | /usr/bin/sed -re '/a\(b/d' | /usr/bin/sed -e s"/.*/Test #58 failed!/" echo 'ab' | /usr/bin/sed -re '/a\(*b/d' | /usr/bin/sed -e s"/.*/Test #59 failed!/" echo 'a((b' | /usr/bin/sed -re '/a\(*b/d' | /usr/bin/sed -e s"/.*/Test #60 failed!/" echo 'a\x' | /usr/bin/sed -re '/a\x/!d' | /usr/bin/sed -e s"/.*/Test #61 failed!/" echo '-' | /usr/bin/sed -re '/abc)/p' | /usr/bin/sed -e '/bad regexp:/d' -e s"/.*/Test #62 failed!/" echo '-' | /usr/bin/sed -re '/(abc/p' | /usr/bin/sed -e '/bad regexp:/d' -e s"/.*/Test #63 failed!/" echo 'abc' | /usr/bin/sed -re '/((a))/d' | /usr/bin/sed -e s"/.*/Test #64 failed!/" echo 'abc' | /usr/bin/sed -re '/(a)b(c)/d' | /usr/bin/sed -e s"/.*/Test #65 failed!/" echo 'aabbabc' | /usr/bin/sed -re '/a+b+c/d' | /usr/bin/sed -e s"/.*/Test #66 failed!/" echo '-' | /usr/bin/sed -re '/a**/d' | /usr/bin/sed -e s"/.*/Test #67 failed!/" echo '-' | /usr/bin/sed -re '/a*?/d' | /usr/bin/sed -e s"/.*/Test #68 failed!/" echo '-' | /usr/bin/sed -re '/(a*)*/d' | /usr/bin/sed -e s"/.*/Test #69 failed!/" echo '-' | /usr/bin/sed -re '/(a*)+/d' | /usr/bin/sed -e s"/.*/Test #70 failed!/" echo '-' | /usr/bin/sed -re '/(a|)*/d' | /usr/bin/sed -e s"/.*/Test #71 failed!/" echo '-' | /usr/bin/sed -re '/(a*|b)*/d' | /usr/bin/sed -e s"/.*/Test #72 failed!/" echo 'ab' | /usr/bin/sed -re '/(a+|b)*/d' | /usr/bin/sed -e s"/.*/Test #73 failed!/" echo 'ab' | /usr/bin/sed -re '/(a+|b)+/d' | /usr/bin/sed -e s"/.*/Test #74 failed!/" echo 'ab' | /usr/bin/sed -re '/(a+|b)?/d' | /usr/bin/sed -e s"/.*/Test #75 failed!/" echo 'cde' | /usr/bin/sed -re '/[^ab]*/d' | /usr/bin/sed -e s"/.*/Test #76 failed!/" echo '-' | /usr/bin/sed -re '/(^)*/d' | /usr/bin/sed -e s"/.*/Test #77 failed!/" echo '-' | /usr/bin/sed -re '/(ab|)*/d' | /usr/bin/sed -e s"/.*/Test #78 failed!/" echo '-' | /usr/bin/sed -re '/)(/p' | /usr/bin/sed -e '/bad regexp:/d' -e s"/.*/Test #79 failed!/" echo '' | /usr/bin/sed -re '/abc/!d' | /usr/bin/sed -e s"/.*/Test #80 failed!/" echo '' | /usr/bin/sed -re '/abc/!d' | /usr/bin/sed -e s"/.*/Test #81 failed!/" echo '' | /usr/bin/sed -re '/a*/d' | /usr/bin/sed -e s"/.*/Test #82 failed!/" echo 'abbbcd' | /usr/bin/sed -re '/([abc])*d/d' | /usr/bin/sed -e s"/.*/Test #83 failed!/" echo 'abcd' | /usr/bin/sed -re '/([abc])*bcd/d' | /usr/bin/sed -e s"/.*/Test #84 failed!/" echo 'e' | /usr/bin/sed -re '/a|b|c|d|e/d' | /usr/bin/sed -e s"/.*/Test #85 failed!/" echo 'ef' | /usr/bin/sed -re '/(a|b|c|d|e)f/d' | /usr/bin/sed -e s"/.*/Test #86 failed!/" echo '-' | /usr/bin/sed -re '/((a*|b))*/d' | /usr/bin/sed -e s"/.*/Test #87 failed!/" echo 'abcdefg' | /usr/bin/sed -re '/abcd*efg/d' | /usr/bin/sed -e s"/.*/Test #88 failed!/" echo 'xabyabbbz' | /usr/bin/sed -re '/ab*/d' | /usr/bin/sed -e s"/.*/Test #89 failed!/" echo 'xayabbbz' | /usr/bin/sed -re '/ab*/d' | /usr/bin/sed -e s"/.*/Test #90 failed!/" echo 'abcde' | /usr/bin/sed -re '/(ab|cd)e/d' | /usr/bin/sed -e s"/.*/Test #91 failed!/" echo 'hij' | /usr/bin/sed -re '/[abhgefdc]ij/d' | /usr/bin/sed -e s"/.*/Test #92 failed!/" echo 'abcde' | /usr/bin/sed -re '/^(ab|cd)e/!d' | /usr/bin/sed -e s"/.*/Test #93 failed!/" echo 'abcdef' | /usr/bin/sed -re '/(abc|)ef/d' | /usr/bin/sed -e s"/.*/Test #94 failed!/" echo 'abcd' | /usr/bin/sed -re '/(a|b)c*d/d' | /usr/bin/sed -e s"/.*/Test #95 failed!/" echo 'abc' | /usr/bin/sed -re '/(ab|ab*)bc/d' | /usr/bin/sed -e s"/.*/Test #96 failed!/" echo 'abc' | /usr/bin/sed -re '/a([bc]*)c*/d' | /usr/bin/sed -e s"/.*/Test #97 failed!/" echo 'abcd' | /usr/bin/sed -re '/a([bc]*)(c*d)/d' | /usr/bin/sed -e s"/.*/Test #98 failed!/" echo 'abcd' | /usr/bin/sed -re '/a([bc]+)(c*d)/d' | /usr/bin/sed -e s"/.*/Test #99 failed!/" echo 'abcd' | /usr/bin/sed -re '/a([bc]*)(c+d)/d' | /usr/bin/sed -e s"/.*/Test #100 failed!/" echo 'adcdcde' | /usr/bin/sed -re '/a[bcd]*dcdcde/d' | /usr/bin/sed -e s"/.*/Test #101 failed!/" echo 'adcdcde' | /usr/bin/sed -re '/a[bcd]+dcdcde/!d' | /usr/bin/sed -e s"/.*/Test #102 failed!/" echo 'abc' | /usr/bin/sed -re '/(ab|a)b*c/d' | /usr/bin/sed -e s"/.*/Test #103 failed!/" echo 'abcd' | /usr/bin/sed -re '/((a)(b)c)(d)/d' | /usr/bin/sed -e s"/.*/Test #104 failed!/" echo 'alpha' | /usr/bin/sed -re '/[A-Za-z_][A-Za-z0-9_]*/d' | /usr/bin/sed -e s"/.*/Test #105 failed!/" echo 'abh' | /usr/bin/sed -re '/^a(bc+|b[eh])g|.h$/d' | /usr/bin/sed -e s"/.*/Test #106 failed!/" echo 'effgz' | /usr/bin/sed -re '/(bc+d$|ef*g.|h?i(j|k))/d' | /usr/bin/sed -e s"/.*/Test #107 failed!/" echo 'ij' | /usr/bin/sed -re '/(bc+d$|ef*g.|h?i(j|k))/d' | /usr/bin/sed -e s"/.*/Test #108 failed!/" echo 'effg' | /usr/bin/sed -re '/(bc+d$|ef*g.|h?i(j|k))/!d' | /usr/bin/sed -e s"/.*/Test #109 failed!/" echo 'bcdd' | /usr/bin/sed -re '/(bc+d$|ef*g.|h?i(j|k))/!d' | /usr/bin/sed -e s"/.*/Test #110 failed!/" echo 'reffgz' | /usr/bin/sed -re '/(bc+d$|ef*g.|h?i(j|k))/d' | /usr/bin/sed -e s"/.*/Test #111 failed!/" echo '-' | /usr/bin/sed -re '/((((((((((a))))))))))/!d' | /usr/bin/sed -e s"/.*/Test #112 failed!/" echo 'a' | /usr/bin/sed -re '/(((((((((a)))))))))/d' | /usr/bin/sed -e s"/.*/Test #113 failed!/" echo 'uh-uh' | /usr/bin/sed -re '/multiple words of text/!d' | /usr/bin/sed -e s"/.*/Test #114 failed!/" echo 'multiple words, yeah' | /usr/bin/sed -re '/multiple words/d' | /usr/bin/sed -e s"/.*/Test #115 failed!/" echo 'abcde' | /usr/bin/sed -re '/(.*)c(.*)/d' | /usr/bin/sed -e s"/.*/Test #116 failed!/" echo '(.*)\)' | /usr/bin/sed -re '/\((.*),/!d' | /usr/bin/sed -e s"/.*/Test #117 failed!/" echo 'ab' | /usr/bin/sed -re '/[k]/!d' | /usr/bin/sed -e s"/.*/Test #118 failed!/" echo 'abcd' | /usr/bin/sed -re '/abcd/d' | /usr/bin/sed -e s"/.*/Test #119 failed!/" echo 'abcd' | /usr/bin/sed -re '/a(bc)d/d' | /usr/bin/sed -e s"/.*/Test #120 failed!/" echo 'ac' | /usr/bin/sed -re '/a[-]?c/d' | /usr/bin/sed -e s"/.*/Test #121 failed!/" echo 'beriberi' | /usr/bin/sed -re '/(....).*\1/d' | /usr/bin/sed -e s"/.*/Test #122 failed!/" dist/sed/PLUGIN/ 42770 765 144 0 6071542673 12053 5ustar lordusersdist/sed/PLUGIN/REQ100660 765 144 7 6070131741 12423 0ustar lordusersrx sed dist/sed/PLUGIN/greet100660 765 144 1006 6071542673 13175 0ustar lordusers She said, "You don't understand what I've said" GNU Sed; I said filters all unwanted ASCII sed -- "No, no no you're wrong. iment. When I was a boy... The less sed the better. everything was right. 'nuf said. Everything was right." I said, "Even though you know what you know I know that I'm ready to leave, cause your making me feel like I've never been born." dist/sed/PLUGIN/greet.~1~100660 765 144 1026 6071407242 13722 0ustar lordusers She said, "You don't understand what I've said" GNU Sed; I said filters all unwanted ASCII sed -- "No, no no you're wrong. iment. When I was a boy... The less sed the better. everything was right. 'nuf said. Everything was right." I said, "Even though you know what you know I know that I'm ready to leave, cause your making me feel like I've never been born."