summaryrefslogtreecommitdiff
path: root/gnu/usr.bin/diff
diff options
context:
space:
mode:
authorjtc <jtc@NetBSD.org>1993-07-08 20:13:23 +0000
committerjtc <jtc@NetBSD.org>1993-07-08 20:13:23 +0000
commitd996bc247f59fdf10cf4b6618131b5062e478999 (patch)
tree4a74e275e3f737762cfe0e24e497cdb45ea14347 /gnu/usr.bin/diff
parenta18c6474da95632c233f93fa4ecd3de9a5949ad2 (diff)
Update to GNU diff 2.3
Diffstat (limited to 'gnu/usr.bin/diff')
-rw-r--r--gnu/usr.bin/diff/COPYING354
-rw-r--r--gnu/usr.bin/diff/Makefile6
-rw-r--r--gnu/usr.bin/diff/analyze.c410
-rw-r--r--gnu/usr.bin/diff/context.c122
-rw-r--r--gnu/usr.bin/diff/diff.c643
-rw-r--r--gnu/usr.bin/diff/diff.h209
-rw-r--r--gnu/usr.bin/diff/dir.c252
-rw-r--r--gnu/usr.bin/diff/ed.c16
-rw-r--r--gnu/usr.bin/diff/getopt.c412
-rw-r--r--gnu/usr.bin/diff/getopt.h109
-rw-r--r--gnu/usr.bin/diff/getopt1.c94
-rw-r--r--gnu/usr.bin/diff/ifdef.c156
-rw-r--r--gnu/usr.bin/diff/io.c973
-rw-r--r--gnu/usr.bin/diff/normal.c5
-rw-r--r--gnu/usr.bin/diff/regex.c6129
-rw-r--r--gnu/usr.bin/diff/regex.h643
-rw-r--r--gnu/usr.bin/diff/side.c283
-rw-r--r--gnu/usr.bin/diff/system.h159
-rw-r--r--gnu/usr.bin/diff/util.c281
-rw-r--r--gnu/usr.bin/diff/version.c2
20 files changed, 7452 insertions, 3806 deletions
diff --git a/gnu/usr.bin/diff/COPYING b/gnu/usr.bin/diff/COPYING
index 9a170375811..a43ea2126fb 100644
--- a/gnu/usr.bin/diff/COPYING
+++ b/gnu/usr.bin/diff/COPYING
@@ -1,38 +1,40 @@
-
GNU GENERAL PUBLIC LICENSE
- Version 1, February 1989
+ Version 2, June 1991
- Copyright (C) 1989 Free Software Foundation, Inc.
- 675 Mass Ave, Cambridge, MA 02139, USA
+ Copyright (C) 1989, 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.
Preamble
- The license agreements of most software companies try to keep users
-at the mercy of those companies. By contrast, our General Public
+ 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. The
-General Public License applies to the Free Software Foundation's
-software and to any other program whose authors commit to using it.
-You can use it for your programs, too.
+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. Specifically, the General Public License is designed to make
-sure that you have the freedom to give away or sell copies of free
-software, 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.
+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 a such a program, whether
+ 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 tell them their rights.
+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,
@@ -45,120 +47,207 @@ 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 Agreement 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 work containing the
-Program or a portion of it, either verbatim or with modifications. Each
-licensee is addressed as "you".
-
- 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
-General Public License and to the absence of any warranty; and give any
-other recipients of the Program a copy of this General Public License
-along with the Program. You may charge a fee for the physical act of
-transferring a copy.
-
- 2. You may modify your copy or copies of the Program or any portion of
-it, and copy and distribute such modifications under the terms of Paragraph
-1 above, provided that you also do the following:
-
- a) cause the modified files to carry prominent notices stating that
- you changed the files and the date of any change; and
-
- b) cause the whole of any work that you distribute or publish, that
- in whole or in part contains the Program or any part thereof, either
- with or without modifications, to be licensed at no charge to all
- third parties under the terms of this General Public License (except
- that you may choose to grant warranty protection to some or all
- third parties, at your option).
-
- c) If the modified program normally reads commands interactively when
- run, you must cause it, when started running for such interactive use
- in the simplest and most usual 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 General
- Public License.
-
- d) 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.
-
-Mere aggregation of another independent work with the Program (or its
-derivative) on a volume of a storage or distribution medium does not bring
-the other work under the scope of these terms.
+ 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.)
- 3. You may copy and distribute the Program (or a portion or derivative of
-it, under Paragraph 2) in object code or executable form under the terms of
-Paragraphs 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
- Paragraphs 1 and 2 above; or,
-
- b) accompany it with a written offer, valid for at least three
- years, to give any third party free (except for a nominal charge
- for the cost of distribution) a complete machine-readable copy of the
- corresponding source code, to be distributed under the terms of
- Paragraphs 1 and 2 above; or,
-
- c) accompany it with the information you received as to where the
- corresponding source code may be obtained. (This alternative is
+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 alone.)
-
-Source code for a work means the preferred form of the work for making
-modifications to it. For an executable file, complete source code means
-all the source code for all modules it contains; but, as a special
-exception, it need not include source code for modules which are standard
-libraries that accompany the operating system on which the executable
-file runs, or for standard header files or definitions files that
-accompany that operating system.
-
- 4. You may not copy, modify, sublicense, distribute or transfer the
-Program except as expressly provided under this General Public License.
-Any attempt otherwise to copy, modify, sublicense, distribute or transfer
-the Program is void, and will automatically terminate your rights to use
-the Program under this License. However, parties who have received
-copies, or rights to use copies, from you under this General Public
-License will not have their licenses terminated so long as such parties
-remain in full compliance.
-
- 5. By copying, distributing or modifying 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.
+ 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.
+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.
- 7. The Free Software Foundation may publish revised and/or new versions
+ 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 the license which applies to it and "any
+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
-the license, you may choose any version ever published by the Free Software
+this License, you may choose any version ever published by the Free Software
Foundation.
- 8. If you wish to incorporate parts of the Program into other free
+ 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
@@ -168,7 +257,7 @@ of promoting the sharing and reuse of software generally.
NO WARRANTY
- 9. BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS 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
@@ -178,7 +267,7 @@ 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.
- 10. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
+ 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
@@ -193,22 +282,21 @@ POSSIBILITY OF SUCH DAMAGES.
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 humanity, the best way to achieve this is to make it
-free software which everyone can redistribute and change under these
-terms.
+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.
+ 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.
<one line to give the program's name and a brief idea of what it does.>
Copyright (C) 19yy <name of author>
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 1, or (at your option)
- any later version.
+ 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
@@ -224,26 +312,28 @@ 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) 19xx name of author
+ 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.
+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 a sample; alter the names:
+necessary. Here is a sample; alter the names:
- Yoyodyne, Inc., hereby disclaims all copyright interest in the
- program `Gnomovision' (a program to direct compilers to make passes
- at assemblers) written by James Hacker.
+ Yoyodyne, Inc., hereby disclaims all copyright interest in the program
+ `Gnomovision' (which makes passes at compilers) written by James Hacker.
<signature of Ty Coon>, 1 April 1989
Ty Coon, President of Vice
-That's all there is to it!
+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.
diff --git a/gnu/usr.bin/diff/Makefile b/gnu/usr.bin/diff/Makefile
index a4f60b83d28..e75c0a00110 100644
--- a/gnu/usr.bin/diff/Makefile
+++ b/gnu/usr.bin/diff/Makefile
@@ -1,7 +1,11 @@
PROG= diff
SRCS= diff.c analyze.c io.c context.c ed.c normal.c ifdef.c util.c dir.c \
- version.c regex.c getopt.c getopt1.c
+ version.c regex.c getopt.c getopt1.c side.c
+CFLAGS+=-DDIRENT=1 -DHAVE_UNISTD_H=1 -DHAVE_DUP2=1 -DHAVE_MEMCHR=1 \
+ -DHAVE_STRERROR=1 -DHAVE_WAITPID=1 -DHAVE_FCNTL_H=1 \
+ -DHAVE_STRING_H=1 -DHAVE_SYS_WAIT_H=1 -DHAVE_TIME_H=1 \
+ -DHAVE_ST_BLKSIZE=1
NOMAN=noman
.include <bsd.prog.mk>
diff --git a/gnu/usr.bin/diff/analyze.c b/gnu/usr.bin/diff/analyze.c
index cf93f4588a6..42167b9b7f2 100644
--- a/gnu/usr.bin/diff/analyze.c
+++ b/gnu/usr.bin/diff/analyze.c
@@ -1,11 +1,11 @@
/* Analyze file differences for GNU DIFF.
- Copyright (C) 1988, 1989 Free Software Foundation, Inc.
+ Copyright (C) 1988, 1989, 1992 Free Software Foundation, Inc.
This file is part of GNU DIFF.
GNU DIFF 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 1, or (at your option)
+the Free Software Foundation; either version 2, or (at your option)
any later version.
GNU DIFF is distributed in the hope that it will be useful,
@@ -23,12 +23,12 @@ the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */
#include "diff.h"
-struct change *find_change ();
+int read_files ();
void finish_output ();
-void print_context_header ();
void print_context_script ();
void print_ed_script ();
void print_ifdef_script ();
+void print_sdiff_script ();
void print_normal_script ();
void print_rcs_script ();
void pr_forward_ed_script ();
@@ -86,7 +86,7 @@ diag (xoff, xlim, yoff, ylim, cost)
int fmin = fmid, fmax = fmid; /* Limits of top-down search. */
int bmin = bmid, bmax = bmid; /* Limits of bottom-up search. */
int c; /* Cost. */
- int odd = fmid - bmid & 1; /* True if southeast corner is on an odd
+ int odd = (fmid - bmid) & 1; /* True if southeast corner is on an odd
diagonal with respect to the northwest. */
fd[fmid] = xoff;
@@ -315,23 +315,24 @@ discard_confusing_lines (filevec)
unsigned int f, i;
char *discarded[2];
int *equiv_count[2];
+ int *p;
/* Allocate our results. */
+ p = (int *) xmalloc ((filevec[0].buffered_lines + filevec[1].buffered_lines)
+ * (2 * sizeof (int)));
for (f = 0; f < 2; f++)
{
- filevec[f].undiscarded
- = (int *) xmalloc (filevec[f].buffered_lines * sizeof (int));
- filevec[f].realindexes
- = (int *) xmalloc (filevec[f].buffered_lines * sizeof (int));
+ filevec[f].undiscarded = p; p += filevec[f].buffered_lines;
+ filevec[f].realindexes = p; p += filevec[f].buffered_lines;
}
/* Set up equiv_count[F][I] as the number of lines in file F
that fall in equivalence class I. */
- equiv_count[0] = (int *) xmalloc (filevec[0].equiv_max * sizeof (int));
- bzero (equiv_count[0], filevec[0].equiv_max * sizeof (int));
- equiv_count[1] = (int *) xmalloc (filevec[1].equiv_max * sizeof (int));
- bzero (equiv_count[1], filevec[1].equiv_max * sizeof (int));
+ p = (int *) xmalloc (filevec[0].equiv_max * (2 * sizeof (int)));
+ equiv_count[0] = p;
+ equiv_count[1] = p + filevec[0].equiv_max;
+ bzero (p, filevec[0].equiv_max * (2 * sizeof (int)));
for (i = 0; i < filevec[0].buffered_lines; ++i)
++equiv_count[0][filevec[0].equivs[i]];
@@ -340,10 +341,10 @@ discard_confusing_lines (filevec)
/* Set up tables of which lines are going to be discarded. */
- discarded[0] = (char *) xmalloc (filevec[0].buffered_lines);
- discarded[1] = (char *) xmalloc (filevec[1].buffered_lines);
- bzero (discarded[0], filevec[0].buffered_lines);
- bzero (discarded[1], filevec[1].buffered_lines);
+ discarded[0] = (char *) xmalloc (filevec[0].buffered_lines
+ + filevec[1].buffered_lines);
+ discarded[1] = discarded[0] + filevec[0].buffered_lines;
+ bzero (discarded[0], filevec[0].buffered_lines + filevec[1].buffered_lines);
/* Mark to be discarded each line that matches no line of the other file.
If a line matches many lines, mark it as provisionally discardable. */
@@ -504,9 +505,7 @@ discard_confusing_lines (filevec)
filevec[f].nondiscarded_lines = j;
}
- free (discarded[1]);
free (discarded[0]);
- free (equiv_count[1]);
free (equiv_count[0]);
}
@@ -544,7 +543,7 @@ shift_boundaries (filevec)
while (1)
{
- int start, end, other_start;
+ int start, other_start;
/* Scan forwards to find beginning of another run of changes.
Also keep track of the corresponding point in the other file. */
@@ -568,8 +567,8 @@ shift_boundaries (filevec)
{
/* Now find the end of this run of changes. */
- while (i < i_end && changed[i] != 0) i++;
- end = i;
+ while (changed[++i] != 0)
+ ;
/* If the first changed line matches the following unchanged one,
and this run does not follow right after a previous run,
@@ -580,17 +579,13 @@ shift_boundaries (filevec)
/* You might ask, how could this run follow right after another?
Only because the previous run was shifted here. */
- if (end != i_end
- && files[f].equivs[start] == files[f].equivs[end]
+ if (i != i_end
+ && files[f].equivs[start] == files[f].equivs[i]
&& !other_changed[j]
- && end != i_end
- && !((preceding >= 0 && start == preceding)
- || (other_preceding >= 0
- && other_start == other_preceding)))
+ && !(start == preceding || other_start == other_preceding))
{
- changed[end++] = 1;
changed[start++] = 0;
- ++i;
+ changed[i] = 1;
/* Since one line-that-matches is now before this run
instead of after, we must advance in the other file
to keep in synch. */
@@ -677,20 +672,10 @@ build_script (filevec)
struct change *script = 0;
char *changed0 = filevec[0].changed_flag;
char *changed1 = filevec[1].changed_flag;
- int len0 = filevec[0].buffered_lines;
- int len1 = filevec[1].buffered_lines;
- int i0 = len0, i1 = len1;
+ int i0 = filevec[0].buffered_lines, i1 = filevec[1].buffered_lines;
/* Note that changedN[-1] does exist, and contains 0. */
-#if 0 /* Unnecessary since a line includes its trailing newline. */
- /* In RCS comparisons, making the existence or nonexistence of trailing
- newlines really matter. */
- if (output_style == OUTPUT_RCS
- && filevec[0].missing_newline != filevec[1].missing_newline)
- changed0[len0 - 1] = changed1[len1 - 1] = 1;
-#endif
-
while (i0 >= 0 || i1 >= 0)
{
if (changed0[i0 - 1] || changed1[i1 - 1])
@@ -723,202 +708,237 @@ diff_2_files (filevec, depth)
int i;
struct change *e, *p;
struct change *script;
- int binary;
int changes;
- /* See if the two named files are actually the same physical file.
- If so, we know they are identical without actually reading them. */
- if (output_style != OUTPUT_IFDEF
- && filevec[0].stat.st_ino == filevec[1].stat.st_ino
- && filevec[0].stat.st_dev == filevec[1].stat.st_dev)
- return 0;
-
- binary = read_files (filevec);
-
- /* If we have detected that file 0 is a binary file,
+ /* If we have detected that either file is binary,
compare the two files as binary. This can happen
only when the first chunk is read.
Also, -q means treat all files as binary. */
- if (binary || no_details_flag)
+ if (read_files (filevec))
{
- int differs = (filevec[0].buffered_chars != filevec[1].buffered_chars
- || bcmp (filevec[0].buffer, filevec[1].buffer,
- filevec[1].buffered_chars));
- if (differs)
- message (binary ? "Binary files %s and %s differ\n"
- : "Files %s and %s differ\n",
- filevec[0].name, filevec[1].name);
+ /* Files with different lengths must be different. */
+ if (filevec[0].stat.st_size != filevec[1].stat.st_size
+ && (filevec[0].desc < 0 || S_ISREG (filevec[0].stat.st_mode))
+ && (filevec[1].desc < 0 || S_ISREG (filevec[1].stat.st_mode)))
+ changes = 1;
- for (i = 0; i < 2; ++i)
- if (filevec[i].buffer)
- free (filevec[i].buffer);
- return differs;
- }
+ /* Standard input equals itself. */
+ else if (filevec[0].desc == filevec[1].desc)
+ changes = 0;
- /* Allocate vectors for the results of comparison:
- a flag for each line of each file, saying whether that line
- is an insertion or deletion.
- Allocate an extra element, always zero, at each end of each vector. */
+ else
+ /* Scan both files, a buffer at a time, looking for a difference. */
+ {
+ /* Allocate same-sized buffers for both files. */
+ int buffer_size = max (STAT_BLOCKSIZE (filevec[0].stat),
+ STAT_BLOCKSIZE (filevec[1].stat));
+ for (i = 0; i < 2; i++)
+ filevec[i].buffer = xrealloc (filevec[i].buffer, buffer_size);
- filevec[0].changed_flag = (char *) xmalloc (filevec[0].buffered_lines + 2);
- filevec[1].changed_flag = (char *) xmalloc (filevec[1].buffered_lines + 2);
- bzero (filevec[0].changed_flag, filevec[0].buffered_lines + 2);
- bzero (filevec[1].changed_flag, filevec[1].buffered_lines + 2);
- filevec[0].changed_flag++;
- filevec[1].changed_flag++;
+ for (;; filevec[0].buffered_chars = filevec[1].buffered_chars = 0)
+ {
+ /* Read a buffer's worth from both files. */
+ for (i = 0; i < 2; i++)
+ if (0 <= filevec[i].desc)
+ while (filevec[i].buffered_chars != buffer_size)
+ {
+ int r = read (filevec[i].desc,
+ filevec[i].buffer
+ + filevec[i].buffered_chars,
+ buffer_size - filevec[i].buffered_chars);
+ if (r == 0)
+ break;
+ if (r < 0)
+ pfatal_with_name (filevec[i].name);
+ filevec[i].buffered_chars += r;
+ }
- /* Some lines are obviously insertions or deletions
- because they don't match anything. Detect them now,
- and avoid even thinking about them in the main comparison algorithm. */
+ /* If the buffers differ, the files differ. */
+ if (filevec[0].buffered_chars != filevec[1].buffered_chars
+ || bcmp (filevec[0].buffer,
+ filevec[1].buffer,
+ filevec[0].buffered_chars) != 0)
+ {
+ changes = 1;
+ break;
+ }
- discard_confusing_lines (filevec);
+ /* If we reach end of file, the files are the same. */
+ if (filevec[0].buffered_chars != buffer_size)
+ {
+ changes = 0;
+ break;
+ }
+ }
+ }
- /* Now do the main comparison algorithm, considering just the
- undiscarded lines. */
+ if (changes)
+ message (no_details_flag ? "Files %s and %s differ\n"
+ : "Binary files %s and %s differ\n",
+ filevec[0].name, filevec[1].name);
+ }
+ else
+ {
+ /* Allocate vectors for the results of comparison:
+ a flag for each line of each file, saying whether that line
+ is an insertion or deletion.
+ Allocate an extra element, always zero, at each end of each vector. */
- xvec = filevec[0].undiscarded;
- yvec = filevec[1].undiscarded;
- diags = filevec[0].nondiscarded_lines + filevec[1].nondiscarded_lines + 3;
- fdiag = (int *) xmalloc (diags * sizeof (int));
- fdiag += filevec[1].nondiscarded_lines + 1;
- bdiag = (int *) xmalloc (diags * sizeof (int));
- bdiag += filevec[1].nondiscarded_lines + 1;
+ filevec[0].changed_flag = (char *) xmalloc (filevec[0].buffered_lines
+ + filevec[1].buffered_lines
+ + 4);
+ bzero (filevec[0].changed_flag, filevec[0].buffered_lines
+ + filevec[1].buffered_lines + 4);
+ filevec[0].changed_flag++;
+ filevec[1].changed_flag = filevec[0].changed_flag
+ + filevec[0].buffered_lines + 2;
- files[0] = filevec[0];
- files[1] = filevec[1];
+ /* Some lines are obviously insertions or deletions
+ because they don't match anything. Detect them now, and
+ avoid even thinking about them in the main comparison algorithm. */
- compareseq (0, filevec[0].nondiscarded_lines,
- 0, filevec[1].nondiscarded_lines);
+ discard_confusing_lines (filevec);
- bdiag -= filevec[1].nondiscarded_lines + 1;
- free (bdiag);
- fdiag -= filevec[1].nondiscarded_lines + 1;
- free (fdiag);
+ /* Now do the main comparison algorithm, considering just the
+ undiscarded lines. */
- /* Modify the results slightly to make them prettier
- in cases where that can validly be done. */
+ xvec = filevec[0].undiscarded;
+ yvec = filevec[1].undiscarded;
+ diags = filevec[0].nondiscarded_lines + filevec[1].nondiscarded_lines + 3;
+ fdiag = (int *) xmalloc (diags * (2 * sizeof (int)));
+ bdiag = fdiag + diags;
+ fdiag += filevec[1].nondiscarded_lines + 1;
+ bdiag += filevec[1].nondiscarded_lines + 1;
- shift_boundaries (filevec);
+ files[0] = filevec[0];
+ files[1] = filevec[1];
- /* Get the results of comparison in the form of a chain
- of `struct change's -- an edit script. */
+ compareseq (0, filevec[0].nondiscarded_lines,
+ 0, filevec[1].nondiscarded_lines);
- if (output_style == OUTPUT_ED)
- script = build_reverse_script (filevec);
- else
- script = build_script (filevec);
+ free (fdiag - (filevec[1].nondiscarded_lines + 1));
- if (script || output_style == OUTPUT_IFDEF)
- {
- setup_output (files[0].name, files[1].name, depth);
+ /* Modify the results slightly to make them prettier
+ in cases where that can validly be done. */
- switch (output_style)
- {
- case OUTPUT_CONTEXT:
- print_context_header (files, 0);
- print_context_script (script, 0);
- break;
-
- case OUTPUT_UNIFIED:
- print_context_header (files, 1);
- print_context_script (script, 1);
- break;
-
- case OUTPUT_ED:
- print_ed_script (script);
- break;
-
- case OUTPUT_FORWARD_ED:
- pr_forward_ed_script (script);
- break;
-
- case OUTPUT_RCS:
- print_rcs_script (script);
- break;
-
- case OUTPUT_NORMAL:
- print_normal_script (script);
- break;
-
- case OUTPUT_IFDEF:
- print_ifdef_script (script);
- break;
- }
+ shift_boundaries (filevec);
- finish_output ();
- }
+ /* Get the results of comparison in the form of a chain
+ of `struct change's -- an edit script. */
- /* Set CHANGES if we had any diffs that were printed.
- If some changes are being ignored, we must scan the script to decide. */
- if (ignore_blank_lines_flag || ignore_regexp)
- {
- struct change *next = script;
- changes = 0;
+ if (output_style == OUTPUT_ED)
+ script = build_reverse_script (filevec);
+ else
+ script = build_script (filevec);
- while (next && changes == 0)
+ if (script || ! no_diff_means_no_output)
{
- struct change *this, *end;
- int first0, last0, first1, last1, deletes, inserts;
+ /* Record info for starting up output,
+ to be used if and when we have some output to print. */
+ setup_output (files[0].name, files[1].name, depth);
- /* Find a set of changes that belong together. */
- this = next;
- end = find_change (next);
+ switch (output_style)
+ {
+ case OUTPUT_CONTEXT:
+ print_context_script (script, 0);
+ break;
+
+ case OUTPUT_UNIFIED:
+ print_context_script (script, 1);
+ break;
+
+ case OUTPUT_ED:
+ print_ed_script (script);
+ break;
+
+ case OUTPUT_FORWARD_ED:
+ pr_forward_ed_script (script);
+ break;
- /* Disconnect them from the rest of the changes,
- making them a hunk, and remember the rest for next iteration. */
- next = end->link;
- end->link = NULL;
+ case OUTPUT_RCS:
+ print_rcs_script (script);
+ break;
- /* Determine whether this hunk was printed. */
- analyze_hunk (this, &first0, &last0, &first1, &last1,
- &deletes, &inserts);
+ case OUTPUT_NORMAL:
+ print_normal_script (script);
+ break;
- /* Reconnect the script so it will all be freed properly. */
- end->link = next;
+ case OUTPUT_IFDEF:
+ print_ifdef_script (script);
+ break;
- if (deletes || inserts)
- changes = 1;
+ case OUTPUT_SDIFF:
+ print_sdiff_script (script);
+ }
+
+ finish_output ();
}
- }
- else
- changes = (script != 0);
- for (i = 1; i >= 0; --i)
- {
- free (filevec[i].realindexes);
- free (filevec[i].undiscarded);
- }
+ /* Set CHANGES if we had any diffs that were printed.
+ If some changes are ignored, we must scan the script to decide. */
+ if (ignore_blank_lines_flag || ignore_regexp_list)
+ {
+ struct change *next = script;
+ changes = 0;
- for (i = 1; i >= 0; --i)
- free (--filevec[i].changed_flag);
+ while (next && changes == 0)
+ {
+ struct change *this, *end;
+ int first0, last0, first1, last1, deletes, inserts;
- for (i = 1; i >= 0; --i)
- free (filevec[i].equivs);
+ /* Find a set of changes that belong together. */
+ this = next;
+ end = find_change (next);
- for (i = 0; i < 2; ++i)
- {
- if (filevec[i].buffer != 0)
- free (filevec[i].buffer);
- free (filevec[i].linbuf);
- }
+ /* Disconnect them from the rest of the changes, making them
+ a hunk, and remember the rest for next iteration. */
+ next = end->link;
+ end->link = NULL;
- for (e = script; e; e = p)
- {
- p = e->link;
- free (e);
- }
+ /* Determine whether this hunk was printed. */
+ analyze_hunk (this, &first0, &last0, &first1, &last1,
+ &deletes, &inserts);
- if (! ROBUST_OUTPUT_STYLE (output_style)
- /* For -D, invent newlines silently. That's ok in C code. */
- && output_style != OUTPUT_IFDEF)
- for (i = 0; i < 2; ++i)
- if (filevec[i].missing_newline)
+ /* Reconnect the script so it will all be freed properly. */
+ end->link = next;
+
+ if (deletes || inserts)
+ changes = 1;
+ }
+ }
+ else
+ changes = (script != 0);
+
+ free (filevec[0].undiscarded);
+
+ free (filevec[0].changed_flag - 1);
+
+ for (i = 1; i >= 0; --i)
+ free (filevec[i].equivs);
+
+ for (i = 0; i < 2; ++i)
+ free (filevec[i].linbuf + filevec[i].linbuf_base);
+
+ for (e = script; e; e = p)
{
- error ("No newline at end of file %s", filevec[i].name, "");
- changes = 2;
+ p = e->link;
+ free (e);
}
+ if (! ROBUST_OUTPUT_STYLE (output_style))
+ for (i = 0; i < 2; ++i)
+ if (filevec[i].missing_newline)
+ {
+ error ("No newline at end of file %s", filevec[i].name, "");
+ changes = 2;
+ }
+ }
+
+ if (filevec[0].buffer != filevec[1].buffer)
+ free (filevec[0].buffer);
+ free (filevec[1].buffer);
+
return changes;
}
diff --git a/gnu/usr.bin/diff/context.c b/gnu/usr.bin/diff/context.c
index d3be6dbc8ac..9b8fc3ec171 100644
--- a/gnu/usr.bin/diff/context.c
+++ b/gnu/usr.bin/diff/context.c
@@ -1,11 +1,11 @@
/* Context-format output routines for GNU DIFF.
- Copyright (C) 1988, 1989 Free Software Foundation, Inc.
+ Copyright (C) 1988, 89, 91, 92 Free Software Foundation, Inc.
This file is part of GNU DIFF.
GNU DIFF 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 1, or (at your option)
+the Free Software Foundation; either version 2, or (at your option)
any later version.
GNU DIFF is distributed in the hope that it will be useful,
@@ -18,7 +18,6 @@ along with GNU DIFF; see the file COPYING. If not, write to
the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */
#include "diff.h"
-#include "regex.h"
static void pr_context_hunk ();
static void pr_unidiff_hunk ();
@@ -75,7 +74,7 @@ print_context_script (script, unidiff_flag)
struct change *script;
int unidiff_flag;
{
- if (ignore_blank_lines_flag || ignore_regexp)
+ if (ignore_blank_lines_flag || ignore_regexp_list)
mark_ignorable (script);
else
{
@@ -84,8 +83,8 @@ print_context_script (script, unidiff_flag)
e->ignore = 0;
}
- find_function_last_search = 0;
- find_function_last_match = -1;
+ find_function_last_search = - files[0].prefix_lines;
+ find_function_last_match = find_function_last_search - 1;
if (unidiff_flag)
print_script (script, find_hunk, pr_unidiff_hunk);
@@ -130,8 +129,9 @@ pr_context_hunk (hunk)
int first0, last0, first1, last1, show_from, show_to, i;
struct change *next;
char *prefix;
- char *function;
+ const char *function;
int function_length;
+ FILE *out;
/* Determine range of line numbers involved in each file. */
@@ -142,29 +142,33 @@ pr_context_hunk (hunk)
/* Include a context's width before and after. */
- first0 = max (first0 - context, 0);
- first1 = max (first1 - context, 0);
- last0 = min (last0 + context, files[0].buffered_lines - 1);
- last1 = min (last1 + context, files[1].buffered_lines - 1);
+ i = - files[0].prefix_lines;
+ first0 = max (first0 - context, i);
+ first1 = max (first1 - context, i);
+ last0 = min (last0 + context, files[0].valid_lines - 1);
+ last1 = min (last1 + context, files[1].valid_lines - 1);
/* If desired, find the preceding function definition line in file 0. */
function = 0;
- if (function_regexp)
+ if (function_regexp_list)
find_function (&files[0], first0, &function, &function_length);
+ begin_output ();
+ out = outfile;
+
/* If we looked for and found a function this is part of,
include its name in the header of the diff section. */
- fprintf (outfile, "***************");
+ fprintf (out, "***************");
if (function)
{
- fprintf (outfile, " ");
- fwrite (function, 1, min (function_length - 1, 40), outfile);
+ fprintf (out, " ");
+ fwrite (function, 1, min (function_length - 1, 40), out);
}
- fprintf (outfile, "\n*** ");
+ fprintf (out, "\n*** ");
print_context_number_range (&files[0], first0, last0);
- fprintf (outfile, " ****\n");
+ fprintf (out, " ****\n");
if (show_from)
{
@@ -191,9 +195,9 @@ pr_context_hunk (hunk)
}
}
- fprintf (outfile, "--- ");
+ fprintf (out, "--- ");
print_context_number_range (&files[1], first1, last1);
- fprintf (outfile, " ----\n");
+ fprintf (out, " ----\n");
if (show_to)
{
@@ -258,9 +262,9 @@ pr_unidiff_hunk (hunk)
{
int first0, last0, first1, last1, show_from, show_to, i, j, k;
struct change *next;
- int lastline;
char *function;
int function_length;
+ FILE *out;
/* Determine range of line numbers involved in each file. */
@@ -271,31 +275,35 @@ pr_unidiff_hunk (hunk)
/* Include a context's width before and after. */
- first0 = max (first0 - context, 0);
- first1 = max (first1 - context, 0);
- last0 = min (last0 + context, files[0].buffered_lines - 1);
- last1 = min (last1 + context, files[1].buffered_lines - 1);
+ i = - files[0].prefix_lines;
+ first0 = max (first0 - context, i);
+ first1 = max (first1 - context, i);
+ last0 = min (last0 + context, files[0].valid_lines - 1);
+ last1 = min (last1 + context, files[1].valid_lines - 1);
/* If desired, find the preceding function definition line in file 0. */
function = 0;
- if (function_regexp)
+ if (function_regexp_list)
find_function (&files[0], first0, &function, &function_length);
- /* If we looked for and found a function this is part of,
- include its name in the header of the diff section. */
+ begin_output ();
+ out = outfile;
- fprintf (outfile, "@@ -");
+ fprintf (out, "@@ -");
print_unidiff_number_range (&files[0], first0, last0);
- fprintf (outfile, " +");
+ fprintf (out, " +");
print_unidiff_number_range (&files[1], first1, last1);
- fprintf (outfile, " @@");
+ fprintf (out, " @@");
+
+ /* If we looked for and found a function this is part of,
+ include its name in the header of the diff section. */
if (function)
{
- putc (' ', outfile);
- fwrite (function, 1, min (function_length - 1, 40), outfile);
+ putc (' ', out);
+ fwrite (function, 1, min (function_length - 1, 40), out);
}
- putc ('\n', outfile);
+ putc ('\n', out);
next = hunk;
i = first0;
@@ -308,7 +316,7 @@ pr_unidiff_hunk (hunk)
if (!next || i < next->line0)
{
- putc (' ', outfile);
+ putc (tab_align_flag ? '\t' : ' ', out);
print_1_line ((char *)0, &files[0].linbuf[i++]);
j++;
}
@@ -319,7 +327,9 @@ pr_unidiff_hunk (hunk)
k = next->deleted;
while (k--)
{
- putc ('-', outfile);
+ putc ('-', out);
+ if (tab_align_flag)
+ putc ('\t', out);
print_1_line ((char *)0, &files[0].linbuf[i++]);
}
@@ -328,7 +338,9 @@ pr_unidiff_hunk (hunk)
k = next->inserted;
while (k--)
{
- putc ('+', outfile);
+ putc ('+', out);
+ if (tab_align_flag)
+ putc ('\t', out);
print_1_line ((char *)0, &files[1].linbuf[j++]);
}
@@ -339,7 +351,7 @@ pr_unidiff_hunk (hunk)
}
}
-/* Scan a (forward-ordered) edit script for the first place that at least
+/* Scan a (forward-ordered) edit script for the first place that more than
2*CONTEXT unchanged lines appear, and return a pointer
to the `struct change' for the last change before those lines. */
@@ -353,7 +365,7 @@ find_hunk (start)
do
{
- /* Computer number of first line in each file beyond this changed. */
+ /* Compute number of first line in each file beyond this changed. */
top0 = start->line0 + start->deleted;
top1 = start->line1 + start->inserted;
prev = start;
@@ -362,7 +374,7 @@ find_hunk (start)
but only CONTEXT if one is ignorable. */
thresh = ((prev->ignore || (start && start->ignore))
? context
- : 2 * context);
+ : 2 * context + 1);
/* It is not supposed to matter which file we check in the end-test.
If it would matter, crash. */
if (start && start->line0 - top0 != start->line1 - top1)
@@ -414,7 +426,7 @@ static void
find_function (file, linenum, linep, lenp)
struct file_data *file;
int linenum;
- char **linep;
+ const char **linep;
int *lenp;
{
int i = linenum;
@@ -424,26 +436,26 @@ find_function (file, linenum, linep, lenp)
while (--i >= last)
{
/* See if this line is what we want. */
-
- if (0 <= re_search (&function_regexp_compiled,
- file->linbuf[i].text,
- file->linbuf[i].length,
- 0, file->linbuf[i].length,
- 0))
- {
- *linep = file->linbuf[i].text;
- *lenp = file->linbuf[i].length;
- find_function_last_match = i;
- return;
- }
+ struct regexp_list *r;
+ const char *line = file->linbuf[i];
+ int len = file->linbuf[i + 1] - line;
+
+ for (r = function_regexp_list; r; r = r->next)
+ if (0 <= re_search (&r->buf, line, len, 0, len, 0))
+ {
+ *linep = line;
+ *lenp = len;
+ find_function_last_match = i;
+ return;
+ }
}
/* If we search back to where we started searching the previous time,
find the line we found last time. */
- if (find_function_last_match >= 0)
+ if (find_function_last_match >= - file->prefix_lines)
{
i = find_function_last_match;
- *linep = file->linbuf[i].text;
- *lenp = file->linbuf[i].length;
+ *linep = file->linbuf[i];
+ *lenp = file->linbuf[i + 1] - *linep;
return;
}
return;
diff --git a/gnu/usr.bin/diff/diff.c b/gnu/usr.bin/diff/diff.c
index 5d4dffdca39..e4eb66f82dd 100644
--- a/gnu/usr.bin/diff/diff.c
+++ b/gnu/usr.bin/diff/diff.c
@@ -1,11 +1,11 @@
/* GNU DIFF main routine.
- Copyright (C) 1988, 1989 Free Software Foundation, Inc.
+ Copyright (C) 1988, 1989, 1992 Free Software Foundation, Inc.
This file is part of GNU DIFF.
GNU DIFF 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 1, or (at your option)
+the Free Software Foundation; either version 2, or (at your option)
any later version.
GNU DIFF is distributed in the hope that it will be useful,
@@ -18,18 +18,34 @@ along with GNU DIFF; see the file COPYING. If not, write to
the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */
/* GNU DIFF was written by Mike Haertel, David Hayes,
- Richard Stallman and Len Tower. */
+ Richard Stallman, Len Tower, and Paul Eggert. */
#define GDIFF_MAIN
-#include "regex.h"
#include "diff.h"
#include "getopt.h"
+#include "fnmatch.h"
+#ifndef DEFAULT_WIDTH
+#define DEFAULT_WIDTH 130
+#endif
+
+#ifndef GUTTER_WIDTH_MINIMUM
+#define GUTTER_WIDTH_MINIMUM 3
+#endif
+
+int diff_dirs ();
+int diff_2_files ();
+
+static int compare_files ();
+static int specify_format ();
+static void add_regexp();
+static void specify_style ();
+static void usage ();
/* Nonzero for -r: if comparing two directories,
compare their common subdirectories recursively. */
-int recursive;
+static int recursive;
/* For debugging: don't do discard_confusing_lines. */
@@ -67,45 +83,134 @@ option_list (optionvec, count)
return result;
}
-/* The numbers 129 and 130 that appear in the fourth element
- for the context and unidiff entries are used as a way of
- telling the big switch in `main' how to process those options. */
+/* Convert STR to a positive integer, storing the result in *OUT.
+ If STR is not a valid integer, return -1 (otherwise 0). */
+static int
+ck_atoi (str, out)
+ char *str;
+ int *out;
+{
+ char *p;
+ for (p = str; *p; p++)
+ if (*p < '0' || *p > '9')
+ return -1;
+
+ *out = atoi (optarg);
+ return 0;
+}
+
+/* Keep track of excluded file name patterns. */
+
+static const char **exclude;
+static int exclude_alloc, exclude_count;
+
+int
+excluded_filename (f)
+ const char *f;
+{
+ int i;
+ for (i = 0; i < exclude_count; i++)
+ if (fnmatch (exclude[i], f, 0) == 0)
+ return 1;
+ return 0;
+}
+
+static void
+add_exclude (pattern)
+ const char *pattern;
+{
+ if (exclude_alloc <= exclude_count)
+ exclude = (const char **)
+ (exclude_alloc == 0
+ ? xmalloc ((exclude_alloc = 64) * sizeof (*exclude))
+ : xrealloc (exclude, (exclude_alloc *= 2) * sizeof (*exclude)));
+
+ exclude[exclude_count++] = pattern;
+}
+
+static int
+add_exclude_file (name)
+ const char *name;
+{
+ struct file_data f;
+ char *p, *q, *lim;
+
+ f.name = optarg;
+ f.desc = strcmp (optarg, "-") == 0 ? 0 : open (optarg, O_RDONLY, 0);
+ if (f.desc < 0 || fstat (f.desc, &f.stat) != 0)
+ return -1;
+
+ sip (&f, 1);
+ slurp (&f);
+
+ for (p = f.buffer, lim = p + f.buffered_chars; p < lim; p = q)
+ {
+ q = memchr (p, '\n', lim - p);
+ if (!q)
+ q = lim;
+ *q++ = 0;
+ add_exclude (p);
+ }
+
+ return close (f.desc);
+}
+
+/* The numbers 129- that appear in the fourth element of some entries
+ tell the big switch in `main' how to process those options. */
static struct option longopts[] =
{
{"ignore-blank-lines", 0, 0, 'B'},
- {"context", 2, 0, 129},
+ {"context", 2, 0, 'C'},
{"ifdef", 1, 0, 'D'},
{"show-function-line", 1, 0, 'F'},
{"speed-large-files", 0, 0, 'H'},
{"ignore-matching-lines", 1, 0, 'I'},
- {"file-label", 1, 0, 'L'},
- {"entire-new-files", 0, 0, 'N'},
- {"new-files", 0, 0, 'N'},
+ {"label", 1, 0, 'L'},
+ {"file-label", 1, 0, 'L'}, /* An alias, no longer recommended */
+ {"new-file", 0, 0, 'N'},
+ {"entire-new-file", 0, 0, 'N'}, /* An alias, no longer recommended */
+ {"unidirectional-new-file", 0, 0, 'P'},
{"starting-file", 1, 0, 'S'},
{"initial-tab", 0, 0, 'T'},
+ {"width", 1, 0, 'W'},
{"text", 0, 0, 'a'},
- {"all-text", 0, 0, 'a'},
- {"ascii", 0, 0, 'a'},
+ {"ascii", 0, 0, 'a'}, /* An alias, no longer recommended */
{"ignore-space-change", 0, 0, 'b'},
{"minimal", 0, 0, 'd'},
{"ed", 0, 0, 'e'},
- {"reversed-ed", 0, 0, 'f'},
+ {"forward-ed", 0, 0, 'f'},
{"ignore-case", 0, 0, 'i'},
- {"print", 0, 0, 'l'},
+ {"paginate", 0, 0, 'l'},
+ {"print", 0, 0, 'l'}, /* An alias, no longer recommended */
{"rcs", 0, 0, 'n'},
{"show-c-function", 0, 0, 'p'},
- {"binary", 0, 0, 'q'},
+ {"binary", 0, 0, 'q'}, /* An alias, no longer recommended */
{"brief", 0, 0, 'q'},
{"recursive", 0, 0, 'r'},
{"report-identical-files", 0, 0, 's'},
{"expand-tabs", 0, 0, 't'},
- {"ignore-all-space", 0, 0, 'w'},
- {"unified", 2, 0, 130},
{"version", 0, 0, 'v'},
+ {"ignore-all-space", 0, 0, 'w'},
+ {"exclude", 1, 0, 'x'},
+ {"exclude-from", 1, 0, 'X'},
+ {"side-by-side", 0, 0, 'y'},
+ {"unified", 2, 0, 'U'},
+ {"left-column", 0, 0, 129},
+ {"suppress-common-lines", 0, 0, 130},
+ {"sdiff-merge-assist", 0, 0, 131},
+ {"old-line-format", 1, 0, 132},
+ {"new-line-format", 1, 0, 133},
+ {"unchanged-line-format", 1, 0, 134},
+ {"old-group-format", 1, 0, 135},
+ {"new-group-format", 1, 0, 136},
+ {"unchanged-group-format", 1, 0, 137},
+ {"changed-group-format", 1, 0, 138},
+ {"horizon-lines", 1, 0, 139},
{0, 0, 0, 0}
};
+int
main (argc, argv)
int argc;
char *argv[];
@@ -113,8 +218,8 @@ main (argc, argv)
int val;
int c;
int prev = -1;
- int longind;
extern char *version_string;
+ int width = DEFAULT_WIDTH;
program = argv[0];
@@ -126,10 +231,11 @@ main (argc, argv)
length_varies = FALSE;
ignore_case_flag = FALSE;
ignore_blank_lines_flag = FALSE;
- ignore_regexp = 0;
- function_regexp = 0;
+ ignore_regexp_list = NULL;
+ function_regexp_list = NULL;
print_file_same_flag = FALSE;
entire_new_file_flag = FALSE;
+ unidirectional_new_file_flag = FALSE;
no_details_flag = FALSE;
context = -1;
line_end_char = '\n';
@@ -137,7 +243,6 @@ main (argc, argv)
tab_expand_flag = FALSE;
recursive = FALSE;
paginate_flag = FALSE;
- ifdef_string = NULL;
heuristic = FALSE;
dir_start_file = NULL;
msg_chain = NULL;
@@ -147,11 +252,9 @@ main (argc, argv)
/* Decode the options. */
while ((c = getopt_long (argc, argv,
- "0123456789abBcC:dD:efF:hHiI:lL:nNpqrsS:tTuvw",
- longopts, &longind)) != EOF)
+ "0123456789abBcC:dD:efF:hHiI:lL:nNpPqrsS:tTuU:vwW:x:X:y",
+ longopts, (int *)0)) != EOF)
{
- if (c == 0) /* Long option. */
- c = longopts[longind].val;
switch (c)
{
/* All digits combine in decimal to specify the context-size. */
@@ -192,26 +295,21 @@ main (argc, argv)
ignore_blank_lines_flag = 1;
break;
- case 'C':
- case 129: /* +context[=lines] */
- case 130: /* +unified[=lines] */
+ case 'C': /* +context[=lines] */
+ case 'U': /* +unified[=lines] */
if (optarg)
{
if (context >= 0)
fatal ("context length specified twice");
- {
- char *p;
- for (p = optarg; *p; p++)
- if (*p < '0' || *p > '9')
- fatal ("invalid context length argument");
- }
- context = atoi (optarg);
+
+ if (ck_atoi (optarg, &context))
+ fatal ("invalid context length argument");
}
/* Falls through. */
case 'c':
/* Make context-style output. */
- specify_style (c == 130 ? OUTPUT_UNIFIED : OUTPUT_CONTEXT);
+ specify_style (c == 'U' ? OUTPUT_UNIFIED : OUTPUT_CONTEXT);
break;
case 'd':
@@ -223,7 +321,25 @@ main (argc, argv)
case 'D':
/* Make merged #ifdef output. */
specify_style (OUTPUT_IFDEF);
- ifdef_string = optarg;
+ {
+ int i, err = 0;
+ static const char C_ifdef_group_formats[] =
+ "#ifndef %s\n%%<#endif /* not %s */\n%c#ifdef %s\n%%>#endif /* %s */\n%c%%=%c#ifndef %s\n%%<#else /* %s */\n%%>#endif /* %s */\n";
+ char *b = xmalloc (sizeof (C_ifdef_group_formats)
+ + 7 * strlen(optarg) - 14 /* 7*"%s" */
+ - 8 /* 5*"%%" + 3*"%c" */);
+ sprintf (b, C_ifdef_group_formats,
+ optarg, optarg, 0,
+ optarg, optarg, 0, 0,
+ optarg, optarg, optarg);
+ for (i = 0; i < 4; i++)
+ {
+ err |= specify_format (&group_format[i], b);
+ b += strlen (b) + 1;
+ }
+ if (err)
+ error ("conflicting #ifdef formats", 0, 0);
+ }
break;
case 'e':
@@ -241,7 +357,7 @@ main (argc, argv)
/* Show, for each set of changes, the previous line that
matches the specified regexp. Currently affects only
context-style output. */
- function_regexp = optarg;
+ add_regexp (&function_regexp_list, optarg);
break;
case 'h':
@@ -265,7 +381,7 @@ main (argc, argv)
case 'I':
/* Ignore changes affecting only lines that match the
specified regexp. */
- ignore_regexp = optarg;
+ add_regexp (&ignore_regexp_list, optarg);
break;
case 'l':
@@ -282,7 +398,7 @@ main (argc, argv)
else
fatal ("too many file label options");
break;
-
+
case 'n':
/* Output RCS-style diffs, like `-f' except that each command
specifies the number of lines affected. */
@@ -298,7 +414,14 @@ main (argc, argv)
case 'p':
/* Make context-style output and show name of last C function. */
specify_style (OUTPUT_CONTEXT);
- function_regexp = "^[_a-zA-Z]";
+ add_regexp (&function_regexp_list, "^[_a-zA-Z$]");
+ break;
+
+ case 'P':
+ /* When comparing directories, if a file appears only in
+ the second directory of the two,
+ treat it as present but empty in the other. */
+ unidirectional_new_file_flag = 1;
break;
case 'q':
@@ -335,21 +458,85 @@ main (argc, argv)
tab_align_flag = 1;
break;
- case 'v':
- printf ("GNU diff version %s\n", version_string);
- break;
-
case 'u':
/* Output the context diff in unidiff format. */
specify_style (OUTPUT_UNIFIED);
break;
+ case 'v':
+ fprintf (stderr, "GNU diff version %s\n", version_string);
+ break;
+
case 'w':
/* Ignore horizontal whitespace when comparing lines. */
ignore_all_space_flag = 1;
length_varies = 1;
break;
+ case 'x':
+ add_exclude (optarg);
+ break;
+
+ case 'X':
+ if (add_exclude_file (optarg) != 0)
+ pfatal_with_name (optarg);
+ break;
+
+ case 'y':
+ /* Use side-by-side (sdiff-style) columnar output. */
+ specify_style (OUTPUT_SDIFF);
+ break;
+
+ case 'W':
+ /* Set the line width for OUTPUT_SDIFF. */
+ if (ck_atoi (optarg, &width) || width <= 0)
+ fatal ("column width must be a positive integer");
+ break;
+
+ case 129:
+ sdiff_left_only = 1;
+ break;
+
+ case 130:
+ sdiff_skip_common_lines = 1;
+ break;
+
+ case 131:
+ /* sdiff-style columns output. */
+ specify_style (OUTPUT_SDIFF);
+ sdiff_help_sdiff = 1;
+ break;
+
+ case 132:
+ case 133:
+ case 134:
+ specify_style (OUTPUT_IFDEF);
+ {
+ const char **form = &line_format[c - 132];
+ if (*form && strcmp (*form, optarg) != 0)
+ error ("conflicting line format", 0, 0);
+ *form = optarg;
+ }
+ break;
+
+ case 135:
+ case 136:
+ case 137:
+ case 138:
+ specify_style (OUTPUT_IFDEF);
+ {
+ const char **form = &group_format[c - 135];
+ if (*form && strcmp (*form, optarg) != 0)
+ error ("conflicting group format", 0, 0);
+ *form = optarg;
+ }
+ break;
+
+ case 139:
+ if (ck_atoi (optarg, &horizon_lines) || horizon_lines < 0)
+ fatal ("horizon must be a nonnegative integer");
+ break;
+
default:
usage ();
}
@@ -359,37 +546,59 @@ main (argc, argv)
if (optind != argc - 2)
usage ();
- if (ignore_regexp)
- {
- char *val;
- bzero (&ignore_regexp_compiled, sizeof ignore_regexp_compiled);
- val = re_compile_pattern (ignore_regexp, strlen (ignore_regexp),
- &ignore_regexp_compiled);
- if (val != 0)
- error ("%s: %s", ignore_regexp, val);
- ignore_regexp_compiled.fastmap = (char *) xmalloc (256);
- }
- if (function_regexp)
- {
- char *val;
- bzero (&function_regexp_compiled, sizeof function_regexp_compiled);
- val = re_compile_pattern (function_regexp, strlen (function_regexp),
- &function_regexp_compiled);
- if (val != 0)
- error ("%s: %s", function_regexp, val);
- function_regexp_compiled.fastmap = (char *) xmalloc (256);
- }
+ {
+ /*
+ * We maximize first the half line width, and then the gutter width,
+ * according to the following constraints:
+ * 1. Two half lines plus a gutter must fit in a line.
+ * 2. If the half line width is nonzero:
+ * a. The gutter width is at least GUTTER_WIDTH_MINIMUM.
+ * b. If tabs are not expanded to spaces,
+ * a half line plus a gutter is an integral number of tabs,
+ * so that tabs in the right column line up.
+ */
+ int t = tab_expand_flag ? 1 : TAB_WIDTH;
+ int off = (width + t + GUTTER_WIDTH_MINIMUM) / (2*t) * t;
+ sdiff_half_width = max (0, min (off - GUTTER_WIDTH_MINIMUM, width - off)),
+ sdiff_column2_offset = sdiff_half_width ? off : width;
+ }
if (output_style != OUTPUT_CONTEXT && output_style != OUTPUT_UNIFIED)
context = 0;
else if (context == -1)
/* Default amount of context for -c. */
context = 3;
+
+ if (output_style == OUTPUT_IFDEF)
+ {
+ int i;
+ for (i = 0; i < sizeof (line_format) / sizeof (*line_format); i++)
+ if (!line_format[i])
+ line_format[i] = "%l\n";
+ if (!group_format[OLD])
+ group_format[OLD]
+ = group_format[UNCHANGED] ? group_format[UNCHANGED] : "%<";
+ if (!group_format[NEW])
+ group_format[NEW]
+ = group_format[UNCHANGED] ? group_format[UNCHANGED] : "%>";
+ if (!group_format[UNCHANGED])
+ group_format[UNCHANGED] = "%=";
+ if (!group_format[CHANGED])
+ group_format[CHANGED] = concat (group_format[OLD],
+ group_format[NEW], "");
+ }
+
+ no_diff_means_no_output =
+ (output_style == OUTPUT_IFDEF ?
+ (!*group_format[UNCHANGED]
+ || (strcmp (group_format[UNCHANGED], "%=") == 0
+ && !*line_format[UNCHANGED]))
+ : output_style == OUTPUT_SDIFF ? sdiff_skip_common_lines : 1);
switch_string = option_list (argv + 1, optind - 1);
- val = compare_files (0, argv[optind], 0, argv[optind + 1], 0);
+ val = compare_files (NULL, argv[optind], NULL, argv[optind + 1], 0);
/* Print any messages that were saved up for last. */
print_message_queue ();
@@ -397,32 +606,76 @@ main (argc, argv)
if (ferror (stdout) || fclose (stdout) != 0)
fatal ("write error");
exit (val);
+ return val;
}
+/* Add the compiled form of regexp PATTERN to REGLIST. */
+
+static void
+add_regexp (reglist, pattern)
+ struct regexp_list **reglist;
+ char *pattern;
+{
+ struct regexp_list *r;
+ const char *m;
+
+ r = (struct regexp_list *) xmalloc (sizeof (*r));
+ bzero (r, sizeof (*r));
+ r->buf.fastmap = (char *) xmalloc (256);
+ m = re_compile_pattern (pattern, strlen (pattern), &r->buf);
+ if (m != 0)
+ error ("%s: %s", pattern, m);
+
+ /* Add to the start of the list, since it's easier than the end. */
+ r->next = *reglist;
+ *reglist = r;
+}
+
+static void
usage ()
{
+ fprintf (stderr, "Usage: %s [options] from-file to-file\n", program);
+ fprintf (stderr, "Options:\n\
+ [-abBcdefhHilnNpPqrstTuvwy] [-C lines] [-D name] [-F regexp]\n\
+ [-I regexp] [-L from-label [-L to-label]] [-S starting-file] [-U lines]\n\
+ [-W columns] [-x pattern] [-X pattern-file] [--exclude=pattern]\n\
+ [--exclude-from=pattern-file] [--ignore-blank-lines] [--context[=lines]]\n\
+ [--ifdef=name] [--show-function-line=regexp] [--speed-large-files]\n\
+ [--label=from-label [--label=to-label]] [--new-file]\n");
fprintf (stderr, "\
-Usage: diff [-#] [-abBcdefhHilnNprstTuvw] [-C lines] [-F regexp] [-I regexp]\n\
- [-L label [-L label]] [-S file] [-D symbol] [+ignore-blank-lines]\n\
- [+context[=lines]] [+unified[=lines]] [+ifdef=symbol]\n\
- [+show-function-line=regexp]\n");
+ [--ignore-matching-lines=regexp] [--unidirectional-new-file]\n\
+ [--starting-file=starting-file] [--initial-tab] [--width=columns]\n\
+ [--text] [--ignore-space-change] [--minimal] [--ed] [--forward-ed]\n\
+ [--ignore-case] [--paginate] [--rcs] [--show-c-function] [--brief]\n\
+ [--recursive] [--report-identical-files] [--expand-tabs] [--version]\n");
fprintf (stderr, "\
- [+speed-large-files] [+ignore-matching-lines=regexp] [+new-file]\n\
- [+initial-tab] [+starting-file=file] [+text] [+all-text] [+ascii]\n\
- [+minimal] [+ignore-space-change] [+ed] [+reversed-ed] [+ignore-case]\n");
- fprintf (stderr, "\
- [+print] [+rcs] [+show-c-function] [+binary] [+brief] [+recursive]\n\
- [+report-identical-files] [+expand-tabs] [+ignore-all-space]\n\
- [+file-label=label [+file-label=label]] [+version] path1 path2\n");
+ [--ignore-all-space] [--side-by-side] [--unified[=lines]]\n\
+ [--left-column] [--suppress-common-lines] [--sdiff-merge-assist]\n\
+ [--old-line-format=format] [--new-line-format=format]\n\
+ [--unchanged-line-format=format]\n\
+ [--old-group-format=format] [--new-group-format=format]\n\
+ [--unchanged-group-format=format] [--changed-group-format=format]\n\
+ [--horizon-lines=lines]\n");
exit (2);
}
+static int
+specify_format (var, value)
+ const char **var;
+ const char *value;
+{
+ int err = *var ? strcmp (*var, value) : 0;
+ *var = value;
+ return err;
+}
+
+static void
specify_style (style)
enum output_style style;
{
if (output_style != OUTPUT_NORMAL
&& output_style != style)
- error ("conflicting specifications of output style");
+ error ("conflicting specifications of output style", 0, 0);
output_style = style;
}
@@ -431,27 +684,28 @@ specify_style (style)
(if DIR0 is 0, then the name is just NAME0, etc.)
This is self-contained; it opens the files and closes them.
- Value is 0 if files are identical, 1 if different,
+ Value is 0 if files are the same, 1 if different,
2 if there is a problem opening them. */
-int
+static int
compare_files (dir0, name0, dir1, name1, depth)
char *dir0, *dir1;
char *name0, *name1;
int depth;
{
- static char Standard_Input[] = "Standard Input";
struct file_data inf[2];
register int i;
int val;
+ int same_files;
int errorcount = 0;
- int stat_result[2];
/* If this is directory comparison, perhaps we have a file
that exists only in one of the directories.
If so, just print a message to that effect. */
- if (! entire_new_file_flag && (name0 == 0 || name1 == 0))
+ if (! ((name0 != 0 && name1 != 0)
+ || (unidirectional_new_file_flag && name1 != 0)
+ || entire_new_file_flag))
{
char *name = name0 == 0 ? name1 : name0;
char *dir = name0 == 0 ? dir1 : dir0;
@@ -476,88 +730,84 @@ compare_files (dir0, name0, dir1, name1, depth)
inf[0].name = dir0 == 0 ? name0 : concat (dir0, "/", name0);
inf[1].name = dir1 == 0 ? name1 : concat (dir1, "/", name1);
- /* Stat the files. Record whether they are directories.
- Record in stat_result whether stat fails. */
+ /* Stat the files. Record whether they are directories. */
for (i = 0; i <= 1; i++)
{
- bzero (&inf[i].stat, sizeof(struct stat));
+ bzero (&inf[i].stat, sizeof (struct stat));
inf[i].dir_p = 0;
- stat_result[i] = 0;
if (inf[i].desc != -1)
{
- char *filename = inf[i].name;
-
- stat_result[i] =
- strcmp (filename, "-")
- ? stat (filename, &inf[i].stat)
- : fstat (0, &inf[i].stat);
-
- if (stat_result[i] < 0)
+ int stat_result;
+
+ if (strcmp (inf[i].name, "-") == 0)
{
- perror_with_name (filename);
+ inf[i].desc = 0;
+ inf[i].name = "Standard Input";
+ stat_result = fstat (0, &inf[i].stat);
+ }
+ else
+ stat_result = stat (inf[i].name, &inf[i].stat);
+
+ if (stat_result != 0)
+ {
+ perror_with_name (inf[i].name);
errorcount = 1;
}
else
- inf[i].dir_p =
- S_IFDIR == (inf[i].stat.st_mode & S_IFMT)
- && strcmp (filename, "-");
+ inf[i].dir_p = S_ISDIR (inf[i].stat.st_mode) && inf[i].desc != 0;
}
}
- /* See if the two named files are actually the same physical file.
- If so, we know they are identical without actually reading them. */
-
- if (output_style != OUTPUT_IFDEF
- && inf[0].stat.st_ino == inf[1].stat.st_ino
- && inf[0].stat.st_dev == inf[1].stat.st_dev
- && stat_result[0] == 0
- && stat_result[1] == 0)
- {
- val = 0;
- goto done;
- }
-
if (name0 == 0)
inf[0].dir_p = inf[1].dir_p;
if (name1 == 0)
inf[1].dir_p = inf[0].dir_p;
- /* Open the files and record their descriptors. */
-
- for (i = 0; i <= 1; i++)
+ if (errorcount == 0 && depth == 0 && inf[0].dir_p != inf[1].dir_p)
{
- if (inf[i].desc == -1)
- ;
- else if (!strcmp (inf[i].name, "-"))
- {
- inf[i].desc = 0;
- inf[i].name = Standard_Input;
- }
- /* Don't bother opening if stat already failed. */
- else if (stat_result[i] == 0 && ! inf[i].dir_p)
- {
- char *filename = inf[i].name;
+ /* If one is a directory, and it was specified in the command line,
+ use the file in that dir with the other file's basename. */
- inf[i].desc = open (filename, O_RDONLY, 0);
- if (0 > inf[i].desc)
- {
- perror_with_name (filename);
- errorcount = 1;
- }
+ int fnm_arg = inf[0].dir_p;
+ int dir_arg = 1 - fnm_arg;
+ char *p = rindex (inf[fnm_arg].name, '/');
+ char *filename = inf[dir_arg].name
+ = concat (inf[dir_arg].name, "/", (p ? p+1 : inf[fnm_arg].name));
+
+ if (inf[fnm_arg].desc == 0)
+ fatal ("can't compare - to a directory");
+
+ if (stat (filename, &inf[dir_arg].stat) != 0)
+ {
+ perror_with_name (filename);
+ errorcount = 1;
}
+ else
+ inf[dir_arg].dir_p = S_ISDIR (inf[dir_arg].stat.st_mode);
}
if (errorcount)
{
- /* If either file should exist but fails to be opened, return 2. */
+ /* If either file should exist but does not, return 2. */
val = 2;
}
- else if (inf[0].dir_p && inf[1].dir_p)
+ else if ((same_files = inf[0].stat.st_ino == inf[1].stat.st_ino
+ && inf[0].stat.st_dev == inf[1].stat.st_dev
+ && inf[0].desc != -1
+ && inf[1].desc != -1)
+ && no_diff_means_no_output)
+ {
+ /* The two named files are actually the same physical file.
+ We know they are identical without actually reading them. */
+
+ val = 0;
+ }
+ else if (inf[0].dir_p & inf[1].dir_p)
{
if (output_style == OUTPUT_IFDEF)
fatal ("-D option not supported with directories");
@@ -574,66 +824,21 @@ compare_files (dir0, name0, dir1, name1, depth)
}
else
{
- val = diff_dirs (inf[0].name, inf[1].name,
- compare_files, depth, 0, 0);
- }
-
- }
- else if (depth == 0 && (inf[0].dir_p || inf[1].dir_p))
- {
-
- /* If only one is a directory, and it was specified in the command line,
- use the file in that dir whose basename matches the other file. */
-
- int dir_arg = (inf[0].dir_p ? 0 : 1);
- int fnm_arg = (inf[0].dir_p ? 1 : 0);
- char *p = rindex (inf[fnm_arg].name, '/');
- char *filename = concat (inf[dir_arg].name, "/",
- (p ? p+1 : inf[fnm_arg].name));
-
- if (inf[fnm_arg].name == Standard_Input)
- fatal ("can't compare - to a directory");
-
- inf[dir_arg].desc = open (filename, O_RDONLY, 0);
-
- if (0 > inf[dir_arg].desc)
- {
- perror_with_name (filename);
- val = 2;
- }
- else
- {
- /* JF: patch from the net to check and make sure we can really free
- this. If it's from argv[], freeing it is a *really* bad idea */
- if (0 != (dir_arg ? dir1 : dir0))
- free (inf[dir_arg].name);
- inf[dir_arg].name = filename;
- if (fstat (inf[dir_arg].desc, &inf[dir_arg].stat) < 0)
- pfatal_with_name (inf[dir_arg].name);
-
- inf[dir_arg].dir_p
- = (S_IFDIR == (inf[dir_arg].stat.st_mode & S_IFMT));
- if (inf[dir_arg].dir_p)
- {
- error ("%s is a directory but %s is not",
- inf[dir_arg].name, inf[fnm_arg].name);
- val = 1;
- }
- else
- val = diff_2_files (inf, depth);
+ val = diff_dirs (inf, compare_files, depth);
}
}
- else if (depth > 0 && (inf[0].dir_p || inf[1].dir_p))
+ else if (inf[0].dir_p | inf[1].dir_p)
{
/* Perhaps we have a subdirectory that exists only in one directory.
If so, just print a message to that effect. */
if (inf[0].desc == -1 || inf[1].desc == -1)
{
- if (entire_new_file_flag && recursive)
- val = diff_dirs (inf[0].name, inf[1].name, compare_files, depth,
- inf[0].desc == -1, inf[1].desc == -1);
+ if (recursive
+ && (entire_new_file_flag
+ || (unidirectional_new_file_flag && inf[0].desc == -1)))
+ val = diff_dirs (inf, compare_files, depth);
else
{
char *dir = (inf[0].desc == -1) ? dir1 : dir0;
@@ -646,34 +851,64 @@ compare_files (dir0, name0, dir1, name1, depth)
/* We have a subdirectory in one directory
and a file in the other. */
- if (inf[0].dir_p)
- message ("%s is a directory but %s is not\n",
- inf[0].name, inf[1].name);
- else
- message ("%s is a directory but %s is not\n",
- inf[1].name, inf[0].name);
+ message ("%s is a directory but %s is not\n",
+ inf[1 - inf[0].dir_p].name, inf[inf[0].dir_p].name);
+
/* This is a difference. */
val = 1;
}
}
+ else if (no_details_flag
+ && inf[0].stat.st_size != inf[1].stat.st_size
+ && (inf[0].desc == -1 || S_ISREG (inf[0].stat.st_mode))
+ && (inf[1].desc == -1 || S_ISREG (inf[1].stat.st_mode)))
+ {
+ message ("Files %s and %s differ\n", inf[0].name, inf[1].name);
+ val = 1;
+ }
else
{
-
- /* Both exist and both are ordinary files. */
-
- val = diff_2_files (inf, depth);
-
+ /* Both exist and neither is a directory. */
+
+ /* Open the files and record their descriptors. */
+
+ if (inf[0].desc == -2)
+ if ((inf[0].desc = open (inf[0].name, O_RDONLY, 0)) < 0)
+ {
+ perror_with_name (inf[0].name);
+ errorcount = 1;
+ }
+ if (inf[1].desc == -2)
+ if (same_files)
+ inf[1].desc = inf[0].desc;
+ else if ((inf[1].desc = open (inf[1].name, O_RDONLY, 0)) < 0)
+ {
+ perror_with_name (inf[1].name);
+ errorcount = 1;
+ }
+
+ /* Compare the files, if no error was found. */
+
+ val = errorcount ? 2 : diff_2_files (inf, depth);
+
+ /* Close the file descriptors. */
+
+ if (inf[0].desc >= 0 && close (inf[0].desc) != 0)
+ {
+ perror_with_name (inf[0].name);
+ val = 2;
+ }
+ if (inf[1].desc >= 0 && inf[0].desc != inf[1].desc
+ && close (inf[1].desc) != 0)
+ {
+ perror_with_name (inf[1].name);
+ val = 2;
+ }
}
/* Now the comparison has been done, if no error prevented it,
and VAL is the value this function will return. */
- if (inf[0].desc >= 0)
- close (inf[0].desc);
- if (inf[1].desc >= 0)
- close (inf[1].desc);
-
- done:
if (val == 0 && !inf[0].dir_p)
{
if (print_file_same_flag)
diff --git a/gnu/usr.bin/diff/diff.h b/gnu/usr.bin/diff/diff.h
index 059522444ee..74f025d9f50 100644
--- a/gnu/usr.bin/diff/diff.h
+++ b/gnu/usr.bin/diff/diff.h
@@ -1,11 +1,11 @@
/* Shared definitions for GNU DIFF
- Copyright (C) 1988, 1989 Free Software Foundation, Inc.
+ Copyright (C) 1988, 89, 91, 92 Free Software Foundation, Inc.
This file is part of GNU DIFF.
GNU DIFF 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 1, or (at your option)
+the Free Software Foundation; either version 2, or (at your option)
any later version.
GNU DIFF is distributed in the hope that it will be useful,
@@ -17,86 +17,23 @@ You should have received a copy of the GNU General Public License
along with GNU DIFF; see the file COPYING. If not, write to
the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */
-
#include <ctype.h>
#include <stdio.h>
-#include <sys/types.h>
-#include <sys/stat.h>
-
-#ifdef USG
-#include <time.h>
-#ifdef HAVE_NDIR
-#ifdef NDIR_IN_SYS
-#include <sys/ndir.h>
-#else
-#include <ndir.h>
-#endif /* not NDIR_IN_SYS */
-#else
-#include <dirent.h>
-#endif /* not HAVE_NDIR */
-#include <fcntl.h>
-#ifndef HAVE_DIRECT
-#define direct dirent
-#endif
-#else /* not USG */
-#include <sys/time.h>
-#include <sys/dir.h>
-#include <sys/file.h>
-#endif
-
-#ifdef USG
-/* Define needed BSD functions in terms of sysV library. */
-
-#define bcopy(s,d,n) memcpy((d),(s),(n))
-#define bcmp(s1,s2,n) memcmp((s1),(s2),(n))
-#define bzero(s,n) memset((s),0,(n))
-
-#ifndef XENIX
-#define dup2(f,t) (close(t),fcntl((f),F_DUPFD,(t)))
-#endif
-
-#define vfork fork
-#define index strchr
-#define rindex strrchr
-#endif
-
-#ifdef sparc
-/* vfork clobbers registers on the Sparc, so don't use it. */
-#define vfork fork
-#endif
-
-#include <errno.h>
-extern int errno;
-extern int sys_nerr;
-extern char *sys_errlist[];
-
-#define EOS (0)
-#define FALSE (0)
-#define TRUE 1
-
-#define min(a,b) ((a) <= (b) ? (a) : (b))
-#define max(a,b) ((a) >= (b) ? (a) : (b))
+#include "system.h"
+#include "regex.h"
#ifndef PR_FILE_NAME
#define PR_FILE_NAME "/bin/pr"
#endif
-/* Support old-fashioned C compilers. */
-#if defined (__STDC__) || defined (__GNUC__)
-#include "limits.h"
-#else
-#define INT_MAX 2147483647
-#define CHAR_BIT 8
-#endif
+/* Character classes. */
+extern const char textchar[];
-/* Support old-fashioned C compilers. */
-#if !defined (__STDC__) && !defined (__GNUC__)
-#define const
-#endif
+/* Is_space is a little broader than ctype.h's isspace,
+ because it also includes backspace and no-break space. */
+#define Is_space(c) (textchar[c] & 2)
-#ifndef O_RDONLY
-#define O_RDONLY 0
-#endif
+#define TAB_WIDTH 8
/* Variables for command line options */
@@ -120,16 +57,20 @@ enum output_style {
/* Like -f, but output a count of changed lines in each "command" (-n). */
OUTPUT_RCS,
/* Output merged #ifdef'd file (-D). */
- OUTPUT_IFDEF };
+ OUTPUT_IFDEF,
+ /* Output sdiff style (-y). */
+ OUTPUT_SDIFF
+};
/* True for output styles that are robust,
i.e. can handle a file that ends in a non-newline. */
-#define ROBUST_OUTPUT_STYLE(S) \
- ((S) == OUTPUT_CONTEXT || (S) == OUTPUT_UNIFIED || (S) == OUTPUT_RCS \
- || (S) == OUTPUT_NORMAL)
+#define ROBUST_OUTPUT_STYLE(S) ((S) != OUTPUT_ED && (S) != OUTPUT_FORWARD_ED)
EXTERN enum output_style output_style;
+/* Nonzero if output cannot be generated for identical files. */
+EXTERN int no_diff_means_no_output;
+
/* Number of lines of context to show in each set of diffs.
This is zero when context is not to be shown. */
EXTERN int context;
@@ -138,6 +79,9 @@ EXTERN int context;
Don't interpret codes over 0177 as implying a "binary file". */
EXTERN int always_text_flag;
+/* Number of lines to keep in identical prefix and suffix. */
+EXTERN int horizon_lines;
+
/* Ignore changes in horizontal whitespace (-b). */
EXTERN int ignore_space_change_flag;
@@ -147,12 +91,6 @@ EXTERN int ignore_all_space_flag;
/* Ignore changes that affect only blank lines (-B). */
EXTERN int ignore_blank_lines_flag;
-/* Ignore changes that affect only lines matching this regexp (-I). */
-EXTERN char *ignore_regexp;
-
-/* Result of regex-compilation of `ignore_regexp'. */
-EXTERN struct re_pattern_buffer ignore_regexp_compiled;
-
/* 1 if lines may match even if their lengths are different.
This depends on various options. */
EXTERN int length_varies;
@@ -163,11 +101,17 @@ EXTERN int ignore_case_flag;
/* File labels for `-c' output headers (-L). */
EXTERN char *file_label[2];
+struct regexp_list
+{
+ struct re_pattern_buffer buf;
+ struct regexp_list *next;
+};
+
/* Regexp to identify function-header lines (-F). */
-EXTERN char *function_regexp;
+EXTERN struct regexp_list *function_regexp_list;
-/* Result of regex-compilation of `function_regexp'. */
-EXTERN struct re_pattern_buffer function_regexp_compiled;
+/* Ignore changes that affect only lines matching this regexp (-I). */
+EXTERN struct regexp_list *ignore_regexp_list;
/* Say only whether files differ, not how (-q). */
EXTERN int no_details_flag;
@@ -196,11 +140,43 @@ EXTERN char *dir_start_file;
Then `patch' would create the file with appropriate contents. */
EXTERN int entire_new_file_flag;
+/* If a file is new (appears in only the second dir)
+ include its entire contents (-P).
+ Then `patch' would create the file with appropriate contents. */
+EXTERN int unidirectional_new_file_flag;
+
/* Pipe each file's output through pr (-l). */
EXTERN int paginate_flag;
-/* String to use for #ifdef (-D). */
-EXTERN char * ifdef_string;
+enum line_class {
+ /* Lines taken from just the first file. */
+ OLD,
+ /* Lines taken from just the second file. */
+ NEW,
+ /* Lines common to both files. */
+ UNCHANGED,
+ /* A hunk containing both old and new lines (line groups only). */
+ CHANGED
+};
+
+/* Line group formats for old, new, unchanged, and changed groups. */
+EXTERN const char *group_format[CHANGED + 1];
+
+/* Line formats for old, new, and unchanged lines. */
+EXTERN const char *line_format[UNCHANGED + 1];
+
+/* If using OUTPUT_SDIFF print extra information to help the sdiff filter. */
+EXTERN int sdiff_help_sdiff;
+
+/* Tell OUTPUT_SDIFF to show only the left version of common lines. */
+EXTERN int sdiff_left_only;
+
+/* Tell OUTPUT_SDIFF to not show common lines. */
+EXTERN int sdiff_skip_common_lines;
+
+/* The half line width and column 2 offset for OUTPUT_SDIFF. */
+EXTERN unsigned sdiff_half_width;
+EXTERN unsigned sdiff_column2_offset;
/* String containing all the command options diff received,
with spaces between and at the beginning but none at the end.
@@ -236,21 +212,13 @@ struct change
/* Structures that describe the input files. */
-/* Data on one line of text. */
-
-struct line_def {
- char *text;
- int length;
- unsigned hash;
-};
-
/* Data on one input file being compared. */
struct file_data {
int desc; /* File descriptor */
char *name; /* File name */
struct stat stat; /* File status from fstat() */
- int dir_p; /* 1 if file is a directory */
+ int dir_p; /* nonzero if file is a directory */
/* Buffer in which text of file is read. */
char * buffer;
@@ -259,26 +227,24 @@ struct file_data {
/* Number of valid characters now in the buffer. */
int buffered_chars;
- /* Array of data on analyzed lines of this chunk of this file. */
- struct line_def *linbuf;
+ /* Array of pointers to lines in the file. */
+ const char **linbuf;
- /* Allocated size of linbuf array (# of elements). */
- int linbufsize;
-
- /* Number of elements of linbuf containing valid data. */
- int buffered_lines;
+ /* linbuf_base <= buffered_lines <= valid_lines <= alloc_lines.
+ linebuf[linbuf_base ... buffered_lines - 1] are possibly differing.
+ linebuf[linbuf_base ... valid_lines - 1] contain valid data.
+ linebuf[linbuf_base ... alloc_lines - 1] are allocated. */
+ int linbuf_base, buffered_lines, valid_lines, alloc_lines;
/* Pointer to end of prefix of this file to ignore when hashing. */
- char *prefix_end;
+ const char *prefix_end;
- /* Count of lines in the prefix. */
+ /* Count of lines in the prefix.
+ There are this many lines in the file before linbuf[0]. */
int prefix_lines;
/* Pointer to start of suffix of this file to ignore when hashing. */
- char *suffix_begin;
-
- /* Count of lines in the suffix. */
- int suffix_lines;
+ const char *suffix_begin;
/* Vector, indexed by line number, containing an equivalence code for
each line. It is this vector that is actually compared with that
@@ -307,18 +273,11 @@ struct file_data {
/* 1 more than the maximum equivalence value used for this or its
sibling file. */
int equiv_max;
-
- /* Table translating diff's internal line numbers
- to actual line numbers in the file.
- This is needed only when some lines have been discarded.
- The allocated size is always linbufsize
- and the number of valid elements is buffered_lines. */
- int *ltran;
};
/* Describe the two files currently being compared. */
-struct file_data files[2];
+EXTERN struct file_data files[2];
/* Queue up one-line messages to be printed at the end,
when -l is specified. Each message is recorded with a `struct msg'. */
@@ -345,7 +304,7 @@ EXTERN FILE *outfile;
/* Declare various functions. */
-#ifdef __STDC__
+#if __STDC__
#define VOID void
#else
#define VOID char
@@ -353,18 +312,24 @@ EXTERN FILE *outfile;
VOID *xmalloc ();
VOID *xrealloc ();
char *concat ();
-void free ();
-char *rindex ();
-char *index ();
+
+int excluded_filename ();
+int sip ();
+
+struct change *find_change ();
void analyze_hunk ();
+void begin_output ();
void error ();
void fatal ();
void message ();
+void output_1_line ();
void perror_with_name ();
void pfatal_with_name ();
void print_1_line ();
+void print_context_header ();
void print_message_queue ();
void print_number_range ();
void print_script ();
+void slurp ();
void translate_range ();
diff --git a/gnu/usr.bin/diff/dir.c b/gnu/usr.bin/diff/dir.c
index 97b3bdf891c..a5e3e2cbe9f 100644
--- a/gnu/usr.bin/diff/dir.c
+++ b/gnu/usr.bin/diff/dir.c
@@ -1,11 +1,11 @@
/* Read, sort and compare two directories. Used for GNU DIFF.
- Copyright (C) 1988, 1989 Free Software Foundation, Inc.
+ Copyright (C) 1988, 1989, 1992 Free Software Foundation, Inc.
This file is part of GNU DIFF.
GNU DIFF 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 1, or (at your option)
+the Free Software Foundation; either version 2, or (at your option)
any later version.
GNU DIFF is distributed in the hope that it will be useful,
@@ -21,87 +21,103 @@ the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */
static int compare_names ();
-/* Read the directory named DIRNAME and return a sorted vector
- of filenames for its contents. NONEX nonzero means this directory is
- known to be nonexistent, so return zero files. */
+/* Read the directory named by DIR and store into DIRDATA a sorted vector
+ of filenames for its contents. DIR->desc == -1 means this directory is
+ known to be nonexistent, so set DIRDATA to an empty vector.
+ Return -1 (setting errno) if error, 0 otherwise. */
struct dirdata
{
- int length; /* # elements in `files' */
- char **files; /* Sorted names of files in the dir */
+ char **files; /* Sorted names of files in dir, terminated by (char *) 0. */
+ char *data; /* Allocated storage for file names. */
};
-static struct dirdata
-dir_sort (dirname, nonex)
- char *dirname;
- int nonex;
+static int
+dir_sort (dir, dirdata)
+ struct file_data *dir;
+ struct dirdata *dirdata;
{
- register DIR *reading;
register struct direct *next;
- struct dirdata dirdata;
+ register int i;
/* Address of block containing the files that are described. */
char **files;
- /* Length of block that `files' points to, measured in files. */
+ /* Number of files in directory. */
int nfiles;
- /* Index of first unused in `files'. */
- int files_index;
+ /* Allocated and used storage for file name data. */
+ char *data;
+ size_t data_alloc, data_used;
- if (nonex)
- {
- dirdata.length = 0;
- dirdata.files = 0;
- return dirdata;
- }
+ dirdata->files = 0;
+ dirdata->data = 0;
+ nfiles = 0;
- /* Open the directory and check for errors. */
- reading = opendir (dirname);
- if (!reading)
+ if (dir->desc != -1)
{
- perror_with_name (dirname);
- dirdata.length = -1;
- return dirdata;
- }
+ /* Open the directory and check for errors. */
+ register DIR *reading = opendir (dir->name);
+ if (!reading)
+ return -1;
- /* Initialize the table of filenames. */
+ /* Initialize the table of filenames. */
- nfiles = 100;
- files = (char **) xmalloc (nfiles * sizeof (char *));
- files_index = 0;
+ data_alloc = max (1, (size_t) dir->stat.st_size);
+ data_used = 0;
+ dirdata->data = data = (char *) xmalloc (data_alloc);
- /* Read the directory entries, and insert the subfiles
- into the `files' table. */
+ /* Read the directory entries, and insert the subfiles
+ into the `data' table. */
- while (next = readdir (reading))
- {
- /* Ignore the files `.' and `..' */
- if (next->d_name[0] == '.'
- && (next->d_name[1] == 0
- || (next->d_name[1] == '.'
- && next->d_name[2] == 0)))
- continue;
-
- if (files_index == nfiles)
+ while ((errno = 0, (next = readdir (reading)) != 0))
+ {
+ char *d_name = next->d_name;
+ size_t d_size;
+
+ /* Ignore the files `.' and `..' */
+ if (d_name[0] == '.'
+ && (d_name[1] == 0 || (d_name[1] == '.' && d_name[2] == 0)))
+ continue;
+
+ if (excluded_filename (d_name))
+ continue;
+
+ d_size = strlen (d_name) + 1;
+ while (data_alloc < data_used + d_size)
+ dirdata->data = data = (char *) xrealloc (data, data_alloc *= 2);
+ bcopy (d_name, data + data_used, d_size);
+ data_used += d_size;
+ nfiles++;
+ }
+ if (errno)
{
- nfiles *= 2;
- files
- = (char **) xrealloc (files, sizeof (char *) * nfiles);
+ int e = errno;
+ closedir (reading);
+ errno = e;
+ return -1;
}
- files[files_index++] = concat (next->d_name, "", "");
+#ifdef VOID_CLOSEDIR
+ closedir (reading);
+#else
+ if (closedir (reading) != 0)
+ return -1;
+#endif
}
- closedir (reading);
+ /* Create the `files' table from the `data' table. */
+ dirdata->files = files = (char **) xmalloc (sizeof (char *) * (nfiles + 1));
+ for (i = 0; i < nfiles; i++)
+ {
+ files[i] = data;
+ data += strlen (data) + 1;
+ }
+ files[nfiles] = 0;
/* Sort the table. */
- qsort (files, files_index, sizeof (char *), compare_names);
+ qsort (files, nfiles, sizeof (char *), compare_names);
- /* Return a description of location and length of the table. */
- dirdata.files = files;
- dirdata.length = files_index;
-
- return dirdata;
+ return 0;
}
/* Sort the files now in the table. */
@@ -113,105 +129,85 @@ compare_names (file1, file2)
return strcmp (*file1, *file2);
}
-/* Compare the contents of two directories named NAME1 and NAME2.
+/* Compare the contents of two directories named in FILEVEC[0] and FILEVEC[1].
This is a top-level routine; it does everything necessary for diff
on two directories.
- NONEX1 nonzero says directory NAME1 doesn't exist, but pretend it is
- empty. Likewise NONEX2.
+ FILEVEC[0].desc == -1 says directory FILEVEC[0] doesn't exist,
+ but pretend it is empty. Likewise for FILEVEC[1].
HANDLE_FILE is a caller-provided subroutine called to handle each file.
It gets five operands: dir and name (rel to original working dir) of file
- in dir 1, dir and name pathname of file in dir 2, and the recursion depth.
+ in dir 0, dir and name pathname of file in dir 1, and the recursion depth.
For a file that appears in only one of the dirs, one of the name-args
to HANDLE_FILE is zero.
- DEPTH is the current depth in recursion.
+ DEPTH is the current depth in recursion, used for skipping top-level
+ files by the -S option.
Returns the maximum of all the values returned by HANDLE_FILE,
or 2 if trouble is encountered in opening files. */
int
-diff_dirs (name1, name2, handle_file, depth, nonex1, nonex2)
- char *name1, *name2;
+diff_dirs (filevec, handle_file, depth)
+ struct file_data filevec[];
int (*handle_file) ();
- int nonex1, nonex2;
+ int depth;
{
- struct dirdata data1, data2;
- register int i1, i2;
- int val = 0;
- int v1;
+ struct dirdata dirdata[2];
+ int val = 0; /* Return value. */
+ int i;
/* Get sorted contents of both dirs. */
- data1 = dir_sort (name1, nonex1);
- data2 = dir_sort (name2, nonex2);
- if (data1.length == -1 || data2.length == -1)
- {
- if (data1.length >= 0)
- free (data1.files);
- if (data2.length >= 0)
- free (data2.files);
- return 2;
- }
-
- i1 = 0;
- i2 = 0;
-
- /* If -Sname was specified, and this is the topmost level of comparison,
- ignore all file names less than the specified starting name. */
-
- if (dir_start_file && depth == 0)
- {
- while (i1 < data1.length && strcmp (data1.files[i1], dir_start_file) < 0)
- i1++;
- while (i2 < data2.length && strcmp (data2.files[i2], dir_start_file) < 0)
- i2++;
- }
-
- /* Loop while files remain in one or both dirs. */
- while (i1 < data1.length || i2 < data2.length)
+ for (i = 0; i < 2; i++)
+ if (dir_sort (&filevec[i], &dirdata[i]) != 0)
+ {
+ perror_with_name (filevec[i].name);
+ val = 2;
+ }
+
+ if (val == 0)
{
- int nameorder;
+ register char **files0 = dirdata[0].files;
+ register char **files1 = dirdata[1].files;
+ char *name0 = filevec[0].name;
+ char *name1 = filevec[1].name;
- /* Compare next name in dir 1 with next name in dir 2.
- At the end of a dir,
- pretend the "next name" in that dir is very large. */
+ /* If `-S name' was given, and this is the topmost level of comparison,
+ ignore all file names less than the specified starting name. */
- if (i1 == data1.length)
- nameorder = 1;
- else if (i2 == data2.length)
- nameorder = -1;
- else
- nameorder = strcmp (data1.files[i1], data2.files[i2]);
-
- if (nameorder == 0)
- {
- /* We have found a file (or subdir) in common between both dirs.
- Compare the two files. */
- v1 = (*handle_file) (name1, data1.files[i1], name2, data2.files[i2],
- depth + 1);
- i1++, i2++;
- }
- if (nameorder < 0)
+ if (dir_start_file && depth == 0)
{
- /* Next filename in dir 1 is less; that is a file in dir 1 only. */
- v1 = (*handle_file) (name1, data1.files[i1], name2, 0, depth + 1);
- i1++;
+ while (*files0 && strcmp (*files0, dir_start_file) < 0)
+ files0++;
+ while (*files1 && strcmp (*files1, dir_start_file) < 0)
+ files1++;
}
- if (nameorder > 0)
+
+ /* Loop while files remain in one or both dirs. */
+ while (*files0 || *files1)
{
- /* Next filename in dir 2 is less; that is a file in dir 2 only. */
- v1 = (*handle_file) (name1, 0, name2, data2.files[i2], depth + 1);
- i2++;
+ /* Compare next name in dir 0 with next name in dir 1.
+ At the end of a dir,
+ pretend the "next name" in that dir is very large. */
+ int nameorder = (!*files0 ? 1 : !*files1 ? -1
+ : strcmp (*files0, *files1));
+ int v1 = (*handle_file) (name0, 0 < nameorder ? 0 : *files0++,
+ name1, nameorder < 0 ? 0 : *files1++,
+ depth + 1);
+ if (v1 > val)
+ val = v1;
}
- if (v1 > val)
- val = v1;
}
- if (data1.files)
- free (data1.files);
- if (data2.files)
- free (data2.files);
+
+ for (i = 0; i < 2; i++)
+ {
+ if (dirdata[i].files)
+ free (dirdata[i].files);
+ if (dirdata[i].data)
+ free (dirdata[i].data);
+ }
return val;
}
diff --git a/gnu/usr.bin/diff/ed.c b/gnu/usr.bin/diff/ed.c
index 59cb99d4ad5..fd051f29dbd 100644
--- a/gnu/usr.bin/diff/ed.c
+++ b/gnu/usr.bin/diff/ed.c
@@ -1,11 +1,11 @@
/* Output routines for ed-script format.
- Copyright (C) 1988, 1989 Free Software Foundation, Inc.
+ Copyright (C) 1988, 89, 91, 92 Free Software Foundation, Inc.
This file is part of GNU DIFF.
GNU DIFF 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 1, or (at your option)
+the Free Software Foundation; either version 2, or (at your option)
any later version.
GNU DIFF is distributed in the hope that it will be useful,
@@ -19,6 +19,8 @@ the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */
#include "diff.h"
+int change_letter ();
+int translate_line_number ();
static void print_rcs_hunk ();
static void print_ed_hunk ();
static void pr_forward_ed_hunk ();
@@ -56,6 +58,8 @@ print_ed_hunk (hunk)
if (!deletes && !inserts)
return;
+ begin_output ();
+
/* Print out the line number header for this hunk */
print_number_range (',', &files[0], f0, l0);
fprintf (outfile, "%c\n", change_letter (inserts, deletes));
@@ -78,8 +82,8 @@ print_ed_hunk (hunk)
so that we will output another ed-command later
to change the double dot into a single dot. */
- if (files[1].linbuf[i].text[0] == '.'
- && files[1].linbuf[i].text[1] == '\n')
+ if (files[1].linbuf[i][0] == '.'
+ && files[1].linbuf[i][1] == '\n')
{
fprintf (outfile, "..\n");
fprintf (outfile, ".\n");
@@ -123,6 +127,8 @@ pr_forward_ed_hunk (hunk)
if (!deletes && !inserts)
return;
+ begin_output ();
+
fprintf (outfile, "%c", change_letter (inserts, deletes));
print_number_range (' ', files, f0, l0);
fprintf (outfile, "\n");
@@ -168,6 +174,8 @@ print_rcs_hunk (hunk)
if (!deletes && !inserts)
return;
+ begin_output ();
+
translate_range (&files[0], f0, l0, &tf0, &tl0);
if (deletes)
diff --git a/gnu/usr.bin/diff/getopt.c b/gnu/usr.bin/diff/getopt.c
index d7677b4163a..a59a0133983 100644
--- a/gnu/usr.bin/diff/getopt.c
+++ b/gnu/usr.bin/diff/getopt.c
@@ -1,10 +1,15 @@
/* Getopt for GNU.
- Copyright (C) 1987, 1989, 1990 Free Software Foundation, Inc.
+ 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!
- 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 1, or (at your option)
- any later version.
+ 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
@@ -13,65 +18,84 @@
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. */
+ Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */
-#ifdef __STDC__
-#define CONST const
-#else
-#define CONST
+/* 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
-/* 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 _POSIX_OPTION_ORDER 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 <stdio.h>
+#ifdef HAVE_CONFIG_H
+#include "config.h"
+#endif
-/* If compiled with GNU C, use the built-in alloca */
#ifdef __GNUC__
#define alloca __builtin_alloca
#else /* not __GNUC__ */
-#ifdef sparc
+#if defined (HAVE_ALLOCA_H) || (defined(sparc) && (defined(sun) || (!defined(USG) && !defined(SVR4) && !defined(__svr4__))))
#include <alloca.h>
#else
+#ifndef _AIX
char *alloca ();
#endif
+#endif /* alloca.h */
#endif /* not __GNUC__ */
-#if defined(STDC_HEADERS) || defined(__GNU_LIBRARY__)
-#include <stdlib.h>
-#include <string.h>
-#define bcopy(s, d, n) memcpy ((d), (s), (n))
-#define index strchr
-#else
-
-#ifdef USG
-#include <string.h>
-#define bcopy(s, d, n) memcpy ((d), (s), (n))
-#define index strchr
-#else
-#ifdef VMS
-#include <string.h>
-#else
-#include <strings.h>
-#endif
-void bcopy ();
+#if !__STDC__ && !defined(const) && IN_GCC
+#define const
#endif
-char *getenv ();
-char *malloc ();
+/* This tells Alpha OSF/1 not to define a getopt prototype in <stdio.h>. */
+#ifndef _NO_PROTO
+#define _NO_PROTO
#endif
+#include <stdio.h>
+
+/* 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 <stdlib.h>
+#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.
@@ -92,6 +116,7 @@ char *optarg = 0;
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
@@ -108,17 +133,23 @@ static char *nextchar;
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
- _POSIX_OPTION_ORDER is defined, PERMUTE otherwise.
+ 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 _POSIX_OPTION_ORDER, or using `+' as the first character
+ 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,
@@ -141,29 +172,47 @@ 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 <string.h>
+#define my_index strchr
+#define my_bcopy(src, dst, n) memcpy ((dst), (src), (n))
+#else
-/* Describe the long-named options requested by the application.
- _GETOPT_LONG_OPTIONS is a vector of `struct option' terminated by an
- element containing a name which is zero.
- The field `has_arg' is 1 if the option takes an argument,
- 2 if it takes an optional argument. */
-
-struct option
-{
- char *name;
- int has_arg;
- int *flag;
- int val;
-};
-
-CONST struct option *_getopt_long_options;
+/* Avoid depending on library functions or files
+ whose names are inconsistent. */
-int _getopt_long_only = 0;
+char *getenv ();
-/* Index in _GETOPT_LONG_OPTIONS of the long-named option actually found.
- Only valid when a long-named option was found. */
+static char *
+my_index (str, chr)
+ const char *str;
+ int chr;
+{
+ while (*str)
+ {
+ if (*str == chr)
+ return (char *) str;
+ str++;
+ }
+ return 0;
+}
-int option_index;
+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. */
@@ -176,26 +225,28 @@ 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.
+ 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.
+ 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. */
+ 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);
+ char **temp = (char **) __alloca (nonopts_size);
/* Interchange the two blocks of data in ARGV. */
- bcopy (&argv[first_nonopt], temp, nonopts_size);
- bcopy (&argv[last_nonopt], &argv[first_nonopt],
- (optind - last_nonopt) * sizeof (char *));
- bcopy (temp, &argv[first_nonopt + optind - last_nonopt], nonopts_size);
+ 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. */
@@ -236,21 +287,40 @@ exchange (argv)
handling the non-option ARGV-elements.
See the comments about RETURN_IN_ORDER and REQUIRE_ORDER, above.
- Long-named options begin with `+' instead of `-'.
+ 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
- otherwise. */
+ 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 (argc, argv, optstring)
+_getopt_internal (argc, argv, optstring, longopts, longind, long_only)
int argc;
- char **argv;
- CONST char *optstring;
+ 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.
@@ -262,7 +332,7 @@ getopt (argc, argv, optstring)
{
first_nonopt = last_nonopt = optind = 1;
- nextchar = 0;
+ nextchar = NULL;
/* Determine how to handle the ordering of options and nonoptions. */
@@ -276,13 +346,13 @@ getopt (argc, argv, optstring)
ordering = REQUIRE_ORDER;
++optstring;
}
- else if (getenv ("_POSIX_OPTION_ORDER") != 0)
+ else if (getenv ("POSIXLY_CORRECT") != NULL)
ordering = REQUIRE_ORDER;
else
ordering = PERMUTE;
}
- if (nextchar == 0 || *nextchar == 0)
+ if (nextchar == NULL || *nextchar == '\0')
{
if (ordering == PERMUTE)
{
@@ -290,7 +360,7 @@ getopt (argc, argv, optstring)
exchange them so that the options come first. */
if (first_nonopt != last_nonopt && last_nonopt != optind)
- exchange (argv);
+ exchange ((char **) argv);
else if (last_nonopt != optind)
first_nonopt = optind;
@@ -298,11 +368,12 @@ getopt (argc, argv, optstring)
and extend the range of non-options previously skipped. */
while (optind < argc
- && (argv[optind][0] != '-'
- || argv[optind][1] == 0)
- && (_getopt_long_options == 0
- || argv[optind][0] != '+'
- || argv[optind][1] == 0))
+ && (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;
}
@@ -317,7 +388,7 @@ getopt (argc, argv, optstring)
optind++;
if (first_nonopt != last_nonopt && last_nonopt != optind)
- exchange (argv);
+ exchange ((char **) argv);
else if (first_nonopt == last_nonopt)
first_nonopt = optind;
last_nonopt = argc;
@@ -340,9 +411,12 @@ getopt (argc, argv, optstring)
/* 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)
- && (_getopt_long_options == 0
- || argv[optind][0] != '+' || argv[optind][1] == 0))
+ 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;
@@ -353,26 +427,30 @@ getopt (argc, argv, optstring)
/* We have found another option-ARGV-element.
Start decoding its characters. */
- nextchar = argv[optind] + 1;
+ nextchar = (argv[optind] + 1
+ + (longopts != NULL && argv[optind][1] == '-'));
}
- if (_getopt_long_options != 0
- && (argv[optind][0] == '+'
- || (_getopt_long_only && argv[optind][0] == '-'))
- )
+ if (longopts != NULL
+ && ((argv[optind][0] == '-'
+ && (argv[optind][1] == '-' || long_only))
+#ifdef GETOPT_COMPAT
+ || argv[optind][0] == '+'
+#endif /* GETOPT_COMPAT */
+ ))
{
- CONST struct option *p;
+ const struct option *p;
char *s = nextchar;
int exact = 0;
int ambig = 0;
- CONST struct option *pfound = 0;
+ const struct option *pfound = NULL;
int indfound;
while (*s && *s != '=')
s++;
/* Test all options for either exact match or abbreviated matches. */
- for (p = _getopt_long_options, option_index = 0; p->name;
+ for (p = longopts, option_index = 0; p->name;
p++, option_index++)
if (!strncmp (p->name, nextchar, s - nextchar))
{
@@ -384,7 +462,7 @@ getopt (argc, argv, optstring)
exact = 1;
break;
}
- else if (pfound == 0)
+ else if (pfound == NULL)
{
/* First nonexact match found. */
pfound = p;
@@ -397,26 +475,39 @@ getopt (argc, argv, optstring)
if (ambig && !exact)
{
- fprintf (stderr, "%s: option `%s' is ambiguous\n",
- argv[0], argv[optind]);
+ if (opterr)
+ fprintf (stderr, "%s: option `%s' is ambiguous\n",
+ argv[0], argv[optind]);
nextchar += strlen (nextchar);
optind++;
return '?';
}
- if (pfound != 0)
+ if (pfound != NULL)
{
option_index = indfound;
optind++;
if (*s)
{
- if (pfound->has_arg > 0)
+ /* 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
{
- fprintf (stderr,
- "%s: option `%c%s' doesn't allow an argument\n",
- argv[0], argv[optind - 1][0], pfound->name);
+ 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 '?';
}
@@ -427,13 +518,16 @@ getopt (argc, argv, optstring)
optarg = argv[optind++];
else
{
- fprintf (stderr, "%s: option `%s' requires an argument\n",
- argv[0], argv[optind - 1]);
+ if (opterr)
+ fprintf (stderr, "%s: option `%s' requires an argument\n",
+ argv[0], argv[optind - 1]);
nextchar += strlen (nextchar);
- return '?';
+ return optstring[0] == ':' ? ':' : '?';
}
}
nextchar += strlen (nextchar);
+ if (longind != NULL)
+ *longind = option_index;
if (pfound->flag)
{
*(pfound->flag) = pfound->val;
@@ -441,17 +535,28 @@ getopt (argc, argv, optstring)
}
return pfound->val;
}
- /* Can't find it as a long option. If this is getopt_long_only,
- and the option starts with '-' and is a valid short
- option, then interpret it as a short option. Otherwise it's
- an error. */
- if (_getopt_long_only == 0 || argv[optind][0] == '+' ||
- index (optstring, *nextchar) == 0)
+ /* 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 != 0)
- fprintf (stderr, "%s: unrecognized option `%c%s'\n",
- argv[0], argv[optind][0], nextchar);
- nextchar += strlen (nextchar);
+ 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 '?';
}
@@ -461,23 +566,28 @@ getopt (argc, argv, optstring)
{
char c = *nextchar++;
- char *temp = index (optstring, c);
+ char *temp = my_index (optstring, c);
/* Increment `optind' when we start to process its last character. */
- if (*nextchar == 0)
- optind++;
+ if (*nextchar == '\0')
+ ++optind;
- if (temp == 0 || c == ':')
+ if (temp == NULL || c == ':')
{
- if (opterr != 0)
+ 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);
+ 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] == ':')
@@ -485,19 +595,19 @@ getopt (argc, argv, optstring)
if (temp[2] == ':')
{
/* This is an option that accepts an argument optionally. */
- if (*nextchar != 0)
+ if (*nextchar != '\0')
{
optarg = nextchar;
optind++;
}
else
optarg = 0;
- nextchar = 0;
+ nextchar = NULL;
}
else
{
/* This is an option that requires an argument. */
- if (*nextchar != 0)
+ if (*nextchar != '\0')
{
optarg = nextchar;
/* If we end this ARGV-element by taking the rest as an arg,
@@ -506,21 +616,47 @@ getopt (argc, argv, optstring)
}
else if (optind == argc)
{
- if (opterr != 0)
- fprintf (stderr, "%s: option `-%c' requires an argument\n",
- argv[0], c);
- c = '?';
+ 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 = 0;
+ 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
diff --git a/gnu/usr.bin/diff/getopt.h b/gnu/usr.bin/diff/getopt.h
index 661acb9273d..45541f5ac0f 100644
--- a/gnu/usr.bin/diff/getopt.h
+++ b/gnu/usr.bin/diff/getopt.h
@@ -1,10 +1,10 @@
-/* declarations for getopt
- Copyright (C) 1989, 1990 Free Software Foundation, Inc.
+/* 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 1, or (at your option)
- any later version.
+ 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
@@ -13,7 +13,14 @@
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. */
+ 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,
@@ -42,16 +49,21 @@ extern int optind;
extern int opterr;
+/* Set to an option character which was unrecognized. */
+
+extern int optopt;
+
/* Describe the long-named options requested by the application.
- _GETOPT_LONG_OPTIONS is a vector of `struct option' terminated by an
- element containing a name which is zero.
+ 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:
- 0 if the option does not take an argument,
- 1 if the option requires an argument,
- 2 if the option takes an optional argument.
+ 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 nonzero, it points to a variable that is set
+ 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.
@@ -64,39 +76,54 @@ extern int opterr;
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;
};
-#ifdef __STDC__
-extern const struct option *_getopt_long_options;
-#else
-extern struct option *_getopt_long_options;
+/* 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
-/* If nonzero, '-' can introduce long-named options.
- Set by getopt_long_only. */
-
-extern int _getopt_long_only;
-
-/* The index in GETOPT_LONG_OPTIONS of the long-named option found.
- Only valid when a long-named option has been found by the most
- recent call to `getopt'. */
-
-extern int option_index;
-
-#ifdef __STDC__
-int getopt (int argc, char **argv, const char *shortopts);
-int getopt_long (int argc, char **argv, const char *shortopts,
- const struct option *longopts, int *longind);
-int getopt_long_only (int argc, char **argv, const char *shortopts,
- const struct option *longopts, int *longind);
-void envopt(int *pargc, char ***pargv, char *optstr);
-#else
-int getopt ();
-int getopt_long ();
-int getopt_long_only ();
-void envopt();
-#endif
+#endif /* _GETOPT_H */
diff --git a/gnu/usr.bin/diff/getopt1.c b/gnu/usr.bin/diff/getopt1.c
index bb73d86e31b..a32615ce36e 100644
--- a/gnu/usr.bin/diff/getopt1.c
+++ b/gnu/usr.bin/diff/getopt1.c
@@ -1,10 +1,11 @@
-/* Getopt for GNU.
- Copyright (C) 1987, 1989 Free Software Foundation, Inc.
+/* 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 1, or (at your option)
- any later version.
+ 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
@@ -13,61 +14,73 @@
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. */
+ Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */
+#ifdef HAVE_CONFIG_H
+#include "config.h"
+#endif
+
#include "getopt.h"
-#ifdef __STDC__
-#define CONST const
+#if !__STDC__ && !defined(const) && IN_GCC
+#define const
+#endif
+
+#include <stdio.h>
+
+/* 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 <stdlib.h>
#else
-#define CONST
+char *getenv ();
#endif
-#if !defined (NULL)
+#ifndef NULL
#define NULL 0
#endif
int
getopt_long (argc, argv, options, long_options, opt_index)
int argc;
- char **argv;
- CONST char *options;
- CONST struct option *long_options;
+ char *const *argv;
+ const char *options;
+ const struct option *long_options;
int *opt_index;
{
- int val;
-
- _getopt_long_options = long_options;
- val = getopt (argc, argv, options);
- if (val == 0 && opt_index != NULL)
- *opt_index = option_index;
- return val;
+ 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 '-' doesn't match a long option,
+/* 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. */
+ instead. */
-int
+int
getopt_long_only (argc, argv, options, long_options, opt_index)
int argc;
- char **argv;
- CONST char *options;
- CONST struct option *long_options;
+ char *const *argv;
+ const char *options;
+ const struct option *long_options;
int *opt_index;
{
- int val;
-
- _getopt_long_options = long_options;
- _getopt_long_only = 1;
- val = getopt (argc, argv, options);
- if (val == 0 && opt_index != NULL)
- *opt_index = option_index;
- return val;
+ return _getopt_internal (argc, argv, options, long_options, opt_index, 1);
}
-
+
+#endif /* _LIBC or not __GNU_LIBRARY__. */
+
#ifdef TEST
#include <stdio.h>
@@ -83,7 +96,6 @@ main (argc, argv)
while (1)
{
int this_option_optind = optind ? optind : 1;
- char *name = '\0';
int option_index = 0;
static struct option long_options[] =
{
@@ -104,7 +116,7 @@ main (argc, argv)
switch (c)
{
case 0:
- printf ("option %s", (long_options[option_index]).name);
+ printf ("option %s", long_options[option_index].name);
if (optarg)
printf (" with arg %s", optarg);
printf ("\n");
@@ -138,6 +150,10 @@ main (argc, argv)
printf ("option c with value `%s'\n", optarg);
break;
+ case 'd':
+ printf ("option d with value `%s'\n", optarg);
+ break;
+
case '?':
break;
diff --git a/gnu/usr.bin/diff/ifdef.c b/gnu/usr.bin/diff/ifdef.c
index bfd07d11ce2..c5dde5cc9ce 100644
--- a/gnu/usr.bin/diff/ifdef.c
+++ b/gnu/usr.bin/diff/ifdef.c
@@ -1,5 +1,5 @@
/* #ifdef-format output routines for GNU DIFF.
- Copyright (C) 1989 Free Software Foundation, Inc.
+ Copyright (C) 1989, 91, 92 Free Software Foundation, Inc.
This file is part of GNU DIFF.
@@ -21,7 +21,9 @@ and this notice must be preserved on all copies. */
#include "diff.h"
+static void format_ifdef ();
static void print_ifdef_hunk ();
+static void print_ifdef_lines ();
struct change *find_change ();
static int next_line;
@@ -32,10 +34,14 @@ void
print_ifdef_script (script)
struct change *script;
{
- next_line = 0;
+ next_line = - files[0].prefix_lines;
print_script (script, find_change, print_ifdef_hunk);
- while (next_line < files[0].buffered_lines)
- print_1_line ("", &files[0].linbuf[next_line++]);
+ if (next_line < files[0].valid_lines)
+ {
+ begin_output ();
+ format_ifdef (group_format[UNCHANGED], next_line, files[0].valid_lines,
+ 0, -1);
+ }
}
/* Print a hunk of an ifdef diff.
@@ -47,39 +53,131 @@ print_ifdef_hunk (hunk)
struct change *hunk;
{
int first0, last0, first1, last1, deletes, inserts;
- register int i;
+ const char *format;
/* Determine range of line numbers involved in each file. */
analyze_hunk (hunk, &first0, &last0, &first1, &last1, &deletes, &inserts);
- if (!deletes && !inserts)
+ if (inserts)
+ format = deletes ? group_format[CHANGED] : group_format[NEW];
+ else if (deletes)
+ format = group_format[OLD];
+ else
return;
- /* Print out lines up to this change. */
- while (next_line < first0)
- print_1_line ("", &files[0].linbuf[next_line++]);
+ begin_output ();
- /* Print out stuff deleted from first file. */
- if (deletes)
- {
- fprintf (outfile, "#ifndef %s\n", ifdef_string);
- for (i = first0; i <= last0; i++)
- print_1_line ("", &files[0].linbuf[i]);
- next_line = i;
- }
+ /* Print lines up to this change. */
+ if (next_line < first0)
+ format_ifdef (group_format[UNCHANGED], next_line, first0, 0, -1);
- /* Print out stuff inserted from second file. */
- if (inserts)
+ /* Print this change. */
+ next_line = last0 + 1;
+ format_ifdef (format, first0, next_line, first1, last1 + 1);
+}
+
+/* Print a set of lines according to FORMAT.
+ Lines BEG0 up to END0 are from the first file.
+ If END1 is -1, then the second file's lines are identical to the first;
+ otherwise, lines BEG1 up to END1 are from the second file. */
+
+static void
+format_ifdef (format, beg0, end0, beg1, end1)
+ const char *format;
+ int beg0, end0, beg1, end1;
+{
+ register FILE *out = outfile;
+ register char c;
+ register const char *f = format;
+
+ while ((c = *f++) != 0)
{
- if (deletes)
- fprintf (outfile, "#else /* %s */\n", ifdef_string);
- else
- fprintf (outfile, "#ifdef %s\n", ifdef_string);
- for (i = first1; i <= last1; i++)
- print_1_line ("", &files[1].linbuf[i]);
- }
+ if (c == '%')
+ switch ((c = *f++))
+ {
+ case 0:
+ return;
+
+ case '<':
+ /* Print lines deleted from first file. */
+ print_ifdef_lines (line_format[OLD], &files[0], beg0, end0);
+ continue;
+
+ case '=':
+ /* Print common lines. */
+ print_ifdef_lines (line_format[UNCHANGED], &files[0], beg0, end0);
+ continue;
+
+ case '>':
+ /* Print lines inserted from second file. */
+ if (end1 == -1)
+ print_ifdef_lines (line_format[NEW], &files[0], beg0, end0);
+ else
+ print_ifdef_lines (line_format[NEW], &files[1], beg1, end1);
+ continue;
+
+ case '0':
+ c = 0;
+ break;
+
+ default:
+ break;
+ }
+ putc (c, out);
+ }
+}
- if (inserts)
- fprintf (outfile, "#endif /* %s */\n", ifdef_string);
+/* Use FORMAT to print each line of CURRENT starting with FROM
+ and continuing up to UPTO. */
+static void
+print_ifdef_lines (format, current, from, upto)
+ const char *format;
+ const struct file_data *current;
+ int from, upto;
+{
+ const char * const *linbuf = current->linbuf;
+
+ /* If possible, use a single fwrite; it's faster. */
+ if (!tab_expand_flag && strcmp (format, "%l\n") == 0)
+ fwrite (linbuf[from], sizeof (char),
+ linbuf[upto] + (linbuf[upto][-1] != '\n') - linbuf[from],
+ outfile);
+ else if (!tab_expand_flag && strcmp (format, "%L") == 0)
+ fwrite (linbuf[from], sizeof (char), linbuf[upto] - linbuf[from], outfile);
else
- fprintf (outfile, "#endif /* not %s */\n", ifdef_string);
+ for (; from < upto; from++)
+ {
+ register FILE *out = outfile;
+ register char c;
+ register const char *f = format;
+
+ while ((c = *f++) != 0)
+ {
+ if (c == '%')
+ switch ((c = *f++))
+ {
+ case 0:
+ goto format_done;
+
+ case 'l':
+ output_1_line (linbuf[from],
+ linbuf[from + 1]
+ - (linbuf[from + 1][-1] == '\n'), 0, 0);
+ continue;
+
+ case 'L':
+ output_1_line (linbuf[from], linbuf[from + 1], 0, 0);
+ continue;
+
+ case '0':
+ c = 0;
+ break;
+
+ default:
+ break;
+ }
+ putc (c, out);
+ }
+
+ format_done:;
+ }
}
diff --git a/gnu/usr.bin/diff/io.c b/gnu/usr.bin/diff/io.c
index 5d53fe18d5a..a0633cd5350 100644
--- a/gnu/usr.bin/diff/io.c
+++ b/gnu/usr.bin/diff/io.c
@@ -1,11 +1,11 @@
/* File I/O for GNU DIFF.
- Copyright (C) 1988, 1989 Free Software Foundation, Inc.
+ Copyright (C) 1988, 1989, 1992 Free Software Foundation, Inc.
This file is part of GNU DIFF.
GNU DIFF 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 1, or (at your option)
+the Free Software Foundation; either version 2, or (at your option)
any later version.
GNU DIFF is distributed in the hope that it will be useful,
@@ -21,318 +21,387 @@ the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */
/* Rotate a value n bits to the left. */
#define UINT_BIT (sizeof (unsigned) * CHAR_BIT)
-#define ROL(v, n) ((v) << (n) | (v) >> UINT_BIT - (n))
+#define ROL(v, n) ((v) << (n) | (v) >> (UINT_BIT - (n)))
/* Given a hash value and a new character, return a new hash value. */
#define HASH(h, c) ((c) + ROL (h, 7))
-/* Current file under consideration. */
-struct file_data *current;
+int line_cmp ();
+
+/* Guess remaining number of lines from number N of lines so far,
+ size S so far, and total size T. */
+#define GUESS_LINES(n,s,t) (((t) - (s)) / ((n) < 10 ? 32 : (s) / ((n)-1)) + 5)
+
+/* Type used for fast prefix comparison in find_identical_ends. */
+typedef long word;
+
+/* Character classes. */
+const char textchar[] = {
+ /* ISO 8859 */
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 2, 2, 1, 2, 2, 2, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 2, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 2, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1,
+ 1, 1, 1, 1, 1, 1, 1, 1
+};
+
+/* Lines are put into equivalence classes (of lines that match in line_cmp).
+ Each equivalence class is represented by one of these structures,
+ but only while the classes are being computed.
+ Afterward, each class is represented by a number. */
+struct equivclass
+{
+ int next; /* Next item in this bucket. */
+ unsigned hash; /* Hash of lines in this class. */
+ const char *line; /* A line that fits this class. */
+ int length; /* The length of that line. */
+};
+
+/* Hash-table: array of buckets, each being a chain of equivalence classes. */
+static int *buckets;
+
+/* Number of buckets in the hash table array. */
+static int nbuckets;
+
+/* Array in which the equivalence classes are allocated.
+ The bucket-chains go through the elements in this array.
+ The number of an equivalence class is its index in this array. */
+static struct equivclass *equivs;
+
+/* Index of first free element in the array `equivs'. */
+static int equivs_index;
+
+/* Number of elements allocated in the array `equivs'. */
+static int equivs_alloc;
/* Check for binary files and compare them for exact identity. */
/* Return 1 if BUF contains a non text character.
SIZE is the number of characters in BUF. */
-static int
-binary_file_p (buf, size)
- char *buf;
- int size;
-{
- static const char textchar[] = {
- /* ISO 8859 */
- 0, 0, 0, 0, 0, 0, 0, 0,
- 0, 1, 1, 1, 1, 1, 0, 0,
- 0, 0, 0, 0, 0, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 0, 0,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 0,
- 0, 0, 0, 0, 0, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 0, 0,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- 1, 1, 1, 1, 1, 1, 1, 1,
- };
- while (--size >= 0)
- if (!textchar[*buf++ & 0377])
- return 1;
- return 0;
-}
-
-int binary_file_threshold = 512;
+#define binary_file_p(buf, size) (size != 0 && memchr (buf, '\0', size) != 0)
-/* Slurp the current file completely into core.
- Return nonzero if it appears to be a binary file. */
+/* Get ready to read the current file.
+ Return nonzero if SKIP_TEST is zero,
+ and if it appears to be a binary file. */
-static int
-slurp ()
+int
+sip (current, skip_test)
+ struct file_data *current;
+ int skip_test;
{
/* If we have a nonexistent file at this stage, treat it as empty. */
if (current->desc < 0)
{
- current->bufsize = 0;
+ /* Leave room for a sentinel. */
+ current->buffer = xmalloc (sizeof (word));
+ current->bufsize = sizeof (word);
current->buffered_chars = 0;
- current->buffer = 0;
- }
- /* If it's a regular file, we can just get the size out of the stat
- block and slurp it in all at once. */
- /* In all cases, we leave room in the buffer for 2 extra chars
- beyond those that current->bufsize describes:
- one for a newline (in case the text does not end with one)
- and one for a sentinel in find_identical_ends. */
- else if ((current->stat.st_mode & S_IFMT) == S_IFREG)
- {
- current->bufsize = current->stat.st_size;
- current->buffer = (char *) xmalloc (current->bufsize + 2);
- current->buffered_chars
- = read (current->desc, current->buffer, current->bufsize);
- if (current->buffered_chars < 0)
- pfatal_with_name (current->name);
}
else
{
- int cc;
+ current->bufsize = current->buffered_chars
+ = STAT_BLOCKSIZE (current->stat);
+
+ if (S_ISREG (current->stat.st_mode))
+ /* Get the size out of the stat block.
+ Allocate enough room for appended newline and sentinel.
+ Allocate at least one block, to prevent overrunning the buffer
+ when comparing growing binary files. */
+ current->bufsize = max (current->bufsize,
+ current->stat.st_size + sizeof (word) + 1);
+
+ current->buffer = xmalloc (current->bufsize);
+ if (skip_test)
+ current->buffered_chars = 0;
+ else
+ {
+ /* Check first part of file to see if it's a binary file. */
+ current->buffered_chars = read (current->desc,
+ current->buffer,
+ current->buffered_chars);
+ if (current->buffered_chars < 0)
+ pfatal_with_name (current->name);
+ return binary_file_p (current->buffer, current->buffered_chars);
+ }
+ }
+
+ return 0;
+}
- current->bufsize = 4096;
- current->buffer = (char *) xmalloc (current->bufsize + 2);
- current->buffered_chars = 0;
-
- /* Not a regular file; read it in a little at a time, growing the
- buffer as necessary. */
- while ((cc = read (current->desc,
- current->buffer + current->buffered_chars,
- current->bufsize - current->buffered_chars))
- > 0)
+/* Slurp the rest of the current file completely into memory. */
+
+void
+slurp (current)
+ struct file_data *current;
+{
+ int cc;
+
+ if (current->desc < 0)
+ /* The file is nonexistent. */
+ ;
+ else if (S_ISREG (current->stat.st_mode))
+ {
+ /* It's a regular file; slurp in the rest all at once. */
+ cc = current->stat.st_size - current->buffered_chars;
+ if (cc)
{
+ cc = read (current->desc,
+ current->buffer + current->buffered_chars,
+ cc);
+ if (cc < 0)
+ pfatal_with_name (current->name);
current->buffered_chars += cc;
+ }
+ }
+ /* It's not a regular file; read it, growing the buffer as needed. */
+ else if (always_text_flag || current->buffered_chars != 0)
+ {
+ for (;;)
+ {
if (current->buffered_chars == current->bufsize)
{
current->bufsize = current->bufsize * 2;
current->buffer = (char *) xrealloc (current->buffer,
- current->bufsize + 2);
+ current->bufsize);
}
+ cc = read (current->desc,
+ current->buffer + current->buffered_chars,
+ current->bufsize - current->buffered_chars);
+ if (cc == 0)
+ break;
+ if (cc < 0)
+ pfatal_with_name (current->name);
+ current->buffered_chars += cc;
}
- if (cc < 0)
- pfatal_with_name (current->name);
+ /* Allocate just enough room for appended newline and sentinel. */
+ current->bufsize = current->buffered_chars + sizeof (word) + 1;
+ current->buffer = (char *) xrealloc (current->buffer, current->bufsize);
}
-
- /* Check first part of file to see if it's a binary file. */
- if (! always_text_flag
- && binary_file_p (current->buffer,
- min (current->buffered_chars, binary_file_threshold)))
- return 1;
-
- /* If not binary, make sure text ends in a newline,
- but remember that we had to add one unless -B is in effect. */
- if (current->buffered_chars > 0
- && current->buffer[current->buffered_chars - 1] != '\n')
- {
- current->missing_newline = !ignore_blank_lines_flag;
- current->buffer[current->buffered_chars++] = '\n';
- }
- else
- current->missing_newline = 0;
-
- /* Don't use uninitialized storage. */
- if (current->buffer != 0)
- current->buffer[current->buffered_chars] = '\0';
-
- return 0;
}
-/* Split the file into lines, simultaneously computing the hash codes for
+/* Split the file into lines, simultaneously computing the equivalence class for
each line. */
-void
-find_and_hash_each_line ()
+static void
+find_and_hash_each_line (current)
+ struct file_data *current;
{
unsigned h;
- int i;
- unsigned char *p = (unsigned char *) current->prefix_end, *ip, c;
-
- /* Attempt to get a good initial guess as to the number of lines. */
- current->linbufsize = current->buffered_chars / 50 + 5;
- current->linbuf
- = (struct line_def *) xmalloc (current->linbufsize * sizeof (struct line_def));
-
- if (function_regexp || output_style == OUTPUT_IFDEF)
- {
- /* If the -C, -D or -F option is used, we need to find the lines
- of the matching prefix. At least we will need to find the last few,
- but since we don't know how many, it's easiest to find them all.
- If -D is specified, we need all the lines of the first file. */
- current->buffered_lines = 0;
- p = (unsigned char *) current->buffer;
- }
- else
+ const unsigned char *p = (const unsigned char *) current->prefix_end;
+ unsigned char c;
+ int i, length, *bucket;
+
+ /* Cache often-used quantities in local variables to help the compiler. */
+ const char **linbuf = current->linbuf;
+ int alloc_lines = current->alloc_lines;
+ int line = 0;
+ int linbuf_base = current->linbuf_base;
+ int *cureqs = (int *) xmalloc (alloc_lines * sizeof (int));
+ struct equivclass *eqs = equivs;
+ int eqs_index = equivs_index;
+ int eqs_alloc = equivs_alloc;
+ const char *suffix_begin = current->suffix_begin;
+ const char *bufend = current->buffer + current->buffered_chars;
+ const char *incomplete_tail
+ = current->missing_newline && ROBUST_OUTPUT_STYLE (output_style)
+ ? bufend : (const char *) 0;
+ int varies = length_varies;
+
+ while ((const char *) p < suffix_begin)
{
- /* Skip the identical prefixes, except be prepared to handle context.
- In fact, handle 1 more preceding line than the context says,
- in case shift_boundaries moves things backwards in this file. */
- current->buffered_lines = current->prefix_lines - context - 1;
- if (current->buffered_lines < 0)
- current->buffered_lines = 0;
- for (i = 0; i < context + 1; ++i)
- /* Unless we are at the beginning, */
- if ((char *) p != current->buffer)
- /* Back up at least 1 char until at the start of a line. */
- while ((char *) --p != current->buffer && p[-1] != '\n')
- ;
- }
+ const char *ip = (const char *) p;
+
+ /* Compute the equivalence class for this line. */
- while ((char *) p < current->suffix_begin)
- {
h = 0;
- ip = p;
- if (current->prefix_end <= (char *) p)
+ /* Hash this line until we find a newline. */
+ if (ignore_case_flag)
{
- /* Hash this line until we find a newline. */
- if (ignore_case_flag)
- {
- if (ignore_all_space_flag)
- while ((c = *p) != '\n')
+ if (ignore_all_space_flag)
+ while ((c = *p++) != '\n')
+ {
+ if (! Is_space (c))
+ h = HASH (h, isupper (c) ? tolower (c) : c);
+ }
+ else if (ignore_space_change_flag)
+ while ((c = *p++) != '\n')
+ {
+ if (c == ' ' || c == '\t')
{
- if (! isspace (c))
- if (isupper (c))
- h = HASH (h, tolower (c));
- else
- h = HASH (h, c);
- ++p;
+ while ((c = *p++) == ' ' || c == '\t')
+ ;
+ if (c == '\n')
+ break;
+ h = HASH (h, ' ');
}
- else if (ignore_space_change_flag)
-
- while ((c = *p) != '\n')
- {
- if (c == ' ' || c == '\t')
- {
- while ((c = *p) == ' ' || c == '\t')
- ++p;
- if (c == '\n')
- break;
- h = HASH (h, ' ');
- }
- /* C is now the first non-space. */
- if (isupper (c))
- h = HASH (h, tolower (c));
- else
- h = HASH (h, c);
- ++p;
- }
- else
- while ((c = *p) != '\n')
- {
- if (isupper (c))
- h = HASH (h, tolower (c));
- else
- h = HASH (h, c);
- ++p;
- }
- }
+ /* C is now the first non-space. */
+ h = HASH (h, isupper (c) ? tolower (c) : c);
+ }
else
- {
- if (ignore_all_space_flag)
- while ((c = *p) != '\n')
- {
- if (! isspace (c))
- h = HASH (h, c);
- ++p;
- }
- else if (ignore_space_change_flag)
- while ((c = *p) != '\n')
- {
- if (c == ' ' || c == '\t')
- {
- while ((c = *p) == ' ' || c == '\t')
- ++p;
- if (c == '\n')
- break;
- h = HASH (h, ' ');
- }
- /* C is not the first non-space. */
- h = HASH (h, c);
- ++p;
- }
- else
- while ((c = *p) != '\n')
+ while ((c = *p++) != '\n')
+ h = HASH (h, isupper (c) ? tolower (c) : c);
+ }
+ else
+ {
+ if (ignore_all_space_flag)
+ while ((c = *p++) != '\n')
+ {
+ if (! Is_space (c))
+ h = HASH (h, c);
+ }
+ else if (ignore_space_change_flag)
+ while ((c = *p++) != '\n')
+ {
+ if (c == ' ' || c == '\t')
{
- h = HASH (h, c);
- ++p;
+ while ((c = *p++) == ' ' || c == '\t')
+ ;
+ if (c == '\n')
+ break;
+ h = HASH (h, ' ');
}
- }
+ /* C is now the first non-space. */
+ h = HASH (h, c);
+ }
+ else
+ while ((c = *p++) != '\n')
+ h = HASH (h, c);
}
- else
- /* This line is part of the matching prefix,
- so we don't need to hash it. */
- while (*p != '\n')
- ++p;
-
+
+ bucket = &buckets[h % nbuckets];
+ length = (const char *) p - ip - ((const char *) p == incomplete_tail);
+ for (i = *bucket; ; i = eqs[i].next)
+ if (!i)
+ {
+ /* Create a new equivalence class in this bucket. */
+ i = eqs_index++;
+ if (i == eqs_alloc)
+ eqs = (struct equivclass *)
+ xrealloc (eqs, (eqs_alloc*=2) * sizeof(*eqs));
+ eqs[i].next = *bucket;
+ eqs[i].hash = h;
+ eqs[i].line = ip;
+ eqs[i].length = length;
+ *bucket = i;
+ break;
+ }
+ else if (eqs[i].hash == h
+ && (eqs[i].length == length || varies)
+ && ! line_cmp (eqs[i].line, eqs[i].length, ip, length))
+ /* Reuse existing equivalence class. */
+ break;
+
/* Maybe increase the size of the line table. */
- if (current->buffered_lines >= current->linbufsize)
+ if (line == alloc_lines)
{
- while (current->buffered_lines >= current->linbufsize)
- current->linbufsize *= 2;
- current->linbuf
- = (struct line_def *) xrealloc (current->linbuf,
- current->linbufsize
- * sizeof (struct line_def));
+ /* Double (alloc_lines - linbuf_base) by adding to alloc_lines. */
+ alloc_lines = 2 * alloc_lines - linbuf_base;
+ cureqs = (int *) xrealloc (cureqs, alloc_lines * sizeof (*cureqs));
+ linbuf = (const char **) xrealloc (linbuf + linbuf_base,
+ (alloc_lines - linbuf_base)
+ * sizeof (*linbuf))
+ - linbuf_base;
}
- current->linbuf[current->buffered_lines].text = (char *) ip;
- current->linbuf[current->buffered_lines].length = p - ip + 1;
- current->linbuf[current->buffered_lines].hash = h;
- ++current->buffered_lines;
- ++p;
+ linbuf[line] = ip;
+ cureqs[line] = i;
+ ++line;
}
- i = 0;
- while ((i < context || output_style == OUTPUT_IFDEF)
- && (char *) p < current->buffer + current->buffered_chars)
+ current->buffered_lines = line;
+
+ for (i = 0; ; i++)
{
- ip = p;
- while (*p++ != '\n')
- ;
- /* Maybe increase the size of the line table. */
- if (current->buffered_lines >= current->linbufsize)
+ /* Record the line start for lines in the suffix that we care about.
+ Record one more line start than lines,
+ so that we can compute the length of any buffered line. */
+ if (line == alloc_lines)
+ {
+ /* Double (alloc_lines - linbuf_base) by adding to alloc_lines. */
+ alloc_lines = 2 * alloc_lines - linbuf_base;
+ linbuf = (const char **) xrealloc (linbuf + linbuf_base,
+ (alloc_lines - linbuf_base)
+ * sizeof (*linbuf))
+ - linbuf_base;
+ }
+ linbuf[line] = (const char *) p;
+
+ if ((const char *) p == bufend)
{
- while (current->buffered_lines >= current->linbufsize)
- current->linbufsize *= 2;
- current->linbuf
- = (struct line_def *) xrealloc (current->linbuf,
- current->linbufsize
- * sizeof (struct line_def));
+ linbuf[line] -= (const char *) p == incomplete_tail;
+ break;
}
- current->linbuf[current->buffered_lines].text = (char *) ip;
- current->linbuf[current->buffered_lines].length = p - ip;
- current->linbuf[current->buffered_lines].hash = 0;
- ++current->buffered_lines;
- ++i;
+
+ if (context <= i && no_diff_means_no_output)
+ break;
+
+ line++;
+
+ while (*p++ != '\n')
+ ;
}
- if (ROBUST_OUTPUT_STYLE (output_style)
- && current->missing_newline
- && current->suffix_begin == current->buffer + current->buffered_chars)
- --current->linbuf[current->buffered_lines - 1].length;
+ /* Done with cache in local variables. */
+ current->linbuf = linbuf;
+ current->valid_lines = line;
+ current->alloc_lines = alloc_lines;
+ current->equivs = cureqs;
+ equivs = eqs;
+ equivs_alloc = eqs_alloc;
+ equivs_index = eqs_index;
+}
+
+/* Prepare the end of the text. Make sure it's initialized.
+ Make sure text ends in a newline,
+ but remember that we had to add one unless -B is in effect. */
+
+static void
+prepare_text_end (current)
+ struct file_data *current;
+{
+ int buffered_chars = current->buffered_chars;
+ char *p = current->buffer;
+
+ if (buffered_chars == 0 || p[buffered_chars - 1] == '\n')
+ current->missing_newline = 0;
+ else
+ {
+ p[buffered_chars++] = '\n';
+ current->buffered_chars = buffered_chars;
+ current->missing_newline = ! ignore_blank_lines_flag;
+ }
+
+ /* Don't use uninitialized storage when planting or using sentinels. */
+ if (p)
+ bzero (p + buffered_chars, sizeof (word));
}
/* Given a vector of two file_data objects, find the identical
@@ -342,170 +411,217 @@ static void
find_identical_ends (filevec)
struct file_data filevec[];
{
- char *p0, *p1, *end0, *beg0;
- int lines;
-
- if (filevec[0].buffered_chars == 0 || filevec[1].buffered_chars == 0)
+ word *w0, *w1;
+ char *p0, *p1, *buffer0, *buffer1;
+ const char *end0, *beg0;
+ const char **linbuf0, **linbuf1;
+ int i, lines;
+ int n0, n1, alloc_lines0, alloc_lines1;
+ int buffered_prefix, prefix_count, prefix_mask;
+ int tem;
+
+ slurp (&filevec[0]);
+ if (filevec[0].desc != filevec[1].desc)
+ slurp (&filevec[1]);
+ else
{
- filevec[0].prefix_end = filevec[0].buffer;
- filevec[1].prefix_end = filevec[1].buffer;
- filevec[0].prefix_lines = filevec[1].prefix_lines = 0;
- filevec[0].suffix_begin = filevec[0].buffer + filevec[0].buffered_chars;
- filevec[1].suffix_begin = filevec[1].buffer + filevec[1].buffered_chars;
- filevec[0].suffix_lines = filevec[1].suffix_lines = 0;
- return;
+ filevec[1].buffer = filevec[0].buffer;
+ filevec[1].bufsize = filevec[0].bufsize;
+ filevec[1].buffered_chars = filevec[0].buffered_chars;
}
+ for (i = 0; i < 2; i++)
+ prepare_text_end (&filevec[i]);
/* Find identical prefix. */
- p0 = filevec[0].buffer;
- p1 = filevec[1].buffer;
- lines = 0;
+ p0 = buffer0 = filevec[0].buffer;
+ p1 = buffer1 = filevec[1].buffer;
- /* Insert end "sentinels", in this case characters that are guaranteed
- to make the equality test false, and thus terminate the loop. */
+ n0 = filevec[0].buffered_chars;
+ n1 = filevec[1].buffered_chars;
- if (filevec[0].buffered_chars < filevec[1].buffered_chars)
- p0[filevec[0].buffered_chars] = ~p1[filevec[0].buffered_chars];
+ if (p0 == p1)
+ /* The buffers are the same; sentinels won't work. */
+ p0 = p1 += n1;
else
- p1[filevec[1].buffered_chars] = ~p0[filevec[1].buffered_chars];
-
- /* Loop until first mismatch, or to the sentinel characters. */
- while (1)
{
- char c = *p0++;
- if (c != *p1++)
- break;
- if (c == '\n')
- ++lines;
- }
+ /* Insert end sentinels, in this case characters that are guaranteed
+ to make the equality test false, and thus terminate the loop. */
- /* Don't count missing newline as part of prefix in RCS mode. */
- if (ROBUST_OUTPUT_STYLE (output_style)
- && ((filevec[0].missing_newline
- && p0 - filevec[0].buffer > filevec[0].buffered_chars)
- ||
- (filevec[1].missing_newline
- && p1 - filevec[1].buffer > filevec[1].buffered_chars)))
- --p0, --p1, --lines;
-
- /* If the sentinel was passed, and lengths are equal, the
- files are identical. */
- if (p0 - filevec[0].buffer > filevec[0].buffered_chars
- && filevec[0].buffered_chars == filevec[1].buffered_chars)
- {
- filevec[0].prefix_end = p0 - 1;
- filevec[1].prefix_end = p1 - 1;
- filevec[0].prefix_lines = filevec[1].prefix_lines = lines;
- filevec[0].suffix_begin = filevec[0].buffer;
- filevec[1].suffix_begin = filevec[1].buffer;
- filevec[0].suffix_lines = filevec[1].suffix_lines = lines;
- return;
+ if (n0 < n1)
+ p0[n0] = ~p1[n0];
+ else
+ p1[n1] = ~p0[n1];
+
+ /* Loop until first mismatch, or to the sentinel characters. */
+
+ /* Compare a word at a time for speed. */
+ w0 = (word *) p0;
+ w1 = (word *) p1;
+ while (*w0++ == *w1++)
+ ;
+ --w0, --w1;
+
+ /* Do the last few bytes of comparison a byte at a time. */
+ p0 = (char *) w0;
+ p1 = (char *) w1;
+ while (*p0++ == *p1++)
+ ;
+ --p0, --p1;
+
+ /* Don't mistakenly count missing newline as part of prefix. */
+ if (ROBUST_OUTPUT_STYLE (output_style)
+ && (buffer0 + n0 - filevec[0].missing_newline < p0)
+ !=
+ (buffer1 + n1 - filevec[1].missing_newline < p1))
+ --p0, --p1;
}
- /* Point at first nonmatching characters. */
- --p0, --p1;
+ /* Now P0 and P1 point at the first nonmatching characters. */
- /* Skip back to last line-beginning in the prefix. */
- while (p0 != filevec[0].buffer && p0[-1] != '\n')
+ /* Skip back to last line-beginning in the prefix,
+ and then discard up to HORIZON_LINES lines from the prefix. */
+ i = horizon_lines;
+ while (p0 != buffer0 && (p0[-1] != '\n' || i--))
--p0, --p1;
/* Record the prefix. */
filevec[0].prefix_end = p0;
filevec[1].prefix_end = p1;
- filevec[0].prefix_lines = filevec[1].prefix_lines = lines;
-
+
/* Find identical suffix. */
/* P0 and P1 point beyond the last chars not yet compared. */
- p0 = filevec[0].buffer + filevec[0].buffered_chars;
- p1 = filevec[1].buffer + filevec[1].buffered_chars;
- lines = 0;
+ p0 = buffer0 + n0;
+ p1 = buffer1 + n1;
if (! ROBUST_OUTPUT_STYLE (output_style)
|| filevec[0].missing_newline == filevec[1].missing_newline)
{
- end0 = p0; /* Addr of last char in file 0. */
+ end0 = p0; /* Addr of last char in file 0. */
/* Get value of P0 at which we should stop scanning backward:
this is when either P0 or P1 points just past the last char
of the identical prefix. */
- if (filevec[0].buffered_chars < filevec[1].buffered_chars)
- beg0 = filevec[0].prefix_end;
- else
- /* Figure out where P0 will be when P1 is at the end of the prefix.
- Thus we only need to test P0. */
- beg0 = (filevec[0].prefix_end
- + filevec[0].buffered_chars - filevec[1].buffered_chars);
+ beg0 = filevec[0].prefix_end + (n0 < n1 ? 0 : n0 - n1);
/* Scan back until chars don't match or we reach that point. */
while (p0 != beg0)
- {
- char c = *--p0;
- if (c != *--p1)
- {
- /* Point at the first char of the matching suffix. */
- ++p0, ++p1;
- break;
- }
- if (c == '\n')
- ++lines;
- }
-
- /* Are we at a line-beginning in both files? */
- if (p0 != end0
- && !((p0 == filevec[0].buffer || p0[-1] == '\n')
- &&
- (p1 == filevec[1].buffer || p1[-1] == '\n')))
- {
- /* No. We counted one line too many. */
- --lines;
- /* Advance to next place that is a line-beginning in both files. */
- do
- {
- ++p0, ++p1;
- }
- while (p0 != end0 && p0[-1] != '\n');
- }
+ if (*--p0 != *--p1)
+ {
+ /* Point at the first char of the matching suffix. */
+ ++p0, ++p1;
+ beg0 = p0;
+ break;
+ }
+
+ /* Are we at a line-beginning in both files? If not, add the rest of
+ this line to the main body. Discard up to HORIZON_LINES lines from
+ the identical suffix. Also, discard one extra line,
+ because shift_boundaries may need it. */
+ i = horizon_lines + !((buffer0 == p0 || p0[-1] == '\n')
+ &&
+ (buffer1 == p1 || p1[-1] == '\n'));
+ while (i-- && p0 != end0)
+ while (*p0++ != '\n')
+ ;
+
+ p1 += p0 - beg0;
}
/* Record the suffix. */
filevec[0].suffix_begin = p0;
filevec[1].suffix_begin = p1;
- filevec[0].suffix_lines = filevec[1].suffix_lines = lines;
-}
-
-/* Lines are put into equivalence classes (of lines that match in line_cmp).
- Each equivalence class is represented by one of these structures,
- but only while the classes are being computed.
- Afterward, each class is represented by a number. */
-struct equivclass
-{
- struct equivclass *next; /* Next item in this bucket. */
- struct line_def line; /* A line that fits this class. */
-};
-/* Hash-table: array of buckets, each being a chain of equivalence classes. */
-static struct equivclass **buckets;
-
-/* Size of the bucket array. */
-static int nbuckets;
+ /* Calculate number of lines of prefix to save.
-/* Array in which the equivalence classes are allocated.
- The bucket-chains go through the elements in this array.
- The number of an equivalence class is its index in this array. */
-static struct equivclass *equivs;
+ prefix_count == 0 means save the whole prefix;
+ we need this with for options like -D that output the whole file.
+ We also need it for options like -F that output some preceding line;
+ at least we will need to find the last few lines,
+ but since we don't know how many, it's easiest to find them all.
-/* Index of first free element in the array `equivs'. */
-static int equivs_index;
+ Otherwise, prefix_count != 0. Save just prefix_count lines at start
+ of the line buffer; they'll be moved to the proper location later.
+ Handle 1 more line than the context says (because we count 1 too many),
+ rounded up to the next power of 2 to speed index computation. */
-/* Size allocated to the array `equivs'. */
-static int equivs_alloc;
+ if (no_diff_means_no_output && ! function_regexp_list)
+ {
+ for (prefix_count = 1; prefix_count < context + 1; prefix_count *= 2)
+ ;
+ prefix_mask = prefix_count - 1;
+ alloc_lines0
+ = prefix_count
+ + GUESS_LINES (0, 0, p0 - filevec[0].prefix_end)
+ + context;
+ }
+ else
+ {
+ prefix_count = 0;
+ prefix_mask = ~0;
+ alloc_lines0 = GUESS_LINES (0, 0, n0);
+ }
+
+ lines = 0;
+ linbuf0 = (const char **) xmalloc (alloc_lines0 * sizeof (*linbuf0));
+
+ /* If the prefix is needed, find the prefix lines. */
+ if (! (no_diff_means_no_output
+ && filevec[0].prefix_end == p0
+ && filevec[1].prefix_end == p1))
+ {
+ p0 = buffer0;
+ end0 = filevec[0].prefix_end;
+ while (p0 != end0)
+ {
+ int l = lines++ & prefix_mask;
+ if (l == alloc_lines0)
+ linbuf0 = (const char **) xrealloc (linbuf0, (alloc_lines0 *= 2)
+ * sizeof(*linbuf0));
+ linbuf0[l] = p0;
+ while (*p0++ != '\n')
+ ;
+ }
+ }
+ buffered_prefix = prefix_count && context < lines ? context : lines;
+ /* Allocate line buffer 1. */
+ tem = prefix_count ? filevec[1].suffix_begin - buffer1 : n1;
+
+ alloc_lines1
+ = (buffered_prefix
+ + GUESS_LINES (lines, filevec[1].prefix_end - buffer1, tem)
+ + context);
+ linbuf1 = (const char **) xmalloc (alloc_lines1 * sizeof (*linbuf1));
+
+ if (buffered_prefix != lines)
+ {
+ /* Rotate prefix lines to proper location. */
+ for (i = 0; i < buffered_prefix; i++)
+ linbuf1[i] = linbuf0[(lines - context + i) & prefix_mask];
+ for (i = 0; i < buffered_prefix; i++)
+ linbuf0[i] = linbuf1[i];
+ }
+
+ /* Initialize line buffer 1 from line buffer 0. */
+ for (i = 0; i < buffered_prefix; i++)
+ linbuf1[i] = linbuf0[i] - buffer0 + buffer1;
+
+ /* Record the line buffer, adjusted so that
+ linbuf*[0] points at the first differing line. */
+ filevec[0].linbuf = linbuf0 + buffered_prefix;
+ filevec[1].linbuf = linbuf1 + buffered_prefix;
+ filevec[0].linbuf_base = filevec[1].linbuf_base = - buffered_prefix;
+ filevec[0].alloc_lines = alloc_lines0 - buffered_prefix;
+ filevec[1].alloc_lines = alloc_lines1 - buffered_prefix;
+ filevec[0].prefix_lines = filevec[1].prefix_lines = lines;
+}
+
/* Largest primes less than some power of two, for nbuckets. Values range
from useful to preposterous. If one of these numbers isn't prime
after all, don't blame it on me, blame it on primes (6) . . . */
-static int primes[] =
+static const int primes[] =
{
509,
1021,
@@ -525,104 +641,55 @@ static int primes[] =
16777213,
33554393,
67108859, /* Preposterously large . . . */
- -1
+ 134217689,
+ 268435399,
+ 536870909,
+ 1073741789,
+ 2147483647,
+ 0
};
-/* Index of current nbuckets in primes. */
-static int primes_index;
-
-/* Find the equiv class associated with line N of the current file. */
-
-static int
-find_equiv_class (n)
- int n;
-{
- int bucket;
- struct equivclass *b, *p = NULL;
-
- /* Equivalence class 0 is permanently allocated to lines that were
- not hashed because they were parts of identical prefixes or
- suffixes. */
- if (n < current->prefix_lines
- || current->linbuf[n].text >= current->suffix_begin)
- return 0;
-
- /* Check through the appropriate bucket to see if there isn't already
- an equivalence class for this line. */
- bucket = current->linbuf[n].hash % nbuckets;
- b = buckets[bucket];
- while (b)
- {
- if (b->line.hash == current->linbuf[n].hash
- && (b->line.length == current->linbuf[n].length
- /* Lines of different lengths can match with certain options. */
- || length_varies)
- && !line_cmp (&b->line, &current->linbuf[n]))
- return b - equivs;
- p = b, b = b->next;
- }
-
- /* Create a new equivalence class in this bucket. */
-
- p = &equivs[equivs_index++];
- p->next = buckets[bucket];
- buckets[bucket] = p;
- p->line = current->linbuf[n];
-
- return equivs_index - 1;
-}
-
/* Given a vector of two file_data objects, read the file associated
with each one, and build the table of equivalence classes.
- Return nonzero if either file appears to be a binary file. */
+ Return 1 if either file appears to be a binary file. */
int
read_files (filevec)
struct file_data filevec[];
{
- int i, j;
- int binary = 0;
- int this_binary;
-
- current = &filevec[0];
- binary = this_binary = slurp ();
+ int i;
+ int skip_test = always_text_flag | no_details_flag;
+ int appears_binary = no_details_flag | sip (&filevec[0], skip_test);
- current = &filevec[1];
- this_binary = slurp ();
- if (binary || this_binary)
+ if (filevec[0].desc != filevec[1].desc)
+ appears_binary |= sip (&filevec[1], skip_test | appears_binary);
+ else
+ {
+ filevec[1].buffer = filevec[0].buffer;
+ filevec[1].bufsize = filevec[0].bufsize;
+ filevec[1].buffered_chars = filevec[0].buffered_chars;
+ }
+ if (appears_binary)
return 1;
find_identical_ends (filevec);
- for (i = 0; i < 2; ++i)
- {
- current = &filevec[i];
- find_and_hash_each_line ();
- }
-
- /* This is guaranteed to be enough space. */
- equivs_alloc = filevec[0].buffered_lines + filevec[1].buffered_lines + 1;
+ equivs_alloc = filevec[0].alloc_lines + filevec[1].alloc_lines + 1;
equivs = (struct equivclass *) xmalloc (equivs_alloc * sizeof (struct equivclass));
/* Equivalence class 0 is permanently safe for lines that were not
hashed. Real equivalence classes start at 1. */
equivs_index = 1;
-
- primes_index = 0;
- while (primes[primes_index] < equivs_alloc / 3)
- primes_index++;
- buckets = (struct equivclass **) xmalloc (primes[primes_index] * sizeof (struct equivclass *));
- bzero (buckets, primes[primes_index] * sizeof (struct equivclass *));
- nbuckets = primes[primes_index];
+ for (i = 0; primes[i] < equivs_alloc / 3; i++)
+ if (! primes[i])
+ abort ();
+ nbuckets = primes[i];
+
+ buckets = (int *) xmalloc (nbuckets * sizeof (*buckets));
+ bzero (buckets, nbuckets * sizeof (*buckets));
for (i = 0; i < 2; ++i)
- {
- current = &filevec[i];
- current->equivs
- = (int *) xmalloc (current->buffered_lines * sizeof (int));
- for (j = 0; j < current->buffered_lines; ++j)
- current->equivs[j] = find_equiv_class (j);
- }
+ find_and_hash_each_line (&filevec[i]);
filevec[0].equiv_max = filevec[1].equiv_max = equivs_index;
diff --git a/gnu/usr.bin/diff/normal.c b/gnu/usr.bin/diff/normal.c
index 63185003066..a0cf4793f6b 100644
--- a/gnu/usr.bin/diff/normal.c
+++ b/gnu/usr.bin/diff/normal.c
@@ -5,7 +5,7 @@ This file is part of GNU DIFF.
GNU DIFF 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 1, or (at your option)
+the Free Software Foundation; either version 2, or (at your option)
any later version.
GNU DIFF is distributed in the hope that it will be useful,
@@ -20,6 +20,7 @@ the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */
#include "diff.h"
+int change_letter ();
void print_normal_hunk ();
void print_number_range ();
struct change *find_change ();
@@ -50,6 +51,8 @@ print_normal_hunk (hunk)
if (!deletes && !inserts)
return;
+ begin_output ();
+
/* Print out the line number header for this hunk */
print_number_range (',', &files[0], first0, last0);
fprintf (outfile, "%c", change_letter (inserts, deletes));
diff --git a/gnu/usr.bin/diff/regex.c b/gnu/usr.bin/diff/regex.c
index 2fc0a43abfb..e8b588207da 100644
--- a/gnu/usr.bin/diff/regex.c
+++ b/gnu/usr.bin/diff/regex.c
@@ -1,9 +1,13 @@
-/* Extended regular expression matching and search library.
- Copyright (C) 1985, 1989-90 Free Software Foundation, Inc.
+/* Extended regular expression matching and search library,
+ version 0.12.
+ (Implements POSIX draft P10003.2/D11.2, except for
+ internationalization features.)
+
+ Copyright (C) 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 1, or (at your option)
+ 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,
@@ -15,34 +19,48 @@
along with this program; if not, write to the Free Software
Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */
+/* AIX requires this to be the first thing in the file. */
+#if defined (_AIX) && !defined (REGEX_MALLOC)
+ #pragma alloca
+#endif
-/* To test, compile with -Dtest. This Dtestable feature turns this into
- a self-contained program which reads a pattern, describes how it
- compiles, then reads a string and searches for it.
-
- On the other hand, if you compile with both -Dtest and -Dcanned you
- can run some tests we've already thought of. */
+#define _GNU_SOURCE
+/* We need this for `regex.h', and perhaps for the Emacs include files. */
+#include <sys/types.h>
-#ifdef emacs
+#ifdef HAVE_CONFIG_H
+#include "config.h"
+#endif
-/* The `emacs' switch turns on certain special matching commands
- that make sense only in emacs. */
+/* The `emacs' switch turns on certain matching commands
+ that make sense only in Emacs. */
+#ifdef emacs
-#include "config.h"
#include "lisp.h"
#include "buffer.h"
#include "syntax.h"
+/* Emacs uses `NULL' as a predicate. */
+#undef NULL
+
#else /* not emacs */
-#if defined (USG) || defined (STDC_HEADERS)
-#ifndef BSTRING
+/* We used to test for `BSTRING' here, but only GCC and Emacs define
+ `BSTRING', as far as I know, and neither of them use this code. */
+#if HAVE_STRING_H || STDC_HEADERS
#include <string.h>
-#define bcopy(s,d,n) memcpy((d),(s),(n))
-#define bcmp(s1,s2,n) memcmp((s1),(s2),(n))
-#define bzero(s,n) memset((s),0,(n))
+#ifndef bcmp
+#define bcmp(s1, s2, n) memcmp ((s1), (s2), (n))
+#endif
+#ifndef bcopy
+#define bcopy(s, d, n) memcpy ((d), (s), (n))
+#endif
+#ifndef bzero
+#define bzero(s, n) memset ((s), 0, (n))
#endif
+#else
+#include <strings.h>
#endif
#ifdef STDC_HEADERS
@@ -52,19 +70,8 @@ char *malloc ();
char *realloc ();
#endif
-/* Make alloca work the best possible way. */
-#ifdef __GNUC__
-#define alloca __builtin_alloca
-#else
-#ifdef sparc
-#include <alloca.h>
-#else
-char *alloca ();
-#endif
-#endif
-
-/* Define the syntax stuff, so we can do the \<, \>, etc. */
+/* Define the syntax stuff for \<, \>, etc. */
/* This must be nonzero for the wordchar and notwordchar pattern
commands in re_match_2. */
@@ -72,17 +79,16 @@ char *alloca ();
#define Sword 1
#endif
-#define SYNTAX(c) re_syntax_table[c]
-
-
#ifdef SYNTAX_TABLE
-char *re_syntax_table;
+extern char *re_syntax_table;
#else /* not SYNTAX_TABLE */
-static char re_syntax_table[256];
+/* How many characters in the character set. */
+#define CHAR_SET_SIZE 256
+static char re_syntax_table[CHAR_SET_SIZE];
static void
init_syntax_once ()
@@ -104,1473 +110,2833 @@ init_syntax_once ()
for (c = '0'; c <= '9'; c++)
re_syntax_table[c] = Sword;
+ re_syntax_table['_'] = Sword;
+
done = 1;
}
-#endif /* SYNTAX_TABLE */
-#endif /* emacs */
+#endif /* not SYNTAX_TABLE */
-/* We write fatal error messages on standard error. */
-#include <stdio.h>
+#define SYNTAX(c) re_syntax_table[c]
+
+#endif /* not emacs */
+
+/* Get the interface, including the syntax bits. */
+#include "regex.h"
-/* isalpha(3) etc. are used for the character classes. */
+/* isalpha etc. are used for the character classes. */
#include <ctype.h>
-/* Sequents are missing isgraph. */
-#ifndef isgraph
-#define isgraph(c) (isprint((c)) && !isspace((c)))
+
+/* Jim Meyering writes:
+
+ "... Some ctype macros are valid only for character codes that
+ isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
+ using /bin/cc or gcc but without giving an ansi option). So, all
+ ctype uses should be through macros like ISPRINT... If
+ STDC_HEADERS is defined, then autoconf has verified that the ctype
+ macros don't need to be guarded with references to isascii. ...
+ Defining isascii to 1 should let any compiler worth its salt
+ eliminate the && through constant folding." */
+#if ! defined (isascii) || defined (STDC_HEADERS)
+#undef isascii
+#define isascii(c) 1
#endif
-/* Get the interface, including the syntax bits. */
-#include "regex.h"
+#ifdef isblank
+#define ISBLANK(c) (isascii (c) && isblank (c))
+#else
+#define ISBLANK(c) ((c) == ' ' || (c) == '\t')
+#endif
+#ifdef isgraph
+#define ISGRAPH(c) (isascii (c) && isgraph (c))
+#else
+#define ISGRAPH(c) (isascii (c) && isprint (c) && !isspace (c))
+#endif
+#define ISPRINT(c) (isascii (c) && isprint (c))
+#define ISDIGIT(c) (isascii (c) && isdigit (c))
+#define ISALNUM(c) (isascii (c) && isalnum (c))
+#define ISALPHA(c) (isascii (c) && isalpha (c))
+#define ISCNTRL(c) (isascii (c) && iscntrl (c))
+#define ISLOWER(c) (isascii (c) && islower (c))
+#define ISPUNCT(c) (isascii (c) && ispunct (c))
+#define ISSPACE(c) (isascii (c) && isspace (c))
+#define ISUPPER(c) (isascii (c) && isupper (c))
+#define ISXDIGIT(c) (isascii (c) && isxdigit (c))
+
+#ifndef NULL
+#define NULL 0
+#endif
-/* These are the command codes that appear in compiled regular
- expressions, one per byte. Some command codes are followed by
- argument bytes. A command code can specify any interpretation
- whatsoever for its arguments. Zero-bytes may appear in the compiled
- regular expression.
+/* We remove any previous definition of `SIGN_EXTEND_CHAR',
+ since ours (we hope) works properly with all combinations of
+ machines, compilers, `char' and `unsigned char' argument types.
+ (Per Bothner suggested the basic approach.) */
+#undef SIGN_EXTEND_CHAR
+#if __STDC__
+#define SIGN_EXTEND_CHAR(c) ((signed char) (c))
+#else /* not __STDC__ */
+/* As in Harbison and Steele. */
+#define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
+#endif
+
+/* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
+ use `alloca' instead of `malloc'. This is because using malloc in
+ re_search* or re_match* could cause memory leaks when C-g is used in
+ Emacs; also, malloc is slower and causes storage fragmentation. On
+ the other hand, malloc is more portable, and easier to debug.
- The value of `exactn' is needed in search.c (search_buffer) in emacs.
+ Because we sometimes use alloca, some routines have to be macros,
+ not functions -- `alloca'-allocated space disappears at the end of the
+ function it is called in. */
+
+#ifdef REGEX_MALLOC
+
+#define REGEX_ALLOCATE malloc
+#define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
+
+#else /* not REGEX_MALLOC */
+
+/* Emacs already defines alloca, sometimes. */
+#ifndef alloca
+
+/* Make alloca work the best possible way. */
+#ifdef __GNUC__
+#define alloca __builtin_alloca
+#else /* not __GNUC__ */
+#if HAVE_ALLOCA_H
+#include <alloca.h>
+#else /* not __GNUC__ or HAVE_ALLOCA_H */
+#ifndef _AIX /* Already did AIX, up at the top. */
+char *alloca ();
+#endif /* not _AIX */
+#endif /* not HAVE_ALLOCA_H */
+#endif /* not __GNUC__ */
+
+#endif /* not alloca */
+
+#define REGEX_ALLOCATE alloca
+
+/* Assumes a `char *destination' variable. */
+#define REGEX_REALLOCATE(source, osize, nsize) \
+ (destination = (char *) alloca (nsize), \
+ bcopy (source, destination, osize), \
+ destination)
+
+#endif /* not REGEX_MALLOC */
+
+
+/* True if `size1' is non-NULL and PTR is pointing anywhere inside
+ `string1' or just past its end. This works if PTR is NULL, which is
+ a good thing. */
+#define FIRST_STRING_P(ptr) \
+ (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
+
+/* (Re)Allocate N items of type T using malloc, or fail. */
+#define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
+#define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
+#define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
+
+#define BYTEWIDTH 8 /* In bits. */
+
+#define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
+
+#define MAX(a, b) ((a) > (b) ? (a) : (b))
+#define MIN(a, b) ((a) < (b) ? (a) : (b))
+
+typedef char boolean;
+#define false 0
+#define true 1
+
+/* These are the command codes that appear in compiled regular
+ expressions. Some opcodes are followed by argument bytes. A
+ command code can specify any interpretation whatsoever for its
+ arguments. Zero bytes may appear in the compiled regular expression.
+
+ The value of `exactn' is needed in search.c (search_buffer) in Emacs.
So regex.h defines a symbol `RE_EXACTN_VALUE' to be 1; the value of
`exactn' we use here must also be 1. */
-enum regexpcode
- {
- unused=0,
- exactn=1, /* Followed by one byte giving n, then by n literal bytes. */
- begline, /* Fail unless at beginning of line. */
- endline, /* Fail unless at end of line. */
- jump, /* Followed by two bytes giving relative address to jump to. */
- on_failure_jump, /* Followed by two bytes giving relative address of
- place to resume at in case of failure. */
- finalize_jump, /* Throw away latest failure point and then jump to
- address. */
- maybe_finalize_jump, /* Like jump but finalize if safe to do so.
- This is used to jump back to the beginning
- of a repeat. If the command that follows
- this jump is clearly incompatible with the
- one at the beginning of the repeat, such that
- we can be sure that there is no use backtracking
- out of repetitions already completed,
- then we finalize. */
- dummy_failure_jump, /* Jump, and push a dummy failure point. This
- failure point will be thrown away if an attempt
- is made to use it for a failure. A + construct
- makes this before the first repeat. Also
- use it as an intermediary kind of jump when
- compiling an or construct. */
- succeed_n, /* Used like on_failure_jump except has to succeed n times;
- then gets turned into an on_failure_jump. The relative
- address following it is useless until then. The
- address is followed by two bytes containing n. */
- jump_n, /* Similar to jump, but jump n times only; also the relative
- address following is in turn followed by yet two more bytes
- containing n. */
- set_number_at, /* Set the following relative location to the
- subsequent number. */
- anychar, /* Matches any (more or less) one character. */
- charset, /* Matches any one char belonging to specified set.
- First following byte is number of bitmap bytes.
- Then come bytes for a bitmap saying which chars are in.
- Bits in each byte are ordered low-bit-first.
- A character is in the set if its bit is 1.
- A character too large to have a bit in the map
- is automatically not in the set. */
- charset_not, /* Same parameters as charset, but match any character
- that is not one of those specified. */
- start_memory, /* Start remembering the text that is matched, for
- storing in a memory register. Followed by one
- byte containing the register number. Register numbers
- must be in the range 0 through RE_NREGS. */
- stop_memory, /* Stop remembering the text that is matched
- and store it in a memory register. Followed by
- one byte containing the register number. Register
- numbers must be in the range 0 through RE_NREGS. */
- duplicate, /* Match a duplicate of something remembered.
- Followed by one byte containing the index of the memory
- register. */
- before_dot, /* Succeeds if before point. */
- at_dot, /* Succeeds if at point. */
- after_dot, /* Succeeds if after point. */
- begbuf, /* Succeeds if at beginning of buffer. */
- endbuf, /* Succeeds if at end of buffer. */
- wordchar, /* Matches any word-constituent character. */
- notwordchar, /* Matches any char that is not a word-constituent. */
- wordbeg, /* Succeeds if at word beginning. */
- wordend, /* Succeeds if at word end. */
- wordbound, /* Succeeds if at a word boundary. */
- notwordbound,/* Succeeds if not at a word boundary. */
- syntaxspec, /* Matches any character whose syntax is specified.
- followed by a byte which contains a syntax code,
- e.g., Sword. */
- notsyntaxspec /* Matches any character whose syntax differs from
- that specified. */
- };
-
+typedef enum
+{
+ no_op = 0,
+
+ /* Followed by one byte giving n, then by n literal bytes. */
+ exactn = 1,
+
+ /* Matches any (more or less) character. */
+ anychar,
+
+ /* Matches any one char belonging to specified set. First
+ following byte is number of bitmap bytes. Then come bytes
+ for a bitmap saying which chars are in. Bits in each byte
+ are ordered low-bit-first. A character is in the set if its
+ bit is 1. A character too large to have a bit in the map is
+ automatically not in the set. */
+ charset,
+
+ /* Same parameters as charset, but match any character that is
+ not one of those specified. */
+ charset_not,
+
+ /* Start remembering the text that is matched, for storing in a
+ register. Followed by one byte with the register number, in
+ the range 0 to one less than the pattern buffer's re_nsub
+ field. Then followed by one byte with the number of groups
+ inner to this one. (This last has to be part of the
+ start_memory only because we need it in the on_failure_jump
+ of re_match_2.) */
+ start_memory,
+
+ /* Stop remembering the text that is matched and store it in a
+ memory register. Followed by one byte with the register
+ number, in the range 0 to one less than `re_nsub' in the
+ pattern buffer, and one byte with the number of inner groups,
+ just like `start_memory'. (We need the number of inner
+ groups here because we don't have any easy way of finding the
+ corresponding start_memory when we're at a stop_memory.) */
+ stop_memory,
+
+ /* Match a duplicate of something remembered. Followed by one
+ byte containing the register number. */
+ duplicate,
+
+ /* Fail unless at beginning of line. */
+ begline,
+
+ /* Fail unless at end of line. */
+ endline,
+
+ /* Succeeds if at beginning of buffer (if emacs) or at beginning
+ of string to be matched (if not). */
+ begbuf,
+
+ /* Analogously, for end of buffer/string. */
+ endbuf,
-/* Number of failure points to allocate space for initially,
- when matching. If this number is exceeded, more space is allocated,
- so it is not a hard limit. */
+ /* Followed by two byte relative address to which to jump. */
+ jump,
+
+ /* Same as jump, but marks the end of an alternative. */
+ jump_past_alt,
+
+ /* Followed by two-byte relative address of place to resume at
+ in case of failure. */
+ on_failure_jump,
+
+ /* Like on_failure_jump, but pushes a placeholder instead of the
+ current string position when executed. */
+ on_failure_keep_string_jump,
+
+ /* Throw away latest failure point and then jump to following
+ two-byte relative address. */
+ pop_failure_jump,
+
+ /* Change to pop_failure_jump if know won't have to backtrack to
+ match; otherwise change to jump. This is used to jump
+ back to the beginning of a repeat. If what follows this jump
+ clearly won't match what the repeat does, such that we can be
+ sure that there is no use backtracking out of repetitions
+ already matched, then we change it to a pop_failure_jump.
+ Followed by two-byte address. */
+ maybe_pop_jump,
+
+ /* Jump to following two-byte address, and push a dummy failure
+ point. This failure point will be thrown away if an attempt
+ is made to use it for a failure. A `+' construct makes this
+ before the first repeat. Also used as an intermediary kind
+ of jump when compiling an alternative. */
+ dummy_failure_jump,
+
+ /* Push a dummy failure point and continue. Used at the end of
+ alternatives. */
+ push_dummy_failure,
+
+ /* Followed by two-byte relative address and two-byte number n.
+ After matching N times, jump to the address upon failure. */
+ succeed_n,
+
+ /* Followed by two-byte relative address, and two-byte number n.
+ Jump to the address N times, then fail. */
+ jump_n,
+
+ /* Set the following two-byte relative address to the
+ subsequent two-byte number. The address *includes* the two
+ bytes of number. */
+ set_number_at,
+
+ wordchar, /* Matches any word-constituent character. */
+ notwordchar, /* Matches any char that is not a word-constituent. */
+
+ wordbeg, /* Succeeds if at word beginning. */
+ wordend, /* Succeeds if at word end. */
+
+ wordbound, /* Succeeds if at a word boundary. */
+ notwordbound /* Succeeds if not at a word boundary. */
-#ifndef NFAILURES
-#define NFAILURES 80
-#endif
+#ifdef emacs
+ ,before_dot, /* Succeeds if before point. */
+ at_dot, /* Succeeds if at point. */
+ after_dot, /* Succeeds if after point. */
-#ifdef CHAR_UNSIGNED
-#define SIGN_EXTEND_CHAR(c) ((c)>(char)127?(c)-256:(c)) /* for IBM RT */
-#endif
-#ifndef SIGN_EXTEND_CHAR
-#define SIGN_EXTEND_CHAR(x) (x)
-#endif
-
+ /* Matches any character whose syntax is specified. Followed by
+ a byte which contains a syntax code, e.g., Sword. */
+ syntaxspec,
+
+ /* Matches any character whose syntax is not that specified. */
+ notsyntaxspec
+#endif /* emacs */
+} re_opcode_t;
+
+/* Common operations on the compiled pattern. */
/* Store NUMBER in two contiguous bytes starting at DESTINATION. */
+
#define STORE_NUMBER(destination, number) \
- { (destination)[0] = (number) & 0377; \
- (destination)[1] = (number) >> 8; }
-
-/* Same as STORE_NUMBER, except increment the destination pointer to
- the byte after where the number is stored. Watch out that values for
- DESTINATION such as p + 1 won't work, whereas p will. */
-#define STORE_NUMBER_AND_INCR(destination, number) \
- { STORE_NUMBER(destination, number); \
- (destination) += 2; }
+ do { \
+ (destination)[0] = (number) & 0377; \
+ (destination)[1] = (number) >> 8; \
+ } while (0)
+/* Same as STORE_NUMBER, except increment DESTINATION to
+ the byte after where the number is stored. Therefore, DESTINATION
+ must be an lvalue. */
+
+#define STORE_NUMBER_AND_INCR(destination, number) \
+ do { \
+ STORE_NUMBER (destination, number); \
+ (destination) += 2; \
+ } while (0)
-/* Put into DESTINATION a number stored in two contingous bytes starting
+/* Put into DESTINATION a number stored in two contiguous bytes starting
at SOURCE. */
+
#define EXTRACT_NUMBER(destination, source) \
- { (destination) = *(source) & 0377; \
- (destination) += SIGN_EXTEND_CHAR (*(char *)((source) + 1)) << 8; }
+ do { \
+ (destination) = *(source) & 0377; \
+ (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
+ } while (0)
+
+#ifdef DEBUG
+static void
+extract_number (dest, source)
+ int *dest;
+ unsigned char *source;
+{
+ int temp = SIGN_EXTEND_CHAR (*(source + 1));
+ *dest = *source & 0377;
+ *dest += temp << 8;
+}
+
+#ifndef EXTRACT_MACROS /* To debug the macros. */
+#undef EXTRACT_NUMBER
+#define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
+#endif /* not EXTRACT_MACROS */
+
+#endif /* DEBUG */
+
+/* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
+ SOURCE must be an lvalue. */
-/* Same as EXTRACT_NUMBER, except increment the pointer for source to
- point to second byte of SOURCE. Note that SOURCE has to be a value
- such as p, not, e.g., p + 1. */
#define EXTRACT_NUMBER_AND_INCR(destination, source) \
- { EXTRACT_NUMBER (destination, source); \
- (source) += 2; }
+ do { \
+ EXTRACT_NUMBER (destination, source); \
+ (source) += 2; \
+ } while (0)
+#ifdef DEBUG
+static void
+extract_number_and_incr (destination, source)
+ int *destination;
+ unsigned char **source;
+{
+ extract_number (destination, *source);
+ *source += 2;
+}
-/* Specify the precise syntax of regexps for compilation. This provides
- for compatibility for various utilities which historically have
- different, incompatible syntaxes.
-
- The argument SYNTAX is a bit-mask comprised of the various bits
- defined in regex.h. */
+#ifndef EXTRACT_MACROS
+#undef EXTRACT_NUMBER_AND_INCR
+#define EXTRACT_NUMBER_AND_INCR(dest, src) \
+ extract_number_and_incr (&dest, &src)
+#endif /* not EXTRACT_MACROS */
-int
-re_set_syntax (syntax)
- int syntax;
+#endif /* DEBUG */
+
+/* If DEBUG is defined, Regex prints many voluminous messages about what
+ it is doing (if the variable `debug' is nonzero). If linked with the
+ main program in `iregex.c', you can enter patterns and strings
+ interactively. And if linked with the main program in `main.c' and
+ the other test files, you can run the already-written tests. */
+
+#ifdef DEBUG
+
+/* We use standard I/O for debugging. */
+#include <stdio.h>
+
+/* It is useful to test things that ``must'' be true when debugging. */
+#include <assert.h>
+
+static int debug = 0;
+
+#define DEBUG_STATEMENT(e) e
+#define DEBUG_PRINT1(x) if (debug) printf (x)
+#define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
+#define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
+#define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
+#define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
+ if (debug) print_partial_compiled_pattern (s, e)
+#define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
+ if (debug) print_double_string (w, s1, sz1, s2, sz2)
+
+
+extern void printchar ();
+
+/* Print the fastmap in human-readable form. */
+
+void
+print_fastmap (fastmap)
+ char *fastmap;
{
- int ret;
+ unsigned was_a_range = 0;
+ unsigned i = 0;
+
+ while (i < (1 << BYTEWIDTH))
+ {
+ if (fastmap[i++])
+ {
+ was_a_range = 0;
+ printchar (i - 1);
+ while (i < (1 << BYTEWIDTH) && fastmap[i])
+ {
+ was_a_range = 1;
+ i++;
+ }
+ if (was_a_range)
+ {
+ printf ("-");
+ printchar (i - 1);
+ }
+ }
+ }
+ putchar ('\n');
+}
- ret = obscure_syntax;
- obscure_syntax = syntax;
- return ret;
+
+/* Print a compiled pattern string in human-readable form, starting at
+ the START pointer into it and ending just before the pointer END. */
+
+void
+print_partial_compiled_pattern (start, end)
+ unsigned char *start;
+ unsigned char *end;
+{
+ int mcnt, mcnt2;
+ unsigned char *p = start;
+ unsigned char *pend = end;
+
+ if (start == NULL)
+ {
+ printf ("(null)\n");
+ return;
+ }
+
+ /* Loop over pattern commands. */
+ while (p < pend)
+ {
+ printf ("%d:\t", p - start);
+
+ switch ((re_opcode_t) *p++)
+ {
+ case no_op:
+ printf ("/no_op");
+ break;
+
+ case exactn:
+ mcnt = *p++;
+ printf ("/exactn/%d", mcnt);
+ do
+ {
+ putchar ('/');
+ printchar (*p++);
+ }
+ while (--mcnt);
+ break;
+
+ case start_memory:
+ mcnt = *p++;
+ printf ("/start_memory/%d/%d", mcnt, *p++);
+ break;
+
+ case stop_memory:
+ mcnt = *p++;
+ printf ("/stop_memory/%d/%d", mcnt, *p++);
+ break;
+
+ case duplicate:
+ printf ("/duplicate/%d", *p++);
+ break;
+
+ case anychar:
+ printf ("/anychar");
+ break;
+
+ case charset:
+ case charset_not:
+ {
+ register int c, last = -100;
+ register int in_range = 0;
+
+ printf ("/charset [%s",
+ (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
+
+ assert (p + *p < pend);
+
+ for (c = 0; c < 256; c++)
+ if (c / 8 < *p
+ && (p[1 + (c/8)] & (1 << (c % 8))))
+ {
+ /* Are we starting a range? */
+ if (last + 1 == c && ! in_range)
+ {
+ putchar ('-');
+ in_range = 1;
+ }
+ /* Have we broken a range? */
+ else if (last + 1 != c && in_range)
+ {
+ printchar (last);
+ in_range = 0;
+ }
+
+ if (! in_range)
+ printchar (c);
+
+ last = c;
+ }
+
+ if (in_range)
+ printchar (last);
+
+ putchar (']');
+
+ p += 1 + *p;
+ }
+ break;
+
+ case begline:
+ printf ("/begline");
+ break;
+
+ case endline:
+ printf ("/endline");
+ break;
+
+ case on_failure_jump:
+ extract_number_and_incr (&mcnt, &p);
+ printf ("/on_failure_jump to %d", p + mcnt - start);
+ break;
+
+ case on_failure_keep_string_jump:
+ extract_number_and_incr (&mcnt, &p);
+ printf ("/on_failure_keep_string_jump to %d", p + mcnt - start);
+ break;
+
+ case dummy_failure_jump:
+ extract_number_and_incr (&mcnt, &p);
+ printf ("/dummy_failure_jump to %d", p + mcnt - start);
+ break;
+
+ case push_dummy_failure:
+ printf ("/push_dummy_failure");
+ break;
+
+ case maybe_pop_jump:
+ extract_number_and_incr (&mcnt, &p);
+ printf ("/maybe_pop_jump to %d", p + mcnt - start);
+ break;
+
+ case pop_failure_jump:
+ extract_number_and_incr (&mcnt, &p);
+ printf ("/pop_failure_jump to %d", p + mcnt - start);
+ break;
+
+ case jump_past_alt:
+ extract_number_and_incr (&mcnt, &p);
+ printf ("/jump_past_alt to %d", p + mcnt - start);
+ break;
+
+ case jump:
+ extract_number_and_incr (&mcnt, &p);
+ printf ("/jump to %d", p + mcnt - start);
+ break;
+
+ case succeed_n:
+ extract_number_and_incr (&mcnt, &p);
+ extract_number_and_incr (&mcnt2, &p);
+ printf ("/succeed_n to %d, %d times", p + mcnt - start, mcnt2);
+ break;
+
+ case jump_n:
+ extract_number_and_incr (&mcnt, &p);
+ extract_number_and_incr (&mcnt2, &p);
+ printf ("/jump_n to %d, %d times", p + mcnt - start, mcnt2);
+ break;
+
+ case set_number_at:
+ extract_number_and_incr (&mcnt, &p);
+ extract_number_and_incr (&mcnt2, &p);
+ printf ("/set_number_at location %d to %d", p + mcnt - start, mcnt2);
+ break;
+
+ case wordbound:
+ printf ("/wordbound");
+ break;
+
+ case notwordbound:
+ printf ("/notwordbound");
+ break;
+
+ case wordbeg:
+ printf ("/wordbeg");
+ break;
+
+ case wordend:
+ printf ("/wordend");
+
+#ifdef emacs
+ case before_dot:
+ printf ("/before_dot");
+ break;
+
+ case at_dot:
+ printf ("/at_dot");
+ break;
+
+ case after_dot:
+ printf ("/after_dot");
+ break;
+
+ case syntaxspec:
+ printf ("/syntaxspec");
+ mcnt = *p++;
+ printf ("/%d", mcnt);
+ break;
+
+ case notsyntaxspec:
+ printf ("/notsyntaxspec");
+ mcnt = *p++;
+ printf ("/%d", mcnt);
+ break;
+#endif /* emacs */
+
+ case wordchar:
+ printf ("/wordchar");
+ break;
+
+ case notwordchar:
+ printf ("/notwordchar");
+ break;
+
+ case begbuf:
+ printf ("/begbuf");
+ break;
+
+ case endbuf:
+ printf ("/endbuf");
+ break;
+
+ default:
+ printf ("?%d", *(p-1));
+ }
+
+ putchar ('\n');
+ }
+
+ printf ("%d:\tend of pattern.\n", p - start);
}
-/* Set by re_set_syntax to the current regexp syntax to recognize. */
-int obscure_syntax = 0;
+void
+print_compiled_pattern (bufp)
+ struct re_pattern_buffer *bufp;
+{
+ unsigned char *buffer = bufp->buffer;
+
+ print_partial_compiled_pattern (buffer, buffer + bufp->used);
+ printf ("%d bytes used/%d bytes allocated.\n", bufp->used, bufp->allocated);
+
+ if (bufp->fastmap_accurate && bufp->fastmap)
+ {
+ printf ("fastmap: ");
+ print_fastmap (bufp->fastmap);
+ }
+
+ printf ("re_nsub: %d\t", bufp->re_nsub);
+ printf ("regs_alloc: %d\t", bufp->regs_allocated);
+ printf ("can_be_null: %d\t", bufp->can_be_null);
+ printf ("newline_anchor: %d\n", bufp->newline_anchor);
+ printf ("no_sub: %d\t", bufp->no_sub);
+ printf ("not_bol: %d\t", bufp->not_bol);
+ printf ("not_eol: %d\t", bufp->not_eol);
+ printf ("syntax: %d\n", bufp->syntax);
+ /* Perhaps we should print the translate table? */
+}
+
+void
+print_double_string (where, string1, size1, string2, size2)
+ const char *where;
+ const char *string1;
+ const char *string2;
+ int size1;
+ int size2;
+{
+ unsigned this_char;
+
+ if (where == NULL)
+ printf ("(null)");
+ else
+ {
+ if (FIRST_STRING_P (where))
+ {
+ for (this_char = where - string1; this_char < size1; this_char++)
+ printchar (string1[this_char]);
+
+ where = string2;
+ }
+
+ for (this_char = where - string2; this_char < size2; this_char++)
+ printchar (string2[this_char]);
+ }
+}
+
+#else /* not DEBUG */
+
+#undef assert
+#define assert(e)
+
+#define DEBUG_STATEMENT(e)
+#define DEBUG_PRINT1(x)
+#define DEBUG_PRINT2(x1, x2)
+#define DEBUG_PRINT3(x1, x2, x3)
+#define DEBUG_PRINT4(x1, x2, x3, x4)
+#define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
+#define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
+
+#endif /* not DEBUG */
-/* Macros for re_compile_pattern, which is found below these definitions. */
+/* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
+ also be assigned to arbitrarily: each pattern buffer stores its own
+ syntax, so it can be changed between regex compilations. */
+reg_syntax_t re_syntax_options = RE_SYNTAX_EMACS;
-#define CHAR_CLASS_MAX_LENGTH 6
-/* Fetch the next character in the uncompiled pattern, translating it if
- necessary. */
+/* Specify the precise syntax of regexps for compilation. This provides
+ for compatibility for various utilities which historically have
+ different, incompatible syntaxes.
+
+ The argument SYNTAX is a bit mask comprised of the various bits
+ defined in regex.h. We return the old syntax. */
+
+reg_syntax_t
+re_set_syntax (syntax)
+ reg_syntax_t syntax;
+{
+ reg_syntax_t ret = re_syntax_options;
+
+ re_syntax_options = syntax;
+ return ret;
+}
+
+/* This table gives an error message for each of the error codes listed
+ in regex.h. Obviously the order here has to be same as there. */
+
+static const char *re_error_msg[] =
+ { NULL, /* REG_NOERROR */
+ "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 */
+ };
+
+/* Subroutine declarations and macros for regex_compile. */
+
+static void store_op1 (), store_op2 ();
+static void insert_op1 (), insert_op2 ();
+static boolean at_begline_loc_p (), at_endline_loc_p ();
+static boolean group_in_compile_stack ();
+static reg_errcode_t compile_range ();
+
+/* 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) \
- {if (p == pend) goto end_of_pattern; \
- c = * (unsigned char *) p++; \
- if (translate) c = translate[c]; }
+ do {if (p == pend) return REG_EEND; \
+ c = (unsigned char) *p++; \
+ if (translate) c = translate[c]; \
+ } while (0)
/* Fetch the next character in the uncompiled pattern, with no
translation. */
#define PATFETCH_RAW(c) \
- {if (p == pend) goto end_of_pattern; \
- c = * (unsigned char *) p++; }
+ do {if (p == pend) return REG_EEND; \
+ c = (unsigned char) *p++; \
+ } while (0)
+/* Go backwards one character in the pattern. */
#define PATUNFETCH p--
+/* If `translate' is non-null, return translate[D], else just D. We
+ cast the subscript to translate because some data is declared as
+ `char *', to avoid warnings when a string constant is passed. But
+ when we use a character as a subscript we must make it unsigned. */
+#define TRANSLATE(d) (translate ? translate[(unsigned char) (d)] : (d))
+
+
+/* Macros for outputting the compiled pattern into `buffer'. */
+
/* If the buffer isn't allocated when it comes in, use this. */
-#define INIT_BUF_SIZE 28
+#define INIT_BUF_SIZE 32
/* Make sure we have at least N more bytes of space in buffer. */
#define GET_BUFFER_SPACE(n) \
- { \
- while (b - bufp->buffer + (n) >= bufp->allocated) \
- EXTEND_BUFFER; \
- }
+ while (b - bufp->buffer + (n) > bufp->allocated) \
+ EXTEND_BUFFER ()
-/* Make sure we have one more byte of buffer space and then add CH to it. */
-#define BUFPUSH(ch) \
- { \
+/* Make sure we have one more byte of buffer space and then add C to it. */
+#define BUF_PUSH(c) \
+ do { \
GET_BUFFER_SPACE (1); \
- *b++ = (char) (ch); \
- }
-
-/* Extend the buffer by twice its current size via reallociation and
- reset the pointers that pointed into the old allocation to point to
- the correct places in the new allocation. If extending the buffer
- results in it being larger than 1 << 16, then flag memory exhausted. */
-#define EXTEND_BUFFER \
- { char *old_buffer = bufp->buffer; \
- if (bufp->allocated == (1L<<16)) goto too_big; \
- bufp->allocated *= 2; \
- if (bufp->allocated > (1L<<16)) bufp->allocated = (1L<<16); \
- bufp->buffer = (char *) realloc (bufp->buffer, bufp->allocated); \
- if (bufp->buffer == 0) \
- goto memory_exhausted; \
- b = (b - old_buffer) + bufp->buffer; \
- if (fixup_jump) \
- fixup_jump = (fixup_jump - old_buffer) + bufp->buffer; \
- if (laststart) \
- laststart = (laststart - old_buffer) + bufp->buffer; \
- begalt = (begalt - old_buffer) + bufp->buffer; \
- if (pending_exact) \
- pending_exact = (pending_exact - old_buffer) + bufp->buffer; \
- }
+ *b++ = (unsigned char) (c); \
+ } while (0)
+
+
+/* Ensure we have two more bytes of buffer space and then append C1 and C2. */
+#define BUF_PUSH_2(c1, c2) \
+ do { \
+ GET_BUFFER_SPACE (2); \
+ *b++ = (unsigned char) (c1); \
+ *b++ = (unsigned char) (c2); \
+ } while (0)
+
+
+/* As with BUF_PUSH_2, except for three bytes. */
+#define BUF_PUSH_3(c1, c2, c3) \
+ do { \
+ GET_BUFFER_SPACE (3); \
+ *b++ = (unsigned char) (c1); \
+ *b++ = (unsigned char) (c2); \
+ *b++ = (unsigned char) (c3); \
+ } while (0)
+
+
+/* Store a jump with opcode OP at LOC to location TO. We store a
+ relative address offset by the three bytes the jump itself occupies. */
+#define STORE_JUMP(op, loc, to) \
+ store_op1 (op, loc, (to) - (loc) - 3)
+
+/* Likewise, for a two-argument jump. */
+#define STORE_JUMP2(op, loc, to, arg) \
+ store_op2 (op, loc, (to) - (loc) - 3, arg)
+
+/* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
+#define INSERT_JUMP(op, loc, to) \
+ insert_op1 (op, loc, (to) - (loc) - 3, b)
+
+/* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
+#define INSERT_JUMP2(op, loc, to, arg) \
+ insert_op2 (op, loc, (to) - (loc) - 3, arg, b)
+
+
+/* This is not an arbitrary limit: the arguments which represent offsets
+ into the pattern are two bytes long. So if 2^16 bytes turns out to
+ be too small, many things would have to change. */
+#define MAX_BUF_SIZE (1L << 16)
+
+
+/* Extend the buffer by twice its current size via realloc and
+ reset the pointers that pointed into the old block to point to the
+ correct places in the new one. If extending the buffer results in it
+ being larger than MAX_BUF_SIZE, then flag memory exhausted. */
+#define EXTEND_BUFFER() \
+ do { \
+ unsigned char *old_buffer = bufp->buffer; \
+ if (bufp->allocated == MAX_BUF_SIZE) \
+ return REG_ESIZE; \
+ bufp->allocated <<= 1; \
+ if (bufp->allocated > MAX_BUF_SIZE) \
+ bufp->allocated = MAX_BUF_SIZE; \
+ bufp->buffer = (unsigned char *) realloc (bufp->buffer, bufp->allocated);\
+ if (bufp->buffer == NULL) \
+ return REG_ESPACE; \
+ /* If the buffer moved, move all the pointers into it. */ \
+ if (old_buffer != bufp->buffer) \
+ { \
+ b = (b - old_buffer) + bufp->buffer; \
+ begalt = (begalt - old_buffer) + bufp->buffer; \
+ if (fixup_alt_jump) \
+ fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\
+ if (laststart) \
+ laststart = (laststart - old_buffer) + bufp->buffer; \
+ if (pending_exact) \
+ pending_exact = (pending_exact - old_buffer) + bufp->buffer; \
+ } \
+ } while (0)
-/* Set the bit for character C in a character set list. */
-#define SET_LIST_BIT(c) (b[(c) / BYTEWIDTH] |= 1 << ((c) % BYTEWIDTH))
-/* 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); \
- } \
- } \
- }
+/* Since we have one byte reserved for the register number argument to
+ {start,stop}_memory, the maximum number of groups we can report
+ things about is what fits in that byte. */
+#define MAX_REGNUM 255
-/* Subroutines for re_compile_pattern. */
-static void store_jump (), insert_jump (), store_jump_n (),
- insert_jump_n (), insert_op_2 ();
+/* But patterns can have more than `MAX_REGNUM' registers. We just
+ ignore the excess. */
+typedef unsigned regnum_t;
-/* re_compile_pattern takes a regular-expression string
- and converts it into a buffer full of byte commands for matching.
+/* Macros for the compile stack. */
- PATTERN is the address of the pattern string
- SIZE is the length of it.
- BUFP is a struct re_pattern_buffer * which points to the info
- on where to store the byte commands.
- This structure contains a char * which points to the
- actual space, which should have been obtained with malloc.
- re_compile_pattern may use realloc to grow the buffer space.
+/* 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;
- The number of bytes of commands can be found out by looking in
- the `struct re_pattern_buffer' that bufp pointed to, after
- re_compile_pattern returns. */
+typedef struct
+{
+ pattern_offset_t begalt_offset;
+ pattern_offset_t fixup_alt_jump;
+ pattern_offset_t inner_group_offset;
+ pattern_offset_t laststart_offset;
+ regnum_t regnum;
+} compile_stack_elt_t;
-char *
-re_compile_pattern (pattern, size, bufp)
- char *pattern;
- int size;
- struct re_pattern_buffer *bufp;
+
+typedef struct
{
- register char *b = bufp->buffer;
- register char *p = pattern;
- char *pend = pattern + size;
- register unsigned c, c1;
- char *p1;
- unsigned char *translate = (unsigned char *) bufp->translate;
+ compile_stack_elt_t *stack;
+ unsigned size;
+ unsigned avail; /* Offset of next open position. */
+} compile_stack_type;
- /* Address of the count-byte of the most recently inserted `exactn'
- command. This makes it possible to tell whether a new exact-match
- character can be added to that command or requires a new `exactn'
- command. */
-
- char *pending_exact = 0;
- /* Address of the place where a forward-jump should go to the end of
- the containing expression. Each alternative of an `or', except the
- last, ends with a forward-jump of this sort. */
+#define INIT_COMPILE_STACK_SIZE 32
- char *fixup_jump = 0;
+#define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
+#define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
- /* Address of start of the most recently finished expression.
- This tells postfix * where to find the start of its operand. */
+/* The next available element. */
+#define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
- char *laststart = 0;
- /* In processing a repeat, 1 means zero matches is allowed. */
+/* Set the bit for character C in a list. */
+#define SET_LIST_BIT(c) \
+ (b[((unsigned char) (c)) / BYTEWIDTH] \
+ |= 1 << (((unsigned char) c) % BYTEWIDTH))
- char zero_times_ok;
- /* In processing a repeat, 1 means many matches is allowed. */
+/* 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) \
+ (STREQ (string, "alpha") || STREQ (string, "upper") \
+ || STREQ (string, "lower") || STREQ (string, "digit") \
+ || STREQ (string, "alnum") || STREQ (string, "xdigit") \
+ || STREQ (string, "space") || STREQ (string, "print") \
+ || STREQ (string, "punct") || STREQ (string, "graph") \
+ || STREQ (string, "cntrl") || STREQ (string, "blank"))
+
+/* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
+ Returns one of error codes defined in `regex.h', or zero for success.
+
+ Assumes the `allocated' (and perhaps `buffer') and `translate'
+ fields are set in BUFP on entry.
+
+ If it succeeds, results are put in BUFP (if it returns an error, the
+ contents of BUFP are undefined):
+ `buffer' is the compiled pattern;
+ `syntax' is set to SYNTAX;
+ `used' is set to the length of the compiled pattern;
+ `fastmap_accurate' is zero;
+ `re_nsub' is the number of subexpressions in PATTERN;
+ `not_bol' and `not_eol' are zero;
+
+ The `fastmap' and `newline_anchor' fields are neither
+ examined nor set. */
- char many_times_ok;
+static reg_errcode_t
+regex_compile (pattern, size, syntax, bufp)
+ const char *pattern;
+ int size;
+ reg_syntax_t syntax;
+ struct re_pattern_buffer *bufp;
+{
+ /* 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, c1;
+
+ /* A random tempory spot in PATTERN. */
+ const char *p1;
- /* Address of beginning of regexp, or inside of last \(. */
+ /* Points to the end of the buffer, where we should append. */
+ register unsigned char *b;
+
+ /* Keeps track of unclosed groups. */
+ compile_stack_type compile_stack;
- char *begalt = b;
+ /* Points to the current (ending) position in the pattern. */
+ const char *p = pattern;
+ const char *pend = pattern + size;
+
+ /* How to translate the characters in the pattern. */
+ char *translate = bufp->translate;
- /* In processing an interval, at least this many matches must be made. */
- int lower_bound;
+ /* Address of the count-byte of the most recently inserted `exactn'
+ command. This makes it possible to tell if a new exact-match
+ character can be added to that command or if the character requires
+ a new `exactn' command. */
+ unsigned char *pending_exact = 0;
- /* In processing an interval, at most this many matches can be made. */
- int upper_bound;
+ /* Address of start of the most recently finished expression.
+ This tells, e.g., postfix * where to find the start of its
+ operand. Reset at the beginning of groups and alternatives. */
+ unsigned char *laststart = 0;
- /* Place in pattern (i.e., the {) to which to go back if the interval
- is invalid. */
- char *beg_interval = 0;
-
- /* Stack of information saved by \( and restored by \).
- Four stack elements are pushed by each \(:
- First, the value of b.
- Second, the value of fixup_jump.
- Third, the value of regnum.
- Fourth, the value of begalt. */
+ /* Address of beginning of regexp, or inside of last group. */
+ unsigned char *begalt;
- int stackb[40];
- int *stackp = stackb;
- int *stacke = stackb + 40;
- int *stackt;
+ /* Place in the uncompiled pattern (i.e., the {) to
+ which to go back if the interval is invalid. */
+ const char *beg_interval;
+
+ /* Address of the place where a forward jump should go to the end of
+ the containing expression. Each alternative of an `or' -- except the
+ last -- ends with a forward jump of this sort. */
+ unsigned char *fixup_alt_jump = 0;
+
+ /* Counts open-groups as they are encountered. Remembered for the
+ matching close-group on the compile stack, so the same register
+ number is put in the stop_memory as the start_memory. */
+ regnum_t regnum = 0;
+
+#ifdef DEBUG
+ DEBUG_PRINT1 ("\nCompiling pattern: ");
+ if (debug)
+ {
+ unsigned debug_count;
+
+ for (debug_count = 0; debug_count < size; debug_count++)
+ printchar (pattern[debug_count]);
+ putchar ('\n');
+ }
+#endif /* DEBUG */
- /* Counts \('s as they are encountered. Remembered for the matching \),
- where it becomes the register number to put in the stop_memory
- command. */
+ /* Initialize the compile stack. */
+ compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
+ if (compile_stack.stack == NULL)
+ return REG_ESPACE;
- int regnum = 1;
+ compile_stack.size = INIT_COMPILE_STACK_SIZE;
+ compile_stack.avail = 0;
+ /* Initialize the pattern buffer. */
+ bufp->syntax = syntax;
bufp->fastmap_accurate = 0;
+ bufp->not_bol = bufp->not_eol = 0;
-#ifndef emacs
-#ifndef SYNTAX_TABLE
+ /* Set `used' to zero, so that if we return an error, the pattern
+ printer (for debugging) will think there's no pattern. We reset it
+ at the end. */
+ bufp->used = 0;
+
+ /* Always count groups, whether or not bufp->no_sub is set. */
+ bufp->re_nsub = 0;
+
+#if !defined (emacs) && !defined (SYNTAX_TABLE)
/* Initialize the syntax table. */
- init_syntax_once();
-#endif
+ init_syntax_once ();
#endif
if (bufp->allocated == 0)
{
- bufp->allocated = INIT_BUF_SIZE;
if (bufp->buffer)
- /* EXTEND_BUFFER loses when bufp->allocated is 0. */
- bufp->buffer = (char *) realloc (bufp->buffer, INIT_BUF_SIZE);
+ { /* If zero allocated, but buffer is non-null, try to realloc
+ enough space. This loses if buffer's address is bogus, but
+ that is the user's responsibility. */
+ RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
+ }
else
- /* Caller did not allocate a buffer. Do it for them. */
- bufp->buffer = (char *) malloc (INIT_BUF_SIZE);
- if (!bufp->buffer) goto memory_exhausted;
- begalt = b = bufp->buffer;
+ { /* Caller did not allocate a buffer. Do it for them. */
+ bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
+ }
+ if (!bufp->buffer) return REG_ESPACE;
+
+ bufp->allocated = INIT_BUF_SIZE;
}
+ begalt = b = bufp->buffer;
+
+ /* Loop through the uncompiled pattern until we're at the end. */
while (p != pend)
{
PATFETCH (c);
switch (c)
- {
- case '$':
- {
- char *p1 = p;
- /* When testing what follows the $,
- look past the \-constructs that don't consume anything. */
- if (! (obscure_syntax & RE_CONTEXT_INDEP_OPS))
- while (p1 != pend)
- {
- if (*p1 == '\\' && p1 + 1 != pend
- && (p1[1] == '<' || p1[1] == '>'
- || p1[1] == '`' || p1[1] == '\''
-#ifdef emacs
- || p1[1] == '='
-#endif
- || p1[1] == 'b' || p1[1] == 'B'))
- p1 += 2;
- else
- break;
- }
- if (obscure_syntax & RE_TIGHT_VBAR)
- {
- if (! (obscure_syntax & RE_CONTEXT_INDEP_OPS) && p1 != pend)
- goto normal_char;
- /* Make operand of last vbar end before this `$'. */
- if (fixup_jump)
- store_jump (fixup_jump, jump, b);
- fixup_jump = 0;
- BUFPUSH (endline);
- break;
- }
- /* $ means succeed if at end of line, but only in special contexts.
- If validly in the middle of a pattern, it is a normal character. */
-
- if ((obscure_syntax & RE_CONTEXTUAL_INVALID_OPS) && p1 != pend)
- goto invalid_pattern;
- if (p1 == pend || *p1 == '\n'
- || (obscure_syntax & RE_CONTEXT_INDEP_OPS)
- || (obscure_syntax & RE_NO_BK_PARENS
- ? *p1 == ')'
- : *p1 == '\\' && p1[1] == ')')
- || (obscure_syntax & RE_NO_BK_VBAR
- ? *p1 == '|'
- : *p1 == '\\' && p1[1] == '|'))
- {
- BUFPUSH (endline);
- break;
- }
- goto normal_char;
+ {
+ 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))
+ BUF_PUSH (begline);
+ else
+ goto normal_char;
}
- case '^':
- /* ^ means succeed if at beg of line, but only if no preceding
- pattern. */
-
- if ((obscure_syntax & RE_CONTEXTUAL_INVALID_OPS) && laststart)
- goto invalid_pattern;
- if (laststart && p - 2 >= pattern && p[-2] != '\n'
- && !(obscure_syntax & RE_CONTEXT_INDEP_OPS))
- goto normal_char;
- if (obscure_syntax & RE_TIGHT_VBAR)
- {
- if (p != pattern + 1
- && ! (obscure_syntax & RE_CONTEXT_INDEP_OPS))
- goto normal_char;
- BUFPUSH (begline);
- begalt = b;
- }
- else
- BUFPUSH (begline);
- break;
+ 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))
+ BUF_PUSH (endline);
+ else
+ goto normal_char;
+ }
+ break;
+
case '+':
- case '?':
- if ((obscure_syntax & RE_BK_PLUS_QM)
- || (obscure_syntax & RE_LIMITED_OPS))
- goto normal_char;
- handle_plus:
- case '*':
- /* If there is no previous pattern, char not special. */
- if (!laststart)
+ case '?':
+ if ((syntax & RE_BK_PLUS_QM)
+ || (syntax & RE_LIMITED_OPS))
+ goto normal_char;
+ handle_plus:
+ case '*':
+ /* If there is no previous pattern... */
+ if (!laststart)
{
- if (obscure_syntax & RE_CONTEXTUAL_INVALID_OPS)
- goto invalid_pattern;
- else if (! (obscure_syntax & RE_CONTEXT_INDEP_OPS))
- goto normal_char;
+ if (syntax & RE_CONTEXT_INVALID_OPS)
+ return REG_BADRPT;
+ else if (!(syntax & RE_CONTEXT_INDEP_OPS))
+ goto normal_char;
}
- /* If there is a sequence of repetition chars,
- collapse it down to just one. */
- zero_times_ok = 0;
- many_times_ok = 0;
- while (1)
- {
- zero_times_ok |= c != '+';
- many_times_ok |= c != '?';
- if (p == pend)
- break;
- PATFETCH (c);
- if (c == '*')
- ;
- else if (!(obscure_syntax & RE_BK_PLUS_QM)
- && (c == '+' || c == '?'))
- ;
- else if ((obscure_syntax & RE_BK_PLUS_QM)
- && c == '\\')
- {
- int c1;
- PATFETCH (c1);
- if (!(c1 == '+' || c1 == '?'))
- {
- PATUNFETCH;
- PATUNFETCH;
- break;
- }
- c = c1;
- }
- else
- {
- PATUNFETCH;
- break;
- }
- }
- /* Star, etc. applied to an empty pattern is equivalent
- to an empty pattern. */
- if (!laststart)
- break;
+ {
+ /* Are we optimizing this jump? */
+ boolean keep_string_p = false;
+
+ /* 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) return REG_EESCAPE;
+
+ PATFETCH (c1);
+ if (!(c1 == '+' || c1 == '?'))
+ {
+ PATUNFETCH;
+ PATUNFETCH;
+ break;
+ }
+
+ c = c1;
+ }
+ else
+ {
+ PATUNFETCH;
+ break;
+ }
- /* Now we know whether or not zero matches is allowed
- and also whether or not two or more matches is allowed. */
- if (many_times_ok)
- {
- /* If more than one repetition is allowed, put in at the
- end a backward relative jump from b to before the next
- jump we're going to put in below (which jumps from
- laststart to after this jump). */
- GET_BUFFER_SPACE (3);
- store_jump (b, maybe_finalize_jump, laststart - 3);
- b += 3; /* Because store_jump put stuff here. */
- }
- /* On failure, jump from laststart to b + 3, which will be the
- end of the buffer after this jump is inserted. */
- GET_BUFFER_SPACE (3);
- insert_jump (on_failure_jump, laststart, b + 3, b);
- pending_exact = 0;
- b += 3;
- if (!zero_times_ok)
- {
- /* At least one repetition is required, so insert a
- dummy-failure before the initial on-failure-jump
- instruction of the loop. This effects a skip over that
- instruction the first time we hit that loop. */
- GET_BUFFER_SPACE (6);
- insert_jump (dummy_failure_jump, laststart, laststart + 6, b);
- b += 3;
- }
+ /* If we get here, we found another repeat character. */
+ }
+
+ /* Star, etc. applied to an empty pattern is equivalent
+ to an empty pattern. */
+ if (!laststart)
+ break;
+
+ /* Now we know whether or not zero matches is allowed
+ and also whether or not two or more matches is allowed. */
+ if (many_times_ok)
+ { /* More than one repetition is allowed, so put in at the
+ end a backward relative jump from `b' to before the next
+ jump we're going to put in below (which jumps from
+ laststart to after this jump).
+
+ But if we are at the `*' in the exact sequence `.*\n',
+ insert an unconditional jump backwards to the .,
+ instead of the beginning of the loop. This way we only
+ push a failure point once, instead of every time
+ through the loop. */
+ assert (p - 1 > pattern);
+
+ /* Allocate the space for the jump. */
+ GET_BUFFER_SPACE (3);
+
+ /* We know we are not at the first character of the pattern,
+ because laststart was nonzero. And we've already
+ incremented `p', by the way, to be the character after
+ the `*'. Do we have to do something analogous here
+ for null bytes, because of RE_DOT_NOT_NULL? */
+ if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
+ && zero_times_ok
+ && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
+ && !(syntax & RE_DOT_NEWLINE))
+ { /* We have .*\n. */
+ STORE_JUMP (jump, b, laststart);
+ keep_string_p = true;
+ }
+ else
+ /* Anything else. */
+ STORE_JUMP (maybe_pop_jump, b, laststart - 3);
+
+ /* We've added more stuff to the buffer. */
+ b += 3;
+ }
+
+ /* On failure, jump from laststart to b + 3, which will be the
+ end of the buffer after this jump is inserted. */
+ GET_BUFFER_SPACE (3);
+ INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
+ : on_failure_jump,
+ laststart, b + 3);
+ pending_exact = 0;
+ b += 3;
+
+ if (!zero_times_ok)
+ {
+ /* At least one repetition is required, so insert a
+ `dummy_failure_jump' before the initial
+ `on_failure_jump' instruction of the loop. This
+ effects a skip over that instruction the first time
+ we hit that loop. */
+ GET_BUFFER_SPACE (3);
+ INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6);
+ b += 3;
+ }
+ }
break;
+
case '.':
- laststart = b;
- BUFPUSH (anychar);
- break;
+ laststart = b;
+ BUF_PUSH (anychar);
+ break;
+
case '[':
- if (p == pend)
- goto invalid_pattern;
- while (b - bufp->buffer
- > bufp->allocated - 3 - (1 << BYTEWIDTH) / BYTEWIDTH)
- EXTEND_BUFFER;
-
- laststart = b;
- if (*p == '^')
- {
- BUFPUSH (charset_not);
- p++;
- }
- else
- BUFPUSH (charset);
- p1 = p;
+ {
+ boolean had_char_class = false;
- BUFPUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
- /* Clear the whole map */
- bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
-
- if ((obscure_syntax & RE_HAT_NOT_NEWLINE) && b[-2] == charset_not)
- SET_LIST_BIT ('\n');
+ if (p == pend) return REG_EBRACK;
+ /* Ensure that we have enough space to push a charset: the
+ opcode, the length count, and the bitset; 34 bytes in all. */
+ GET_BUFFER_SPACE (34);
- /* Read in characters and ranges, setting map bits. */
- while (1)
- {
- PATFETCH (c);
-
- /* If set, \ escapes characters when inside [...]. */
- if ((obscure_syntax & RE_AWK_CLASS_HACK) && c == '\\')
- {
- PATFETCH(c1);
- SET_LIST_BIT (c1);
- continue;
- }
- if (c == ']')
- {
- if (p == p1 + 1)
- {
- /* If this is an empty bracket expression. */
- if ((obscure_syntax & RE_NO_EMPTY_BRACKETS)
- && p == pend)
- goto invalid_pattern;
- }
- else
- /* Stop if this isn't merely a ] inside a bracket
- expression, but rather the end of a bracket
- expression. */
- break;
- }
- /* Get a range. */
- if (p[0] == '-' && p[1] != ']')
- {
- PATFETCH (c1);
- PATFETCH (c1);
-
- if ((obscure_syntax & RE_NO_EMPTY_RANGES) && c > c1)
- goto invalid_pattern;
-
- if ((obscure_syntax & RE_NO_HYPHEN_RANGE_END)
- && c1 == '-' && *p != ']')
- goto invalid_pattern;
-
- while (c <= c1)
- {
- SET_LIST_BIT (c);
- c++;
- }
- }
- else if ((obscure_syntax & RE_CHAR_CLASSES)
- && c == '[' && p[0] == ':')
- {
- /* Longest valid character class word has six characters. */
- char str[CHAR_CLASS_MAX_LENGTH];
- PATFETCH (c);
- c1 = 0;
- /* If no ] at end. */
- if (p == pend)
- goto invalid_pattern;
- while (1)
- {
- /* Don't translate the ``character class'' characters. */
- PATFETCH_RAW (c);
- if (c == ':' || c == ']' || p == pend
- || c1 == CHAR_CLASS_MAX_LENGTH)
- break;
- str[c1++] = c;
- }
- str[c1] = '\0';
- if (p == pend
- || c == ']' /* End of the bracket expression. */
- || p[0] != ']'
- || p + 1 == pend
- || (strcmp (str, "alpha") != 0
- && strcmp (str, "upper") != 0
- && strcmp (str, "lower") != 0
- && strcmp (str, "digit") != 0
- && strcmp (str, "alnum") != 0
- && strcmp (str, "xdigit") != 0
- && strcmp (str, "space") != 0
- && strcmp (str, "print") != 0
- && strcmp (str, "punct") != 0
- && strcmp (str, "graph") != 0
- && strcmp (str, "cntrl") != 0))
- {
- /* Undo the ending character, the letters, and leave
- the leading : and [ (but set bits for them). */
- c1++;
- while (c1--)
- PATUNFETCH;
- SET_LIST_BIT ('[');
- SET_LIST_BIT (':');
- }
- else
- {
- /* The ] at the end of the character class. */
- PATFETCH (c);
- if (c != ']')
- goto invalid_pattern;
- for (c = 0; c < (1 << BYTEWIDTH); c++)
- {
- if ((strcmp (str, "alpha") == 0 && isalpha (c))
- || (strcmp (str, "upper") == 0 && isupper (c))
- || (strcmp (str, "lower") == 0 && islower (c))
- || (strcmp (str, "digit") == 0 && isdigit (c))
- || (strcmp (str, "alnum") == 0 && isalnum (c))
- || (strcmp (str, "xdigit") == 0 && isxdigit (c))
- || (strcmp (str, "space") == 0 && isspace (c))
- || (strcmp (str, "print") == 0 && isprint (c))
- || (strcmp (str, "punct") == 0 && ispunct (c))
- || (strcmp (str, "graph") == 0 && isgraph (c))
- || (strcmp (str, "cntrl") == 0 && iscntrl (c)))
- SET_LIST_BIT (c);
- }
- }
- }
- else
- SET_LIST_BIT (c);
- }
+ laststart = b;
- /* Discard any character set/class bitmap bytes that are all
- 0 at the end of the map. Decrement the map-length byte too. */
- while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
- b[-1]--;
- b += b[-1];
+ /* We test `*p == '^' twice, instead of using an if
+ statement, so we only need one BUF_PUSH. */
+ BUF_PUSH (*p == '^' ? charset_not : charset);
+ if (*p == '^')
+ p++;
+
+ /* Remember the first position in the bracket expression. */
+ p1 = p;
+
+ /* Push the number of bytes in the bitmap. */
+ BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
+
+ /* Clear the whole map. */
+ bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
+
+ /* charset_not matches newline according to a syntax bit. */
+ if ((re_opcode_t) b[-2] == charset_not
+ && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
+ SET_LIST_BIT ('\n');
+
+ /* Read in characters and ranges, setting map bits. */
+ for (;;)
+ {
+ if (p == pend) return REG_EBRACK;
+
+ PATFETCH (c);
+
+ /* \ might escape characters inside [...] and [^...]. */
+ if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
+ {
+ if (p == pend) return REG_EESCAPE;
+
+ PATFETCH (c1);
+ SET_LIST_BIT (c1);
+ 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)
+ break;
+
+ /* Look ahead to see if it's a range when the last thing
+ was a character class. */
+ if (had_char_class && c == '-' && *p != ']')
+ return REG_ERANGE;
+
+ /* 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 (&p, pend, translate, syntax, b);
+ if (ret != REG_NOERROR) return ret;
+ }
+
+ 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 (&p, pend, translate, syntax, b);
+ if (ret != REG_NOERROR) return ret;
+ }
+
+ /* See if we're at the beginning of a possible character
+ class. */
+
+ else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
+ { /* Leave room for the null. */
+ char str[CHAR_CLASS_MAX_LENGTH + 1];
+
+ PATFETCH (c);
+ c1 = 0;
+
+ /* If pattern is `[[:'. */
+ if (p == pend) return REG_EBRACK;
+
+ 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;
+ boolean is_alnum = STREQ (str, "alnum");
+ boolean is_alpha = STREQ (str, "alpha");
+ boolean is_blank = STREQ (str, "blank");
+ boolean is_cntrl = STREQ (str, "cntrl");
+ boolean is_digit = STREQ (str, "digit");
+ boolean is_graph = STREQ (str, "graph");
+ boolean is_lower = STREQ (str, "lower");
+ boolean is_print = STREQ (str, "print");
+ boolean is_punct = STREQ (str, "punct");
+ boolean is_space = STREQ (str, "space");
+ boolean is_upper = STREQ (str, "upper");
+ boolean is_xdigit = STREQ (str, "xdigit");
+
+ if (!IS_CHAR_CLASS (str)) return REG_ECTYPE;
+
+ /* Throw away the ] at the end of the character
+ class. */
+ PATFETCH (c);
+
+ if (p == pend) return REG_EBRACK;
+
+ for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
+ {
+ if ( (is_alnum && ISALNUM (ch))
+ || (is_alpha && ISALPHA (ch))
+ || (is_blank && ISBLANK (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)))
+ SET_LIST_BIT (ch);
+ }
+ had_char_class = true;
+ }
+ else
+ {
+ c1++;
+ while (c1--)
+ PATUNFETCH;
+ SET_LIST_BIT ('[');
+ SET_LIST_BIT (':');
+ had_char_class = false;
+ }
+ }
+ else
+ {
+ had_char_class = false;
+ SET_LIST_BIT (c);
+ }
+ }
+
+ /* Discard any (non)matching list bytes that are all 0 at the
+ end of the map. Decrease the map-length byte too. */
+ while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
+ b[-1]--;
+ b += b[-1];
+ }
break;
+
case '(':
- if (! (obscure_syntax & RE_NO_BK_PARENS))
- goto normal_char;
- else
- goto handle_open;
+ 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 ')':
- if (! (obscure_syntax & RE_NO_BK_PARENS))
- goto normal_char;
- else
- goto handle_close;
case '\n':
- if (! (obscure_syntax & RE_NEWLINE_OR))
- goto normal_char;
- else
- goto handle_bar;
+ if (syntax & RE_NEWLINE_ALT)
+ goto handle_alt;
+ else
+ goto normal_char;
+
case '|':
- if ((obscure_syntax & RE_CONTEXTUAL_INVALID_OPS)
- && (! laststart || p == pend))
- goto invalid_pattern;
- else if (! (obscure_syntax & RE_NO_BK_VBAR))
- goto normal_char;
- else
- goto handle_bar;
+ if (syntax & RE_NO_BK_VBAR)
+ goto handle_alt;
+ else
+ goto normal_char;
- case '{':
- if (! ((obscure_syntax & RE_NO_BK_CURLY_BRACES)
- && (obscure_syntax & RE_INTERVALS)))
- goto normal_char;
- else
+
+ case '{':
+ if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
goto handle_interval;
-
+ else
+ goto normal_char;
+
+
case '\\':
- if (p == pend) goto invalid_pattern;
- PATFETCH_RAW (c);
- switch (c)
- {
- case '(':
- if (obscure_syntax & RE_NO_BK_PARENS)
- goto normal_backsl;
- handle_open:
- if (stackp == stacke) goto nesting_too_deep;
-
- /* Laststart should point to the start_memory that we are about
- to push (unless the pattern has RE_NREGS or more ('s). */
- *stackp++ = b - bufp->buffer;
- if (regnum < RE_NREGS)
- {
- BUFPUSH (start_memory);
- BUFPUSH (regnum);
- }
- *stackp++ = fixup_jump ? fixup_jump - bufp->buffer + 1 : 0;
- *stackp++ = regnum++;
- *stackp++ = begalt - bufp->buffer;
- fixup_jump = 0;
- laststart = 0;
- begalt = b;
- break;
-
- case ')':
- if (obscure_syntax & RE_NO_BK_PARENS)
- goto normal_backsl;
- handle_close:
- if (stackp == stackb) goto unmatched_close;
- begalt = *--stackp + bufp->buffer;
- if (fixup_jump)
- store_jump (fixup_jump, jump, b);
- if (stackp[-1] < RE_NREGS)
- {
- BUFPUSH (stop_memory);
- BUFPUSH (stackp[-1]);
- }
- stackp -= 2;
- fixup_jump = *stackp ? *stackp + bufp->buffer - 1 : 0;
- laststart = *--stackp + bufp->buffer;
- break;
-
- case '|':
- if ((obscure_syntax & RE_LIMITED_OPS)
- || (obscure_syntax & RE_NO_BK_VBAR))
- goto normal_backsl;
- handle_bar:
- if (obscure_syntax & RE_LIMITED_OPS)
- goto normal_char;
- /* Insert before the previous alternative a jump which
- jumps to this alternative if the former fails. */
- GET_BUFFER_SPACE (6);
- insert_jump (on_failure_jump, begalt, b + 6, b);
- pending_exact = 0;
- b += 3;
- /* The alternative before the previous alternative has a
- jump after it which gets executed if it gets matched.
- Adjust that jump so it will jump to the previous
- alternative's analogous jump (put in below, which in
- turn will jump to the next (if any) alternative's such
- jump, etc.). The last such jump jumps to the correct
- final destination. */
- if (fixup_jump)
- store_jump (fixup_jump, jump, b);
+ if (p == pend) return REG_EESCAPE;
+
+ /* 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:
+ bufp->re_nsub++;
+ regnum++;
+
+ if (COMPILE_STACK_FULL)
+ {
+ RETALLOC (compile_stack.stack, compile_stack.size << 1,
+ compile_stack_elt_t);
+ if (compile_stack.stack == NULL) return REG_ESPACE;
+
+ compile_stack.size <<= 1;
+ }
+
+ /* These are the values to restore when we hit end of this
+ group. They are all relative offsets, so that if the
+ whole pattern moves because of realloc, they will still
+ be valid. */
+ COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer;
+ COMPILE_STACK_TOP.fixup_alt_jump
+ = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0;
+ COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer;
+ COMPILE_STACK_TOP.regnum = regnum;
+
+ /* We will eventually replace the 0 with the number of
+ groups inner to this one. But do not push a
+ start_memory for groups beyond the last one we can
+ represent in the compiled pattern. */
+ if (regnum <= MAX_REGNUM)
+ {
+ COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2;
+ BUF_PUSH_3 (start_memory, regnum, 0);
+ }
- /* Leave space for a jump after previous alternative---to be
- filled in later. */
- fixup_jump = b;
- b += 3;
+ compile_stack.avail++;
+ fixup_alt_jump = 0;
laststart = 0;
- begalt = b;
- break;
+ begalt = b;
+ /* If we've reached MAX_REGNUM groups, then this open
+ won't actually generate any code, so we'll have to
+ clear pending_exact explicitly. */
+ pending_exact = 0;
+ break;
- case '{':
- if (! (obscure_syntax & RE_INTERVALS)
- /* Let \{ be a literal. */
- || ((obscure_syntax & RE_INTERVALS)
- && (obscure_syntax & RE_NO_BK_CURLY_BRACES))
- /* If it's the string "\{". */
- || (p - 2 == pattern && p == pend))
- goto normal_backsl;
- handle_interval:
- beg_interval = p - 1; /* The {. */
- /* If there is no previous pattern, this isn't an interval. */
- if (!laststart)
- {
- if (obscure_syntax & RE_CONTEXTUAL_INVALID_OPS)
- goto invalid_pattern;
- else
- goto normal_backsl;
+
+ case ')':
+ if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
+
+ if (COMPILE_STACK_EMPTY)
+ if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
+ goto normal_backslash;
+ else
+ return REG_ERPAREN;
+
+ handle_close:
+ if (fixup_alt_jump)
+ { /* Push a dummy failure point at the end of the
+ alternative for a possible future
+ `pop_failure_jump' to pop. See comments at
+ `push_dummy_failure' in `re_match_2'. */
+ BUF_PUSH (push_dummy_failure);
+
+ /* We allocated space for this jump when we assigned
+ to `fixup_alt_jump', in the `handle_alt' case below. */
+ STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
}
- /* It also isn't an interval if not preceded by an re
- matching a single character or subexpression, or if
- the current type of intervals can't handle back
- references and the previous thing is a back reference. */
- if (! (*laststart == anychar
- || *laststart == charset
- || *laststart == charset_not
- || *laststart == start_memory
- || (*laststart == exactn && laststart[1] == 1)
- || (! (obscure_syntax & RE_NO_BK_REFS)
- && *laststart == duplicate)))
- {
- if (obscure_syntax & RE_NO_BK_CURLY_BRACES)
- goto normal_char;
-
- /* Posix extended syntax is handled in previous
- statement; this is for Posix basic syntax. */
- if (obscure_syntax & RE_INTERVALS)
- goto invalid_pattern;
+
+ /* See similar code for backslashed left paren above. */
+ if (COMPILE_STACK_EMPTY)
+ if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
+ goto normal_char;
+ else
+ return REG_ERPAREN;
+
+ /* Since we just checked for an empty stack above, this
+ ``can't happen''. */
+ assert (compile_stack.avail != 0);
+ {
+ /* 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;
+
+ compile_stack.avail--;
+ begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset;
+ fixup_alt_jump
+ = COMPILE_STACK_TOP.fixup_alt_jump
+ ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1
+ : 0;
+ laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset;
+ this_group_regnum = COMPILE_STACK_TOP.regnum;
+ /* If we've reached MAX_REGNUM groups, then this open
+ won't actually generate any code, so we'll have to
+ clear pending_exact explicitly. */
+ pending_exact = 0;
+
+ /* We're at the end of the group, so now we know how many
+ groups were inside this one. */
+ if (this_group_regnum <= MAX_REGNUM)
+ {
+ unsigned char *inner_group_loc
+ = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset;
- goto normal_backsl;
- }
- lower_bound = -1; /* So can see if are set. */
- upper_bound = -1;
- GET_UNSIGNED_NUMBER (lower_bound);
- if (c == ',')
- {
- GET_UNSIGNED_NUMBER (upper_bound);
- if (upper_bound < 0)
- upper_bound = RE_DUP_MAX;
- }
- if (upper_bound < 0)
- upper_bound = lower_bound;
- if (! (obscure_syntax & RE_NO_BK_CURLY_BRACES))
- {
- if (c != '\\')
- goto invalid_pattern;
- PATFETCH (c);
- }
- if (c != '}' || lower_bound < 0 || upper_bound > RE_DUP_MAX
- || lower_bound > upper_bound
- || ((obscure_syntax & RE_NO_BK_CURLY_BRACES)
- && p != pend && *p == '{'))
- {
- if (obscure_syntax & RE_NO_BK_CURLY_BRACES)
- goto unfetch_interval;
- else
- goto invalid_pattern;
- }
+ *inner_group_loc = regnum - this_group_regnum;
+ BUF_PUSH_3 (stop_memory, this_group_regnum,
+ regnum - this_group_regnum);
+ }
+ }
+ break;
- /* If upper_bound is zero, don't want to succeed at all;
- jump from laststart to b + 3, which will be the end of
- the buffer after this jump is inserted. */
-
- if (upper_bound == 0)
- {
- GET_BUFFER_SPACE (3);
- insert_jump (jump, laststart, b + 3, b);
- b += 3;
- }
- /* Otherwise, after lower_bound number of succeeds, jump
- to after the jump_n which will be inserted at the end
- of the buffer, and insert that jump_n. */
- else
- { /* Set to 5 if only one repetition is allowed and
- hence no jump_n is inserted at the current end of
- the buffer; then only space for the succeed_n is
- needed. Otherwise, need space for both the
- succeed_n and the jump_n. */
-
- unsigned slots_needed = upper_bound == 1 ? 5 : 10;
-
- GET_BUFFER_SPACE (slots_needed);
- /* Initialize the succeed_n to n, even though it will
- be set by its attendant set_number_at, because
- re_compile_fastmap will need to know it. Jump to
- what the end of buffer will be after inserting
- this succeed_n and possibly appending a jump_n. */
- insert_jump_n (succeed_n, laststart, b + slots_needed,
- b, lower_bound);
- b += 5; /* Just increment for the succeed_n here. */
-
- /* More than one repetition is allowed, so put in at
- the end of the buffer a backward jump from b to the
- succeed_n we put in above. By the time we've gotten
- to this jump when matching, we'll have matched once
- already, so jump back only upper_bound - 1 times. */
-
- if (upper_bound > 1)
- {
- store_jump_n (b, jump_n, laststart, upper_bound - 1);
- b += 5;
- /* When hit this when matching, reset the
- preceding jump_n's n to upper_bound - 1. */
- BUFPUSH (set_number_at);
- GET_BUFFER_SPACE (2);
- STORE_NUMBER_AND_INCR (b, -5);
- STORE_NUMBER_AND_INCR (b, upper_bound - 1);
- }
- /* When hit this when matching, set the succeed_n's n. */
- GET_BUFFER_SPACE (5);
- insert_op_2 (set_number_at, laststart, b, 5, lower_bound);
- b += 5;
- }
+ case '|': /* `\|'. */
+ if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
+ goto normal_backslash;
+ handle_alt:
+ if (syntax & RE_LIMITED_OPS)
+ goto normal_char;
+
+ /* Insert before the previous alternative a jump which
+ jumps to this alternative if the former fails. */
+ GET_BUFFER_SPACE (3);
+ INSERT_JUMP (on_failure_jump, begalt, b + 6);
pending_exact = 0;
- beg_interval = 0;
+ b += 3;
+
+ /* The alternative before this one has a jump after it
+ which gets executed if it gets matched. Adjust that
+ jump so it will jump to this alternative's analogous
+ jump (put in below, which in turn will jump to the next
+ (if any) alternative's such jump, etc.). The last such
+ jump jumps to the correct final destination. A picture:
+ _____ _____
+ | | | |
+ | v | v
+ a | b | c
+
+ If we are at `b', then fixup_alt_jump right now points to a
+ three-byte space after `a'. We'll put in the jump, set
+ fixup_alt_jump to right after `b', and leave behind three
+ bytes which we'll fill in when we get to after `c'. */
+
+ if (fixup_alt_jump)
+ STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
+
+ /* Mark and leave space for a jump after this alternative,
+ to be filled in later either by next alternative or
+ when know we're at the end of a series of alternatives. */
+ fixup_alt_jump = b;
+ GET_BUFFER_SPACE (3);
+ b += 3;
+
+ laststart = 0;
+ begalt = b;
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 = -1, upper_bound = -1;
+
+ beg_interval = p - 1;
+
+ if (p == pend)
+ {
+ if (syntax & RE_NO_BK_BRACES)
+ goto unfetch_interval;
+ else
+ return REG_EBRACE;
+ }
+
+ 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 `{1}' => match exactly once. */
+ 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
+ return REG_BADBR;
+ }
+
+ if (!(syntax & RE_NO_BK_BRACES))
+ {
+ if (c != '\\') return REG_EBRACE;
+
+ PATFETCH (c);
+ }
+
+ if (c != '}')
+ {
+ if (syntax & RE_NO_BK_BRACES)
+ goto unfetch_interval;
+ else
+ return REG_BADBR;
+ }
+
+ /* We just parsed a valid interval. */
+
+ /* If it's invalid to have no preceding re. */
+ if (!laststart)
+ {
+ if (syntax & RE_CONTEXT_INVALID_OPS)
+ return REG_BADRPT;
+ else if (syntax & RE_CONTEXT_INDEP_OPS)
+ laststart = b;
+ else
+ goto unfetch_interval;
+ }
+
+ /* If the upper bound is zero, don't want to succeed at
+ all; jump from `laststart' to `b + 3', which will be
+ the end of the buffer after we insert the jump. */
+ if (upper_bound == 0)
+ {
+ GET_BUFFER_SPACE (3);
+ INSERT_JUMP (jump, laststart, b + 3);
+ b += 3;
+ }
+
+ /* Otherwise, we have a nontrivial interval. When
+ we're all done, the pattern will look like:
+ set_number_at <jump count> <upper bound>
+ set_number_at <succeed_n count> <lower bound>
+ succeed_n <after jump addr> <succed_n count>
+ <body of loop>
+ jump_n <succeed_n addr> <jump count>
+ (The upper bound and `jump_n' are omitted if
+ `upper_bound' is 1, though.) */
+ else
+ { /* If the upper bound is > 1, we need to insert
+ more at the end of the loop. */
+ unsigned nbytes = 10 + (upper_bound > 1) * 10;
+
+ GET_BUFFER_SPACE (nbytes);
+
+ /* Initialize lower bound of the `succeed_n', even
+ though it will be set during matching by its
+ attendant `set_number_at' (inserted next),
+ because `re_compile_fastmap' needs to know.
+ Jump to the `jump_n' we might insert below. */
+ INSERT_JUMP2 (succeed_n, laststart,
+ b + 5 + (upper_bound > 1) * 5,
+ lower_bound);
+ b += 5;
+
+ /* Code to initialize the lower bound. Insert
+ before the `succeed_n'. The `5' is the last two
+ bytes of this `set_number_at', plus 3 bytes of
+ the following `succeed_n'. */
+ insert_op2 (set_number_at, laststart, 5, lower_bound, b);
+ b += 5;
+
+ if (upper_bound > 1)
+ { /* More than one repetition is allowed, so
+ append a backward jump to the `succeed_n'
+ that starts this interval.
+
+ When we've reached this during matching,
+ we'll have matched the interval once, so
+ jump back only `upper_bound - 1' times. */
+ STORE_JUMP2 (jump_n, b, laststart + 5,
+ upper_bound - 1);
+ b += 5;
+
+ /* The location we want to set is the second
+ parameter of the `jump_n'; that is `b-2' as
+ an absolute address. `laststart' will be
+ the `set_number_at' we're about to insert;
+ `laststart+3' the number to set, the source
+ for the relative address. But we are
+ inserting into the middle of the pattern --
+ so everything is getting moved up by 5.
+ Conclusion: (b - 2) - (laststart + 3) + 5,
+ i.e., b - laststart.
+
+ We insert this at the beginning of the loop
+ so that if we fail during matching, we'll
+ reinitialize the bounds. */
+ insert_op2 (set_number_at, laststart, b - laststart,
+ upper_bound - 1, b);
+ b += 5;
+ }
+ }
+ pending_exact = 0;
+ beg_interval = NULL;
+ }
+ break;
+
unfetch_interval:
- /* If an invalid interval, match the characters as literals. */
- if (beg_interval)
- p = beg_interval;
- else
+ /* If an invalid interval, match the characters as literals. */
+ assert (beg_interval);
+ p = beg_interval;
+ beg_interval = NULL;
+
+ /* normal_char and normal_backslash need `c'. */
+ PATFETCH (c);
+
+ if (!(syntax & RE_NO_BK_BRACES))
{
- fprintf (stderr,
- "regex: no interval beginning to which to backtrack.\n");
- exit (1);
+ if (p > pattern && p[-1] == '\\')
+ goto normal_backslash;
}
-
- beg_interval = 0;
- PATFETCH (c); /* normal_char expects char in `c'. */
- goto normal_char;
- break;
+ goto normal_char;
#ifdef emacs
- case '=':
- BUFPUSH (at_dot);
- break;
-
- case 's':
- laststart = b;
- BUFPUSH (syntaxspec);
- PATFETCH (c);
- BUFPUSH (syntax_spec_code[c]);
- break;
-
- case 'S':
- laststart = b;
- BUFPUSH (notsyntaxspec);
- PATFETCH (c);
- BUFPUSH (syntax_spec_code[c]);
- break;
+ /* There is no way to specify the before_dot and after_dot
+ operators. rms says this is ok. --karl */
+ case '=':
+ BUF_PUSH (at_dot);
+ break;
+
+ case 's':
+ laststart = b;
+ PATFETCH (c);
+ BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
+ break;
+
+ case 'S':
+ laststart = b;
+ PATFETCH (c);
+ BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
+ break;
#endif /* emacs */
- case 'w':
- laststart = b;
- BUFPUSH (wordchar);
- break;
-
- case 'W':
- laststart = b;
- BUFPUSH (notwordchar);
- break;
-
- case '<':
- BUFPUSH (wordbeg);
- break;
-
- case '>':
- BUFPUSH (wordend);
- break;
-
- case 'b':
- BUFPUSH (wordbound);
- break;
-
- case 'B':
- BUFPUSH (notwordbound);
- break;
-
- case '`':
- BUFPUSH (begbuf);
- break;
-
- case '\'':
- BUFPUSH (endbuf);
- break;
-
- case '1':
- case '2':
- case '3':
- case '4':
- case '5':
- case '6':
- case '7':
- case '8':
- case '9':
- if (obscure_syntax & RE_NO_BK_REFS)
+
+ case 'w':
+ laststart = b;
+ BUF_PUSH (wordchar);
+ break;
+
+
+ case 'W':
+ laststart = b;
+ BUF_PUSH (notwordchar);
+ break;
+
+
+ case '<':
+ BUF_PUSH (wordbeg);
+ break;
+
+ case '>':
+ BUF_PUSH (wordend);
+ break;
+
+ case 'b':
+ BUF_PUSH (wordbound);
+ break;
+
+ case 'B':
+ BUF_PUSH (notwordbound);
+ break;
+
+ case '`':
+ BUF_PUSH (begbuf);
+ break;
+
+ case '\'':
+ BUF_PUSH (endbuf);
+ 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)
- {
- if (obscure_syntax & RE_NO_EMPTY_BK_REF)
- goto invalid_pattern;
- else
- goto normal_char;
- }
+
+ if (c1 > regnum)
+ return REG_ESUBREG;
+
/* Can't back reference to a subexpression if inside of it. */
- for (stackt = stackp - 2; stackt > stackb; stackt -= 4)
- if (*stackt == c1)
- goto normal_char;
- laststart = b;
- BUFPUSH (duplicate);
- BUFPUSH (c1);
- break;
-
- case '+':
- case '?':
- if (obscure_syntax & RE_BK_PLUS_QM)
- goto handle_plus;
- else
- goto normal_backsl;
+ if (group_in_compile_stack (compile_stack, c1))
+ goto normal_char;
+
+ laststart = b;
+ BUF_PUSH_2 (duplicate, c1);
break;
+
+ case '+':
+ case '?':
+ if (syntax & RE_BK_PLUS_QM)
+ goto handle_plus;
+ else
+ goto normal_backslash;
+
default:
- normal_backsl:
- /* 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. */
- if (translate) c = translate[c];
- goto normal_char;
- }
- break;
+ 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:
- normal_char: /* Expects the character in `c'. */
- if (!pending_exact || pending_exact + *pending_exact + 1 != b
- || *pending_exact == 0177 || *p == '*' || *p == '^'
- || ((obscure_syntax & RE_BK_PLUS_QM)
+ /* Expects the character in `c'. */
+ normal_char:
+ /* If no exactn currently being built. */
+ if (!pending_exact
+
+ /* If last exactn not at current position. */
+ || pending_exact + *pending_exact + 1 != b
+
+ /* We have only one byte following the exactn for the count. */
+ || *pending_exact == (1 << BYTEWIDTH) - 1
+
+ /* If followed by a repetition operator. */
+ || *p == '*' || *p == '^'
+ || ((syntax & RE_BK_PLUS_QM)
? *p == '\\' && (p[1] == '+' || p[1] == '?')
: (*p == '+' || *p == '?'))
- || ((obscure_syntax & RE_INTERVALS)
- && ((obscure_syntax & RE_NO_BK_CURLY_BRACES)
+ || ((syntax & RE_INTERVALS)
+ && ((syntax & RE_NO_BK_BRACES)
? *p == '{'
: (p[0] == '\\' && p[1] == '{'))))
{
- laststart = b;
- BUFPUSH (exactn);
- pending_exact = b;
- BUFPUSH (0);
- }
- BUFPUSH (c);
- (*pending_exact)++;
- }
- }
+ /* Start building a new exactn. */
+
+ laststart = b;
- if (fixup_jump)
- store_jump (fixup_jump, jump, b);
+ BUF_PUSH_2 (exactn, 0);
+ pending_exact = b - 1;
+ }
+
+ BUF_PUSH (c);
+ (*pending_exact)++;
+ break;
+ } /* switch (c) */
+ } /* while p != pend */
- if (stackp != stackb) goto unmatched_open;
+
+ /* Through the pattern now. */
+
+ if (fixup_alt_jump)
+ STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
+ if (!COMPILE_STACK_EMPTY)
+ return REG_EPAREN;
+
+ free (compile_stack.stack);
+
+ /* We have succeeded; set the length of the buffer. */
bufp->used = b - bufp->buffer;
- return 0;
- invalid_pattern:
- return "Invalid regular expression";
+#ifdef DEBUG
+ if (debug)
+ {
+ DEBUG_PRINT1 ("\nCompiled pattern: \n");
+ print_compiled_pattern (bufp);
+ }
+#endif /* DEBUG */
- unmatched_open:
- return "Unmatched \\(";
+ return REG_NOERROR;
+} /* regex_compile */
+
+/* Subroutines for `regex_compile'. */
- unmatched_close:
- return "Unmatched \\)";
+/* Store OP at LOC followed by two-byte integer parameter ARG. */
- end_of_pattern:
- return "Premature end of regular expression";
+static void
+store_op1 (op, loc, arg)
+ re_opcode_t op;
+ unsigned char *loc;
+ int arg;
+{
+ *loc = (unsigned char) op;
+ STORE_NUMBER (loc + 1, arg);
+}
- nesting_too_deep:
- return "Nesting too deep";
- too_big:
- return "Regular expression too big";
+/* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
- memory_exhausted:
- return "Memory exhausted";
+static void
+store_op2 (op, loc, arg1, arg2)
+ re_opcode_t op;
+ unsigned char *loc;
+ int arg1, arg2;
+{
+ *loc = (unsigned char) op;
+ STORE_NUMBER (loc + 1, arg1);
+ STORE_NUMBER (loc + 3, arg2);
}
-/* Store a jump of the form <OPCODE> <relative address>.
- Store in the location FROM a jump operation to jump to relative
- address FROM - TO. OPCODE is the opcode to store. */
+/* Copy the bytes from LOC to END to open up three bytes of space at LOC
+ for OP followed by two-byte integer parameter ARG. */
static void
-store_jump (from, opcode, to)
- char *from, *to;
- char opcode;
+insert_op1 (op, loc, arg, end)
+ re_opcode_t op;
+ unsigned char *loc;
+ int arg;
+ unsigned char *end;
{
- from[0] = opcode;
- STORE_NUMBER(from + 1, to - (from + 3));
-}
+ register unsigned char *pfrom = end;
+ register unsigned char *pto = end + 3;
+ while (pfrom != loc)
+ *--pto = *--pfrom;
+
+ store_op1 (op, loc, arg);
+}
-/* Open up space before char FROM, and insert there a jump to TO.
- CURRENT_END gives the end of the storage not in use, so we know
- how much data to copy up. OP is the opcode of the jump to insert.
- If you call this function, you must zero out pending_exact. */
+/* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
static void
-insert_jump (op, from, to, current_end)
- char op;
- char *from, *to, *current_end;
+insert_op2 (op, loc, arg1, arg2, end)
+ re_opcode_t op;
+ unsigned char *loc;
+ int arg1, arg2;
+ unsigned char *end;
{
- register char *pfrom = current_end; /* Copy from here... */
- register char *pto = current_end + 3; /* ...to here. */
+ register unsigned char *pfrom = end;
+ register unsigned char *pto = end + 5;
- while (pfrom != from)
+ while (pfrom != loc)
*--pto = *--pfrom;
- store_jump (from, op, to);
+
+ store_op2 (op, loc, arg1, arg2);
}
-/* Store a jump of the form <opcode> <relative address> <n> .
+/* 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 ^. */
- Store in the location FROM a jump operation to jump to relative
- address FROM - TO. OPCODE is the opcode to store, N is a number the
- jump uses, say, to decide how many times to jump.
-
- If you call this function, you must zero out pending_exact. */
+static boolean
+at_begline_loc_p (pattern, p, syntax)
+ const char *pattern, *p;
+ reg_syntax_t syntax;
+{
+ const char *prev = p - 2;
+ boolean 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));
+}
-static void
-store_jump_n (from, opcode, to, n)
- char *from, *to;
- char opcode;
- unsigned n;
+
+/* 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'. */
+
+static boolean
+at_endline_loc_p (p, pend, syntax)
+ const char *p, *pend;
+ int syntax;
{
- from[0] = opcode;
- STORE_NUMBER (from + 1, to - (from + 3));
- STORE_NUMBER (from + 3, n);
+ const char *next = p;
+ boolean next_backslash = *next == '\\';
+ const char *next_next = p + 1 < pend ? p + 1 : NULL;
+
+ 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 == '|');
}
-/* Similar to insert_jump, but handles a jump which needs an extra
- number to handle minimum and maximum cases. Open up space at
- location FROM, and insert there a jump to TO. CURRENT_END gives the
- end of the storage in use, so we know how much data to copy up. OP is
- the opcode of the jump to insert.
+/* Returns true if REGNUM is in one of COMPILE_STACK's elements and
+ false if it's not. */
- If you call this function, you must zero out pending_exact. */
+static boolean
+group_in_compile_stack (compile_stack, regnum)
+ compile_stack_type compile_stack;
+ regnum_t regnum;
+{
+ int this_element;
-static void
-insert_jump_n (op, from, to, current_end, n)
- char op;
- char *from, *to, *current_end;
- unsigned n;
+ for (this_element = compile_stack.avail - 1;
+ this_element >= 0;
+ this_element--)
+ if (compile_stack.stack[this_element].regnum == regnum)
+ return true;
+
+ return false;
+}
+
+
+/* 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. */
+
+static reg_errcode_t
+compile_range (p_ptr, pend, translate, syntax, b)
+ const char **p_ptr, *pend;
+ char *translate;
+ reg_syntax_t syntax;
+ unsigned char *b;
{
- register char *pfrom = current_end; /* Copy from here... */
- register char *pto = current_end + 5; /* ...to here. */
+ unsigned this_char;
- while (pfrom != from)
- *--pto = *--pfrom;
- store_jump_n (from, op, to, n);
+ const char *p = *p_ptr;
+ int range_start, range_end;
+
+ if (p == pend)
+ return REG_ERANGE;
+
+ /* Even though the pattern is a signed `char *', we need to fetch
+ with unsigned char *'s; if the high bit of the pattern character
+ is set, the range endpoints will be negative if we fetch using a
+ signed char *.
+
+ We also want to fetch the endpoints without translating them; the
+ appropriate translation is done in the bit-setting loop below. */
+ range_start = ((unsigned char *) p)[-2];
+ range_end = ((unsigned char *) p)[0];
+
+ /* Have to increment the pointer into the pattern string, so the
+ caller isn't still at the ending character. */
+ (*p_ptr)++;
+
+ /* If the start is after the end, the range is empty. */
+ if (range_start > range_end)
+ return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
+
+ /* Here we see why `this_char' has to be larger than an `unsigned
+ char' -- the range is inclusive, so if `range_end' == 0xff
+ (assuming 8-bit characters), we would otherwise go into an infinite
+ loop, since all characters <= 0xff. */
+ for (this_char = range_start; this_char <= range_end; this_char++)
+ {
+ SET_LIST_BIT (TRANSLATE (this_char));
+ }
+
+ return REG_NOERROR;
}
+
+/* Failure stack declarations and macros; both re_compile_fastmap and
+ re_match_2 use a failure stack. These have to be macros because of
+ REGEX_ALLOCATE. */
+
+/* Number of failure points for which to initially allocate space
+ when matching. If this number is exceeded, we allocate more
+ space, so it is not a hard limit. */
+#ifndef INIT_FAILURE_ALLOC
+#define INIT_FAILURE_ALLOC 5
+#endif
-/* Open up space at location THERE, and insert operation OP followed by
- NUM_1 and NUM_2. CURRENT_END gives the end of the storage in use, so
- we know how much data to copy up.
+/* Roughly the maximum number of failure points on the stack. Would be
+ exactly that if always used MAX_FAILURE_SPACE each time we failed.
+ This is a variable only so users of regex can assign to it; we never
+ change it ourselves. */
+int re_max_failures = 2000;
- If you call this function, you must zero out pending_exact. */
+typedef const unsigned char *fail_stack_elt_t;
-static void
-insert_op_2 (op, there, current_end, num_1, num_2)
- char op;
- char *there, *current_end;
- int num_1, num_2;
+typedef struct
{
- register char *pfrom = current_end; /* Copy from here... */
- register char *pto = current_end + 5; /* ...to here. */
+ fail_stack_elt_t *stack;
+ unsigned size;
+ unsigned avail; /* Offset of next open position. */
+} fail_stack_type;
- while (pfrom != there)
- *--pto = *--pfrom;
-
- there[0] = op;
- STORE_NUMBER (there + 1, num_1);
- STORE_NUMBER (there + 3, num_2);
-}
+#define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
+#define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
+#define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
+#define FAIL_STACK_TOP() (fail_stack.stack[fail_stack.avail])
+
+
+/* Initialize `fail_stack'. Do `return -2' if the alloc fails. */
+
+#define INIT_FAIL_STACK() \
+ do { \
+ fail_stack.stack = (fail_stack_elt_t *) \
+ REGEX_ALLOCATE (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \
+ \
+ if (fail_stack.stack == NULL) \
+ return -2; \
+ \
+ fail_stack.size = INIT_FAILURE_ALLOC; \
+ fail_stack.avail = 0; \
+ } while (0)
+
+
+/* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
+
+ Return 1 if succeeds, and 0 if either ran out of memory
+ allocating space for it or it was already too large.
+
+ REGEX_REALLOCATE requires `destination' be declared. */
+
+#define DOUBLE_FAIL_STACK(fail_stack) \
+ ((fail_stack).size > re_max_failures * MAX_FAILURE_ITEMS \
+ ? 0 \
+ : ((fail_stack).stack = (fail_stack_elt_t *) \
+ REGEX_REALLOCATE ((fail_stack).stack, \
+ (fail_stack).size * sizeof (fail_stack_elt_t), \
+ ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \
+ \
+ (fail_stack).stack == NULL \
+ ? 0 \
+ : ((fail_stack).size <<= 1, \
+ 1)))
+
+
+/* Push PATTERN_OP on FAIL_STACK.
+
+ Return 1 if was able to do so and 0 if ran out of memory allocating
+ space to do so. */
+#define PUSH_PATTERN_OP(pattern_op, fail_stack) \
+ ((FAIL_STACK_FULL () \
+ && !DOUBLE_FAIL_STACK (fail_stack)) \
+ ? 0 \
+ : ((fail_stack).stack[(fail_stack).avail++] = pattern_op, \
+ 1))
+
+/* This pushes an item onto the failure stack. Must be a four-byte
+ value. Assumes the variable `fail_stack'. Probably should only
+ be called from within `PUSH_FAILURE_POINT'. */
+#define PUSH_FAILURE_ITEM(item) \
+ fail_stack.stack[fail_stack.avail++] = (fail_stack_elt_t) item
+
+/* The complement operation. Assumes `fail_stack' is nonempty. */
+#define POP_FAILURE_ITEM() fail_stack.stack[--fail_stack.avail]
+
+/* Used to omit pushing failure point id's when we're not debugging. */
+#ifdef DEBUG
+#define DEBUG_PUSH PUSH_FAILURE_ITEM
+#define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_ITEM ()
+#else
+#define DEBUG_PUSH(item)
+#define DEBUG_POP(item_addr)
+#endif
+
+
+/* Push the information about the state we will need
+ if we ever fail back to it.
+
+ Requires variables fail_stack, regstart, regend, reg_info, and
+ num_regs be declared. DOUBLE_FAIL_STACK requires `destination' be
+ declared.
+
+ Does `return FAILURE_CODE' if runs out of memory. */
+
+#define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
+ do { \
+ char *destination; \
+ /* Must be int, so when we don't save any registers, the arithmetic \
+ of 0 + -1 isn't done as unsigned. */ \
+ int this_reg; \
+ \
+ DEBUG_STATEMENT (failure_id++); \
+ DEBUG_STATEMENT (nfailure_points_pushed++); \
+ DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
+ DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
+ DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
+ \
+ DEBUG_PRINT2 (" slots needed: %d\n", NUM_FAILURE_ITEMS); \
+ DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
+ \
+ /* Ensure we have enough space allocated for what we will push. */ \
+ while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
+ { \
+ if (!DOUBLE_FAIL_STACK (fail_stack)) \
+ return failure_code; \
+ \
+ DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
+ (fail_stack).size); \
+ DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
+ } \
+ \
+ /* Push the info, starting with the registers. */ \
+ DEBUG_PRINT1 ("\n"); \
+ \
+ for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
+ this_reg++) \
+ { \
+ DEBUG_PRINT2 (" Pushing reg: %d\n", this_reg); \
+ DEBUG_STATEMENT (num_regs_pushed++); \
+ \
+ DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
+ PUSH_FAILURE_ITEM (regstart[this_reg]); \
+ \
+ DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
+ PUSH_FAILURE_ITEM (regend[this_reg]); \
+ \
+ DEBUG_PRINT2 (" info: 0x%x\n ", reg_info[this_reg]); \
+ DEBUG_PRINT2 (" match_null=%d", \
+ REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
+ DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
+ DEBUG_PRINT2 (" matched_something=%d", \
+ MATCHED_SOMETHING (reg_info[this_reg])); \
+ DEBUG_PRINT2 (" ever_matched=%d", \
+ EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
+ DEBUG_PRINT1 ("\n"); \
+ PUSH_FAILURE_ITEM (reg_info[this_reg].word); \
+ } \
+ \
+ DEBUG_PRINT2 (" Pushing low active reg: %d\n", lowest_active_reg);\
+ PUSH_FAILURE_ITEM (lowest_active_reg); \
+ \
+ DEBUG_PRINT2 (" Pushing high active reg: %d\n", highest_active_reg);\
+ PUSH_FAILURE_ITEM (highest_active_reg); \
+ \
+ DEBUG_PRINT2 (" Pushing pattern 0x%x: ", pattern_place); \
+ DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
+ PUSH_FAILURE_ITEM (pattern_place); \
+ \
+ DEBUG_PRINT2 (" Pushing string 0x%x: `", string_place); \
+ DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
+ size2); \
+ DEBUG_PRINT1 ("'\n"); \
+ PUSH_FAILURE_ITEM (string_place); \
+ \
+ DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
+ DEBUG_PUSH (failure_id); \
+ } while (0)
+
+/* This is the number of items that are pushed and popped on the stack
+ for each register. */
+#define NUM_REG_ITEMS 3
+
+/* Individual items aside from the registers. */
+#ifdef DEBUG
+#define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
+#else
+#define NUM_NONREG_ITEMS 4
+#endif
+
+/* We push at most this many items on the stack. */
+#define MAX_FAILURE_ITEMS ((num_regs - 1) * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
+
+/* We actually push this many items. */
+#define NUM_FAILURE_ITEMS \
+ ((highest_active_reg - lowest_active_reg + 1) * NUM_REG_ITEMS \
+ + NUM_NONREG_ITEMS)
+
+/* How many items can still be added to the stack without overflowing it. */
+#define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
+/* Pops what PUSH_FAIL_STACK pushes.
+
+ We restore into the parameters, all of which should be lvalues:
+ STR -- the saved data position.
+ PAT -- the saved pattern position.
+ LOW_REG, HIGH_REG -- the highest and lowest active registers.
+ REGSTART, REGEND -- arrays of string positions.
+ REG_INFO -- array of information about each subexpression.
+
+ Also assumes the variables `fail_stack' and (if debugging), `bufp',
+ `pend', `string1', `size1', `string2', and `size2'. */
+
+#define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
+{ \
+ DEBUG_STATEMENT (fail_stack_elt_t failure_id;) \
+ int this_reg; \
+ const unsigned char *string_temp; \
+ \
+ assert (!FAIL_STACK_EMPTY ()); \
+ \
+ /* Remove failure points and point to how many regs pushed. */ \
+ DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
+ DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
+ DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
+ \
+ assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
+ \
+ DEBUG_POP (&failure_id); \
+ DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
+ \
+ /* If the saved string location is NULL, it came from an \
+ on_failure_keep_string_jump opcode, and we want to throw away the \
+ saved NULL, thus retaining our current position in the string. */ \
+ string_temp = POP_FAILURE_ITEM (); \
+ if (string_temp != NULL) \
+ str = (const char *) string_temp; \
+ \
+ DEBUG_PRINT2 (" Popping string 0x%x: `", str); \
+ DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
+ DEBUG_PRINT1 ("'\n"); \
+ \
+ pat = (unsigned char *) POP_FAILURE_ITEM (); \
+ DEBUG_PRINT2 (" Popping pattern 0x%x: ", pat); \
+ DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
+ \
+ /* Restore register info. */ \
+ high_reg = (unsigned) POP_FAILURE_ITEM (); \
+ DEBUG_PRINT2 (" Popping high active reg: %d\n", high_reg); \
+ \
+ low_reg = (unsigned) POP_FAILURE_ITEM (); \
+ DEBUG_PRINT2 (" Popping low active reg: %d\n", low_reg); \
+ \
+ for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
+ { \
+ DEBUG_PRINT2 (" Popping reg: %d\n", this_reg); \
+ \
+ reg_info[this_reg].word = POP_FAILURE_ITEM (); \
+ DEBUG_PRINT2 (" info: 0x%x\n", reg_info[this_reg]); \
+ \
+ regend[this_reg] = (const char *) POP_FAILURE_ITEM (); \
+ DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
+ \
+ regstart[this_reg] = (const char *) POP_FAILURE_ITEM (); \
+ DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
+ } \
+ \
+ DEBUG_STATEMENT (nfailure_points_popped++); \
+} /* POP_FAILURE_POINT */
-/* Given a pattern, compute a fastmap from it. The fastmap records
- which of the (1 << BYTEWIDTH) possible characters can start a string
- that matches the pattern. This fastmap is used by re_search to skip
- quickly over totally implausible text.
+/* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
+ BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
+ characters can start a string that matches the pattern. This fastmap
+ is used by re_search to skip quickly over impossible starting points.
- The caller must supply the address of a (1 << BYTEWIDTH)-byte data
- area as bufp->fastmap.
- The other components of bufp describe the pattern to be used. */
+ The caller must supply the address of a (1 << BYTEWIDTH)-byte data
+ area as BUFP->fastmap.
+
+ We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
+ the pattern buffer.
-void
+ Returns 0 if we succeed, -2 if an internal error. */
+
+int
re_compile_fastmap (bufp)
struct re_pattern_buffer *bufp;
{
- unsigned char *pattern = (unsigned char *) bufp->buffer;
- int size = bufp->used;
+ int j, k;
+ fail_stack_type fail_stack;
+#ifndef REGEX_MALLOC
+ char *destination;
+#endif
+ /* We don't push any register information onto the failure stack. */
+ unsigned num_regs = 0;
+
register char *fastmap = bufp->fastmap;
- register unsigned char *p = pattern;
+ unsigned char *pattern = bufp->buffer;
+ unsigned long size = bufp->used;
+ const unsigned char *p = pattern;
register unsigned char *pend = pattern + size;
- register int j, k;
- unsigned char *translate = (unsigned char *) bufp->translate;
- unsigned char *stackb[NFAILURES];
- unsigned char **stackp = stackb;
+ /* Assume that each path through the pattern can be null until
+ proven otherwise. We set this false at the bottom of switch
+ statement, to which we get only if a particular path doesn't
+ match the empty string. */
+ boolean path_can_be_null = true;
- unsigned is_a_succeed_n;
+ /* We aren't doing a `succeed_n' to begin with. */
+ boolean succeed_n_p = false;
- bzero (fastmap, (1 << BYTEWIDTH));
- bufp->fastmap_accurate = 1;
+ assert (fastmap != NULL && p != NULL);
+
+ INIT_FAIL_STACK ();
+ bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
+ bufp->fastmap_accurate = 1; /* It will be when we're done. */
bufp->can_be_null = 0;
- while (p)
+ while (p != pend || !FAIL_STACK_EMPTY ())
{
- is_a_succeed_n = 0;
if (p == pend)
- {
- bufp->can_be_null = 1;
- break;
+ {
+ bufp->can_be_null |= path_can_be_null;
+
+ /* Reset for next path. */
+ path_can_be_null = true;
+
+ p = fail_stack.stack[--fail_stack.avail];
}
+
+ /* We should never be about to go beyond the end of the pattern. */
+ assert (p < pend);
+
#ifdef SWITCH_ENUM_BUG
- switch ((int) ((enum regexpcode) *p++))
+ switch ((int) ((re_opcode_t) *p++))
#else
- switch ((enum regexpcode) *p++)
+ switch ((re_opcode_t) *p++)
#endif
{
+
+ /* I guess the idea here is to simply not bother with a fastmap
+ if a backreference is used, since it's too hard to figure out
+ the fastmap for the corresponding group. Setting
+ `can_be_null' stops `re_search_2' from using the fastmap, so
+ that is all we do. */
+ case duplicate:
+ bufp->can_be_null = 1;
+ return 0;
+
+
+ /* Following are the cases which match a character. These end
+ with `break'. */
+
case exactn:
- if (translate)
- fastmap[translate[p[1]]] = 1;
- else
- fastmap[p[1]] = 1;
+ fastmap[p[1]] = 1;
break;
- case begline:
- case before_dot:
+
+ case charset:
+ for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
+ if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
+ fastmap[j] = 1;
+ break;
+
+
+ case charset_not:
+ /* Chars beyond end of map must be allowed. */
+ for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
+ fastmap[j] = 1;
+
+ for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
+ if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
+ fastmap[j] = 1;
+ break;
+
+
+ case wordchar:
+ for (j = 0; j < (1 << BYTEWIDTH); j++)
+ if (SYNTAX (j) == Sword)
+ fastmap[j] = 1;
+ break;
+
+
+ case notwordchar:
+ for (j = 0; j < (1 << BYTEWIDTH); j++)
+ if (SYNTAX (j) != Sword)
+ fastmap[j] = 1;
+ break;
+
+
+ case anychar:
+ /* `.' matches anything ... */
+ for (j = 0; j < (1 << BYTEWIDTH); j++)
+ fastmap[j] = 1;
+
+ /* ... except perhaps newline. */
+ if (!(bufp->syntax & RE_DOT_NEWLINE))
+ fastmap['\n'] = 0;
+
+ /* Return if we have already set `can_be_null'; if we have,
+ then the fastmap is irrelevant. Something's wrong here. */
+ else if (bufp->can_be_null)
+ return 0;
+
+ /* Otherwise, have to check alternative paths. */
+ break;
+
+
+#ifdef emacs
+ case syntaxspec:
+ k = *p++;
+ for (j = 0; j < (1 << BYTEWIDTH); j++)
+ if (SYNTAX (j) == (enum syntaxcode) k)
+ fastmap[j] = 1;
+ break;
+
+
+ case notsyntaxspec:
+ k = *p++;
+ for (j = 0; j < (1 << BYTEWIDTH); j++)
+ if (SYNTAX (j) != (enum syntaxcode) k)
+ fastmap[j] = 1;
+ break;
+
+
+ /* All cases after this match the empty string. These end with
+ `continue'. */
+
+
+ case before_dot:
case at_dot:
case after_dot:
+ continue;
+#endif /* not emacs */
+
+
+ case no_op:
+ case begline:
+ case endline:
case begbuf:
case endbuf:
case wordbound:
case notwordbound:
case wordbeg:
case wordend:
+ case push_dummy_failure:
continue;
- case endline:
- if (translate)
- fastmap[translate['\n']] = 1;
- else
- fastmap['\n'] = 1;
-
- if (bufp->can_be_null != 1)
- bufp->can_be_null = 2;
- break;
case jump_n:
- case finalize_jump:
- case maybe_finalize_jump:
+ case pop_failure_jump:
+ case maybe_pop_jump:
case jump:
+ case jump_past_alt:
case dummy_failure_jump:
EXTRACT_NUMBER_AND_INCR (j, p);
p += j;
if (j > 0)
continue;
- /* Jump backward reached implies we just went through
- the body of a loop and matched nothing.
- Opcode jumped to should be an on_failure_jump.
- Just treat it like an ordinary jump.
- For a * loop, it has pushed its failure point already;
- If so, discard that as redundant. */
-
- if ((enum regexpcode) *p != on_failure_jump
- && (enum regexpcode) *p != succeed_n)
+
+ /* Jump backward implies we just went through the body of a
+ loop and matched nothing. Opcode jumped to should be
+ `on_failure_jump' or `succeed_n'. Just treat it like an
+ ordinary jump. For a * loop, it has pushed its failure
+ point already; if so, discard that as redundant. */
+ if ((re_opcode_t) *p != on_failure_jump
+ && (re_opcode_t) *p != succeed_n)
continue;
+
p++;
EXTRACT_NUMBER_AND_INCR (j, p);
p += j;
- if (stackp != stackb && *stackp == p)
- stackp--;
- continue;
+ /* If what's on the stack is where we are now, pop it. */
+ if (!FAIL_STACK_EMPTY ()
+ && fail_stack.stack[fail_stack.avail - 1] == p)
+ fail_stack.avail--;
+
+ continue;
+
+
case on_failure_jump:
+ case on_failure_keep_string_jump:
handle_on_failure_jump:
EXTRACT_NUMBER_AND_INCR (j, p);
- *++stackp = p + j;
- if (is_a_succeed_n)
- EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
- continue;
+
+ /* For some patterns, e.g., `(a?)?', `p+j' here points to the
+ end of the pattern. We don't want to push such a point,
+ since when we restore it above, entering the switch will
+ increment `p' past the end of the pattern. We don't need
+ to push such a point since we obviously won't find any more
+ fastmap entries beyond `pend'. Such a pattern can match
+ the null string, though. */
+ if (p + j < pend)
+ {
+ if (!PUSH_PATTERN_OP (p + j, fail_stack))
+ return -2;
+ }
+ else
+ bufp->can_be_null = 1;
+
+ if (succeed_n_p)
+ {
+ EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
+ succeed_n_p = false;
+ }
+
+ continue;
+
case succeed_n:
- is_a_succeed_n = 1;
/* Get to the number of times to succeed. */
p += 2;
- /* Increment p past the n for when k != 0. */
+
+ /* Increment p past the n for when k != 0. */
EXTRACT_NUMBER_AND_INCR (k, p);
if (k == 0)
{
p -= 4;
+ succeed_n_p = true; /* Spaghetti code alert. */
goto handle_on_failure_jump;
}
continue;
-
+
+
case set_number_at:
p += 4;
continue;
- case start_memory:
- case stop_memory:
- p++;
- continue;
- case duplicate:
- bufp->can_be_null = 1;
- fastmap['\n'] = 1;
- case anychar:
- for (j = 0; j < (1 << BYTEWIDTH); j++)
- if (j != '\n')
- fastmap[j] = 1;
- if (bufp->can_be_null)
- return;
- /* Don't return; check the alternative paths
- so we can set can_be_null if appropriate. */
- break;
-
- case wordchar:
- for (j = 0; j < (1 << BYTEWIDTH); j++)
- if (SYNTAX (j) == Sword)
- fastmap[j] = 1;
- break;
-
- case notwordchar:
- for (j = 0; j < (1 << BYTEWIDTH); j++)
- if (SYNTAX (j) != Sword)
- fastmap[j] = 1;
- break;
-
-#ifdef emacs
- case syntaxspec:
- k = *p++;
- for (j = 0; j < (1 << BYTEWIDTH); j++)
- if (SYNTAX (j) == (enum syntaxcode) k)
- fastmap[j] = 1;
- break;
+ case start_memory:
+ case stop_memory:
+ p += 2;
+ continue;
- case notsyntaxspec:
- k = *p++;
- for (j = 0; j < (1 << BYTEWIDTH); j++)
- if (SYNTAX (j) != (enum syntaxcode) k)
- fastmap[j] = 1;
- break;
-#endif /* not emacs */
- case charset:
- for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
- if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
- {
- if (translate)
- fastmap[translate[j]] = 1;
- else
- fastmap[j] = 1;
- }
- break;
+ default:
+ abort (); /* We have listed all the cases. */
+ } /* switch *p++ */
+
+ /* Getting here means we have found the possible starting
+ characters for one path of the pattern -- and that the empty
+ string does not match. We need not follow this path further.
+ Instead, look at the next alternative (remembered on the
+ stack), or quit if no more. The test at the top of the loop
+ does these things. */
+ path_can_be_null = false;
+ p = pend;
+ } /* while p */
+
+ /* Set `can_be_null' for the last path (also the first path, if the
+ pattern is empty). */
+ bufp->can_be_null |= path_can_be_null;
+ return 0;
+} /* re_compile_fastmap */
+
+/* Set REGS to hold NUM_REGS registers, storing them in STARTS and
+ ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
+ this memory for recording register information. STARTS and ENDS
+ must be allocated using the malloc library routine, and must each
+ be at least NUM_REGS * sizeof (regoff_t) bytes long.
- case charset_not:
- /* Chars beyond end of map must be allowed */
- for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
- if (translate)
- fastmap[translate[j]] = 1;
- else
- fastmap[j] = 1;
+ If NUM_REGS == 0, then subsequent matches should allocate their own
+ register data.
- for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
- if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
- {
- if (translate)
- fastmap[translate[j]] = 1;
- else
- fastmap[j] = 1;
- }
- break;
- }
+ Unless this function is called, the first search or match using
+ PATTERN_BUFFER will allocate its own register data, without
+ freeing the old data. */
- /* Get here means we have successfully found the possible starting
- characters of one path of the pattern. We need not follow this
- path any farther. Instead, look at the next alternative
- remembered in the stack. */
- if (stackp != stackb)
- p = *stackp--;
- else
- break;
+void
+re_set_registers (bufp, regs, num_regs, starts, ends)
+ struct re_pattern_buffer *bufp;
+ struct re_registers *regs;
+ unsigned num_regs;
+ regoff_t *starts, *ends;
+{
+ if (num_regs)
+ {
+ bufp->regs_allocated = REGS_REALLOCATE;
+ regs->num_regs = num_regs;
+ regs->start = starts;
+ regs->end = ends;
+ }
+ else
+ {
+ bufp->regs_allocated = REGS_UNALLOCATED;
+ regs->num_regs = 0;
+ regs->start = regs->end = (regoff_t) 0;
}
}
-
-
+/* Searching routines. */
+
/* Like re_search_2, below, but only one string is specified, and
doesn't let you say where to stop matching. */
int
-re_search (pbufp, string, size, startpos, range, regs)
- struct re_pattern_buffer *pbufp;
- char *string;
+re_search (bufp, string, size, startpos, range, regs)
+ struct re_pattern_buffer *bufp;
+ const char *string;
int size, startpos, range;
struct re_registers *regs;
{
- return re_search_2 (pbufp, (char *) 0, 0, string, size, startpos, range,
+ return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
regs, size);
}
-/* Using the compiled pattern in PBUFP->buffer, first tries to match the
+/* Using the compiled pattern in BUFP->buffer, first tries to match the
virtual concatenation of STRING1 and STRING2, starting first at index
- STARTPOS, then at STARTPOS + 1, and so on. RANGE is the number of
- places to try before giving up. If RANGE is negative, it searches
- backwards, i.e., the starting positions tried are STARTPOS, STARTPOS
- - 1, etc. STRING1 and STRING2 are of SIZE1 and SIZE2, respectively.
+ STARTPOS, then at STARTPOS + 1, and so on.
+
+ STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
+
+ RANGE is how far to scan while trying to match. RANGE = 0 means try
+ only at STARTPOS; in general, the last start tried is STARTPOS +
+ RANGE.
+
In REGS, return the indices of the virtual concatenation of STRING1
- and STRING2 that matched the entire PBUFP->buffer and its contained
- subexpressions. Do not consider matching one past the index MSTOP in
- the virtual concatenation of STRING1 and STRING2.
+ and STRING2 that matched the entire BUFP->buffer and its contained
+ subexpressions.
+
+ Do not consider matching one past the index STOP in the virtual
+ concatenation of STRING1 and STRING2.
- The value returned is the position in the strings at which the match
- was found, or -1 if no match was found, or -2 if error (such as
- failure stack overflow). */
+ We return either the position in the strings at which the match was
+ found, -1 if no match, or -2 if error (such as failure
+ stack overflow). */
int
-re_search_2 (pbufp, string1, size1, string2, size2, startpos, range,
- regs, mstop)
- struct re_pattern_buffer *pbufp;
- char *string1, *string2;
+re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
+ struct re_pattern_buffer *bufp;
+ const char *string1, *string2;
int size1, size2;
int startpos;
- register int range;
+ int range;
struct re_registers *regs;
- int mstop;
+ int stop;
{
- register char *fastmap = pbufp->fastmap;
- register unsigned char *translate = (unsigned char *) pbufp->translate;
+ int val;
+ register char *fastmap = bufp->fastmap;
+ register char *translate = bufp->translate;
int total_size = size1 + size2;
int endpos = startpos + range;
- int val;
- /* Check for out-of-range starting position. */
- if (startpos < 0 || startpos > total_size)
+ /* Check for out-of-range STARTPOS. */
+ if (startpos < 0 || startpos > total_size)
return -1;
- /* Fix up range if it would eventually take startpos outside of the
- virtual concatenation of string1 and string2. */
+ /* Fix up RANGE if it might eventually take us outside
+ the virtual concatenation of STRING1 and STRING2. */
if (endpos < -1)
range = -1 - startpos;
else if (endpos > total_size)
range = total_size - startpos;
- /* Update the fastmap now if not correct already. */
- if (fastmap && !pbufp->fastmap_accurate)
- re_compile_fastmap (pbufp);
-
/* If the search isn't to be a backwards one, don't waste time in a
- long search for a pattern that says it is anchored. */
- if (pbufp->used > 0 && (enum regexpcode) pbufp->buffer[0] == begbuf
- && range > 0)
+ search for a pattern that must be anchored. */
+ if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0)
{
if (startpos > 0)
return -1;
@@ -1578,63 +2944,68 @@ re_search_2 (pbufp, string1, size1, string2, size2, startpos, range,
range = 1;
}
- while (1)
+ /* Update the fastmap now if not correct already. */
+ if (fastmap && !bufp->fastmap_accurate)
+ if (re_compile_fastmap (bufp) == -2)
+ return -2;
+
+ /* Loop through the string, looking for a place to start matching. */
+ for (;;)
{
/* If a fastmap is supplied, skip quickly over characters that
- cannot possibly be the start of a match. Note, however, that
- if the pattern can possibly match the null string, we must
- test it at each starting point so that we take the first null
- string we get. */
-
- if (fastmap && startpos < total_size && pbufp->can_be_null != 1)
+ cannot be the start of a match. If the pattern can match the
+ null string, however, we don't need to skip characters; we want
+ the first null string. */
+ if (fastmap && startpos < total_size && !bufp->can_be_null)
{
if (range > 0) /* Searching forwards. */
{
+ register const char *d;
register int lim = 0;
- register unsigned char *p;
int irange = range;
- if (startpos < size1 && startpos + range >= size1)
- lim = range - (size1 - startpos);
- p = ((unsigned char *)
- &(startpos >= size1 ? string2 - size1 : string1)[startpos]);
+ if (startpos < size1 && startpos + range >= size1)
+ lim = range - (size1 - startpos);
+
+ d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
+
+ /* Written out as an if-else to avoid testing `translate'
+ inside the loop. */
+ if (translate)
+ while (range > lim
+ && !fastmap[(unsigned char)
+ translate[(unsigned char) *d++]])
+ range--;
+ else
+ while (range > lim && !fastmap[(unsigned char) *d++])
+ range--;
- while (range > lim && !fastmap[translate
- ? translate[*p++]
- : *p++])
- range--;
startpos += irange - range;
}
else /* Searching backwards. */
{
- register unsigned char c;
-
- if (string1 == 0 || startpos >= size1)
- c = string2[startpos - size1];
- else
- c = string1[startpos];
+ register char c = (size1 == 0 || startpos >= size1
+ ? string2[startpos - size1]
+ : string1[startpos]);
- c &= 0xff;
- if (translate ? !fastmap[translate[c]] : !fastmap[c])
+ if (!fastmap[(unsigned char) TRANSLATE (c)])
goto advance;
}
}
- if (range >= 0 && startpos == total_size
- && fastmap && pbufp->can_be_null == 0)
+ /* If can't match the null string, and that's all we have left, fail. */
+ if (range >= 0 && startpos == total_size && fastmap
+ && !bufp->can_be_null)
return -1;
- val = re_match_2 (pbufp, string1, size1, string2, size2, startpos,
- regs, mstop);
+ val = re_match_2 (bufp, string1, size1, string2, size2,
+ startpos, regs, stop);
if (val >= 0)
return startpos;
+
if (val == -2)
return -2;
-#ifdef C_ALLOCA
- alloca (0);
-#endif /* C_ALLOCA */
-
advance:
if (!range)
break;
@@ -1650,243 +3021,229 @@ re_search_2 (pbufp, string1, size1, string2, size2, startpos, range,
}
}
return -1;
-}
-
-
+} /* re_search_2 */
-#ifndef emacs /* emacs never uses this. */
-int
-re_match (pbufp, string, size, pos, regs)
- struct re_pattern_buffer *pbufp;
- char *string;
- int size, pos;
- struct re_registers *regs;
-{
- return re_match_2 (pbufp, (char *) 0, 0, string, size, pos, regs, size);
-}
-#endif /* not emacs */
+/* Declarations and macros for re_match_2. */
-
-/* The following are used for re_match_2, defined below: */
-
-/* Roughly the maximum number of failure points on the stack. Would be
- exactly that if always pushed MAX_NUM_FAILURE_ITEMS each time we failed. */
-
-int re_max_failures = 2000;
-
-/* Routine used by re_match_2. */
static int bcmp_translate ();
-
-
-/* Structure and accessing macros used in re_match_2: */
-
-struct register_info
+static boolean alt_match_null_string_p (),
+ common_op_match_null_string_p (),
+ group_match_null_string_p ();
+
+/* Structure for per-register (a.k.a. per-group) information.
+ This must not be longer than one word, because we push this value
+ onto the failure stack. Other register information, such as the
+ starting and ending positions (which are addresses), and the list of
+ inner groups (which is a bits list) are maintained in separate
+ variables.
+
+ We are making a (strictly speaking) nonportable assumption here: that
+ the compiler will pack our bit fields into something that fits into
+ the type of `word', i.e., is something that fits into one item on the
+ failure stack. */
+typedef union
{
- unsigned is_active : 1;
- unsigned matched_something : 1;
-};
-
-#define IS_ACTIVE(R) ((R).is_active)
-#define MATCHED_SOMETHING(R) ((R).matched_something)
-
-
-/* Macros used by re_match_2: */
-
-
-/* I.e., regstart, regend, and reg_info. */
-
-#define NUM_REG_ITEMS 3
-
-/* We push at most this many things on the stack whenever we
- fail. The `+ 2' refers to PATTERN_PLACE and STRING_PLACE, which are
- arguments to the PUSH_FAILURE_POINT macro. */
-
-#define MAX_NUM_FAILURE_ITEMS (RE_NREGS * NUM_REG_ITEMS + 2)
-
-
-/* We push this many things on the stack whenever we fail. */
-
-#define NUM_FAILURE_ITEMS (last_used_reg * NUM_REG_ITEMS + 2)
-
-
-/* This pushes most of the information about the current state we will want
- if we ever fail back to it. */
-
-#define PUSH_FAILURE_POINT(pattern_place, string_place) \
- { \
- short last_used_reg, this_reg; \
- \
- /* Find out how many registers are active or have been matched. \
- (Aside from register zero, which is only set at the end.) */ \
- for (last_used_reg = RE_NREGS - 1; last_used_reg > 0; last_used_reg--)\
- if (regstart[last_used_reg] != (unsigned char *) -1) \
- break; \
- \
- if (stacke - stackp < NUM_FAILURE_ITEMS) \
- { \
- unsigned char **stackx; \
- if (stacke - stackb > re_max_failures * MAX_NUM_FAILURE_ITEMS) \
- return -2; \
- \
- /* Roughly double the size of the stack. */ \
- stackx = (unsigned char **) alloca (2 * MAX_NUM_FAILURE_ITEMS \
- * (stacke - stackb) \
- * sizeof (unsigned char *));\
- /* Only copy what is in use. */ \
- bcopy (stackb, stackx, (stackp - stackb) * sizeof (char *)); \
- stackp = stackx + (stackp - stackb); \
- stackb = stackx; \
- stacke = stackb + 2 * MAX_NUM_FAILURE_ITEMS * (stacke - stackb);\
- } \
- \
- /* Now push the info for each of those registers. */ \
- for (this_reg = 1; this_reg <= last_used_reg; this_reg++) \
- { \
- *stackp++ = regstart[this_reg]; \
- *stackp++ = regend[this_reg]; \
- *stackp++ = (unsigned char *) &reg_info[this_reg]; \
- } \
- \
- /* Push how many registers we saved. */ \
- *stackp++ = (unsigned char *) last_used_reg; \
- \
- *stackp++ = pattern_place; \
- *stackp++ = string_place; \
- }
-
-
-/* This pops what PUSH_FAILURE_POINT pushes. */
-
-#define POP_FAILURE_POINT() \
- { \
- int temp; \
- stackp -= 2; /* Remove failure points. */ \
- temp = (int) *--stackp; /* How many regs pushed. */ \
- temp *= NUM_REG_ITEMS; /* How much to take off the stack. */ \
- stackp -= temp; /* Remove the register info. */ \
- }
-
+ fail_stack_elt_t word;
+ struct
+ {
+ /* This field is one if this group can match the empty string,
+ zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
+#define MATCH_NULL_UNSET_VALUE 3
+ unsigned match_null_string_p : 2;
+ unsigned is_active : 1;
+ unsigned matched_something : 1;
+ unsigned ever_matched_something : 1;
+ } bits;
+} register_info_type;
+
+#define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
+#define IS_ACTIVE(R) ((R).bits.is_active)
+#define MATCHED_SOMETHING(R) ((R).bits.matched_something)
+#define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
+
+
+/* Call this when have matched a real character; it sets `matched' flags
+ for the subexpressions which we are currently inside. Also records
+ that those subexprs have matched. */
+#define SET_REGS_MATCHED() \
+ do \
+ { \
+ unsigned r; \
+ for (r = lowest_active_reg; r <= highest_active_reg; r++) \
+ { \
+ MATCHED_SOMETHING (reg_info[r]) \
+ = EVER_MATCHED_SOMETHING (reg_info[r]) \
+ = 1; \
+ } \
+ } \
+ while (0)
+
+
+/* This converts PTR, a pointer into one of the search strings `string1'
+ and `string2' into an offset from the beginning of that string. */
+#define POINTER_TO_OFFSET(ptr) \
+ (FIRST_STRING_P (ptr) ? (ptr) - string1 : (ptr) - string2 + size1)
+
+/* Registers are set to a sentinel when they haven't yet matched. */
+#define REG_UNSET_VALUE ((char *) -1)
+#define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
+
+
+/* Macros for dealing with the split strings in re_match_2. */
#define MATCHING_IN_FIRST_STRING (dend == end_match_1)
-/* Is true if there is a first string and if PTR is pointing anywhere
- inside it or just past the end. */
-
-#define IS_IN_FIRST_STRING(ptr) \
- (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
-
/* Call before fetching a character with *d. This switches over to
string2 if necessary. */
+#define PREFETCH() \
+ while (d == dend) \
+ { \
+ /* End of string2 => fail. */ \
+ if (dend == end_match_2) \
+ goto fail; \
+ /* End of string1 => advance to string2. */ \
+ d = string2; \
+ dend = end_match_2; \
+ }
-#define PREFETCH \
- while (d == dend) \
- { \
- /* end of string2 => fail. */ \
- if (dend == end_match_2) \
- goto fail; \
- /* end of string1 => advance to string2. */ \
- d = string2; \
- dend = end_match_2; \
- }
-
-
-/* Call this when have matched something; it sets `matched' flags for the
- registers corresponding to the subexpressions of which we currently
- are inside. */
-#define SET_REGS_MATCHED \
- { unsigned this_reg; \
- for (this_reg = 0; this_reg < RE_NREGS; this_reg++) \
- { \
- if (IS_ACTIVE(reg_info[this_reg])) \
- MATCHED_SOMETHING(reg_info[this_reg]) = 1; \
- else \
- MATCHED_SOMETHING(reg_info[this_reg]) = 0; \
- } \
- }
/* Test if at very beginning or at very end of the virtual concatenation
- of string1 and string2. If there is only one string, we've put it in
- string2. */
-
-#define AT_STRINGS_BEG (d == (size1 ? string1 : string2) || !size2)
-#define AT_STRINGS_END (d == end2)
-
-#define AT_WORD_BOUNDARY \
- (AT_STRINGS_BEG || AT_STRINGS_END || IS_A_LETTER (d - 1) != IS_A_LETTER (d))
-
-/* We have two special cases to check for:
- 1) if we're past the end of string1, we have to look at the first
- character in string2;
- 2) if we're before the beginning of string2, we have to look at the
- last character in string1; we assume there is a string1, so use
- this in conjunction with AT_STRINGS_BEG. */
-#define IS_A_LETTER(d) \
- (SYNTAX ((d) == end1 ? *string2 : (d) == string2 - 1 ? *(end1 - 1) : *(d))\
+ of `string1' and `string2'. If only one string, it's `string2'. */
+#define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
+#define AT_STRINGS_END(d) ((d) == end2)
+
+
+/* Test if D points to a character which is word-constituent. We have
+ two special cases to check for: if past the end of string1, look at
+ the first character in string2; and if before the beginning of
+ string2, look at the last character in string1. */
+#define WORDCHAR_P(d) \
+ (SYNTAX ((d) == end1 ? *string2 \
+ : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
== Sword)
+/* Test if the character before D and the one at D differ with respect
+ to being word-constituent. */
+#define AT_WORD_BOUNDARY(d) \
+ (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
+ || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
+
+
+/* Free everything we malloc. */
+#ifdef REGEX_MALLOC
+#define FREE_VAR(var) if (var) free (var); var = NULL
+#define FREE_VARIABLES() \
+ do { \
+ FREE_VAR (fail_stack.stack); \
+ FREE_VAR (regstart); \
+ FREE_VAR (regend); \
+ FREE_VAR (old_regstart); \
+ FREE_VAR (old_regend); \
+ FREE_VAR (best_regstart); \
+ FREE_VAR (best_regend); \
+ FREE_VAR (reg_info); \
+ FREE_VAR (reg_dummy); \
+ FREE_VAR (reg_info_dummy); \
+ } while (0)
+#else /* not REGEX_MALLOC */
+/* Some MIPS systems (at least) want this to free alloca'd storage. */
+#define FREE_VARIABLES() alloca (0)
+#endif /* not REGEX_MALLOC */
+
+
+/* These values must meet several constraints. They must not be valid
+ register values; since we have a limit of 255 registers (because
+ we use only one byte in the pattern for the register number), we can
+ use numbers larger than 255. They must differ by 1, because of
+ NUM_FAILURE_ITEMS above. And the value for the lowest register must
+ be larger than the value for the highest register, so we do not try
+ to actually save any registers when none are active. */
+#define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
+#define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
+
+/* Matching routines. */
-/* Match the pattern described by PBUFP against the virtual
- concatenation of STRING1 and STRING2, which are of SIZE1 and SIZE2,
- respectively. Start the match at index POS in the virtual
- concatenation of STRING1 and STRING2. In REGS, return the indices of
- the virtual concatenation of STRING1 and STRING2 that matched the
- entire PBUFP->buffer and its contained subexpressions. Do not
- consider matching one past the index MSTOP in the virtual
- concatenation of STRING1 and STRING2.
+#ifndef emacs /* Emacs never uses this. */
+/* re_match is like re_match_2 except it takes only a single string. */
- If pbufp->fastmap is nonzero, then it had better be up to date.
+int
+re_match (bufp, string, size, pos, regs)
+ struct re_pattern_buffer *bufp;
+ const char *string;
+ int size, pos;
+ struct re_registers *regs;
+ {
+ return re_match_2 (bufp, NULL, 0, string, size, pos, regs, size);
+}
+#endif /* not emacs */
- The reason that the data to match are specified as two components
- which are to be regarded as concatenated is so this function can be
- used directly on the contents of an Emacs buffer.
- -1 is returned if there is no match. -2 is returned if there is an
- error (such as match stack overflow). Otherwise the value is the
- length of the substring which was matched. */
+/* re_match_2 matches the compiled pattern in BUFP against the
+ the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
+ and SIZE2, respectively). We start matching at POS, and stop
+ matching at STOP.
+
+ If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
+ store offsets for the substring each group matched in REGS. See the
+ documentation for exactly how many groups we fill.
+
+ We return -1 if no match, -2 if an internal error (such as the
+ failure stack overflowing). Otherwise, we return the length of the
+ matched substring. */
int
-re_match_2 (pbufp, string1_arg, size1, string2_arg, size2, pos, regs, mstop)
- struct re_pattern_buffer *pbufp;
- char *string1_arg, *string2_arg;
+re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
+ struct re_pattern_buffer *bufp;
+ const char *string1, *string2;
int size1, size2;
int pos;
struct re_registers *regs;
- int mstop;
+ int stop;
{
- register unsigned char *p = (unsigned char *) pbufp->buffer;
+ /* General temporaries. */
+ int mcnt;
+ unsigned char *p1;
- /* Pointer to beyond end of buffer. */
- register unsigned char *pend = p + pbufp->used;
-
- unsigned char *string1 = (unsigned char *) string1_arg;
- unsigned char *string2 = (unsigned char *) string2_arg;
- unsigned char *end1; /* Just past end of first string. */
- unsigned char *end2; /* Just past end of second string. */
+ /* Just past the end of the corresponding string. */
+ const char *end1, *end2;
/* Pointers into string1 and string2, just past the last characters in
each to consider matching. */
- unsigned char *end_match_1, *end_match_2;
-
- register unsigned char *d, *dend;
- register int mcnt; /* Multipurpose. */
- unsigned char *translate = (unsigned char *) pbufp->translate;
- unsigned is_a_jump_n = 0;
+ const char *end_match_1, *end_match_2;
- /* Failure point stack. Each place that can handle a failure further
- down the line pushes a failure point on this stack. It consists of
- restart, regend, and reg_info for all registers corresponding to the
- subexpressions we're currently inside, plus the number of such
- registers, and, finally, two char *'s. The first char * is where to
- resume scanning the pattern; the second one is where to resume
- scanning the strings. If the latter is zero, the failure point is a
- ``dummy''; if a failure happens and the failure point is a dummy, it
- gets discarded and the next next one is tried. */
-
- unsigned char *initial_stack[MAX_NUM_FAILURE_ITEMS * NFAILURES];
- unsigned char **stackb = initial_stack;
- unsigned char **stackp = stackb;
- unsigned char **stacke = &stackb[MAX_NUM_FAILURE_ITEMS * NFAILURES];
+ /* Where we are in the data, and the end of the current string. */
+ const char *d, *dend;
+
+ /* Where we are in the pattern, and the end of the pattern. */
+ unsigned char *p = bufp->buffer;
+ register unsigned char *pend = p + bufp->used;
+
+ /* We use this to map every character in the string. */
+ char *translate = bufp->translate;
+
+ /* Failure point stack. Each place that can handle a failure further
+ down the line pushes a failure point on this stack. It consists of
+ restart, regend, and reg_info for all registers corresponding to
+ the subexpressions we're currently inside, plus the number of such
+ registers, and, finally, two char *'s. The first char * is where
+ to resume scanning the pattern; the second one is where to resume
+ scanning the strings. If the latter is zero, the failure point is
+ a ``dummy''; if a failure happens and the failure point is a dummy,
+ it gets discarded and the next next one is tried. */
+ fail_stack_type fail_stack;
+#ifdef DEBUG
+ static unsigned failure_id = 0;
+ unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
+#endif
+ /* We fill all the registers internally, independent of what we
+ return, for use in backreferences. The number here includes
+ an element for register zero. */
+ unsigned num_regs = bufp->re_nsub + 1;
+
+ /* The currently active registers. */
+ unsigned lowest_active_reg = NO_LOWEST_ACTIVE_REG;
+ unsigned highest_active_reg = NO_HIGHEST_ACTIVE_REG;
/* Information on the contents of registers. These are pointers into
the input strings; they record just what was matched (on this
@@ -1895,9 +3252,14 @@ re_match_2 (pbufp, string1_arg, size1, string2_arg, size2, pos, regs, mstop)
matching and the regnum-th regend points to right after where we
stopped matching the regnum-th subexpression. (The zeroth register
keeps track of what the whole pattern matches.) */
-
- unsigned char *regstart[RE_NREGS];
- unsigned char *regend[RE_NREGS];
+ const char **regstart, **regend;
+
+ /* If a group that's operated upon by a repetition operator fails to
+ match anything, then the register for its start will need to be
+ restored because it will have been set to wherever in the string we
+ are when we last see its open-group operator. Similarly for a
+ register's end. */
+ const char **old_regstart, **old_regend;
/* The is_active field of reg_info helps us keep track of which (possibly
nested) subexpressions we are currently in. The matched_something
@@ -1905,38 +3267,97 @@ re_match_2 (pbufp, string1_arg, size1, string2_arg, size2, pos, regs, mstop)
matched any of the pattern so far this time through the reg_num-th
subexpression. These two fields get reset each time through any
loop their register is in. */
-
- struct register_info reg_info[RE_NREGS];
-
+ register_info_type *reg_info;
/* The following record the register info as found in the above
variables when we find a match better than any we've seen before.
This happens as we backtrack through the failure points, which in
- turn happens only if we have not yet matched the entire string. */
-
- unsigned best_regs_set = 0;
- unsigned char *best_regstart[RE_NREGS];
- unsigned char *best_regend[RE_NREGS];
+ turn happens only if we have not yet matched the entire string. */
+ unsigned best_regs_set = false;
+ const char **best_regstart, **best_regend;
+
+ /* Logically, this is `best_regend[0]'. But we don't want to have to
+ allocate space for that if we're not allocating space for anything
+ else (see below). Also, we never need info about register 0 for
+ any of the other register vectors, and it seems rather a kludge to
+ treat `best_regend' differently than the rest. So we keep track of
+ the end of the best match so far in a separate variable. We
+ initialize this to NULL so that when we backtrack the first time
+ and need to test it, it's not garbage. */
+ const char *match_end = NULL;
+
+ /* Used when we pop values we don't care about. */
+ const char **reg_dummy;
+ register_info_type *reg_info_dummy;
+
+#ifdef DEBUG
+ /* Counts the total number of registers pushed. */
+ unsigned num_regs_pushed = 0;
+#endif
+ DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
+
+ INIT_FAIL_STACK ();
+
+ /* Do not bother to initialize all the register variables if there are
+ no groups in the pattern, as it takes a fair amount of time. If
+ there are groups, we include space for register 0 (the whole
+ pattern), even though we never use it, since it simplifies the
+ array indexing. We should fix this. */
+ if (bufp->re_nsub)
+ {
+ regstart = REGEX_TALLOC (num_regs, const char *);
+ regend = REGEX_TALLOC (num_regs, const char *);
+ old_regstart = REGEX_TALLOC (num_regs, const char *);
+ old_regend = REGEX_TALLOC (num_regs, const char *);
+ best_regstart = REGEX_TALLOC (num_regs, const char *);
+ best_regend = REGEX_TALLOC (num_regs, const char *);
+ reg_info = REGEX_TALLOC (num_regs, register_info_type);
+ reg_dummy = REGEX_TALLOC (num_regs, const char *);
+ reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type);
+
+ if (!(regstart && regend && old_regstart && old_regend && reg_info
+ && best_regstart && best_regend && reg_dummy && reg_info_dummy))
+ {
+ FREE_VARIABLES ();
+ return -2;
+ }
+ }
+#ifdef REGEX_MALLOC
+ else
+ {
+ /* We must initialize all our variables to NULL, so that
+ `FREE_VARIABLES' doesn't try to free them. */
+ regstart = regend = old_regstart = old_regend = best_regstart
+ = best_regend = reg_dummy = NULL;
+ reg_info = reg_info_dummy = (register_info_type *) NULL;
+ }
+#endif /* REGEX_MALLOC */
+ /* The starting position is bogus. */
+ if (pos < 0 || pos > size1 + size2)
+ {
+ FREE_VARIABLES ();
+ return -1;
+ }
+
/* Initialize subexpression text positions to -1 to mark ones that no
- \( or ( and \) or ) has been seen for. Also set all registers to
- inactive and mark them as not having matched anything or ever
- failed. */
- for (mcnt = 0; mcnt < RE_NREGS; mcnt++)
+ start_memory/stop_memory has been seen for. Also initialize the
+ register information struct. */
+ for (mcnt = 1; mcnt < num_regs; mcnt++)
{
- regstart[mcnt] = regend[mcnt] = (unsigned char *) -1;
+ regstart[mcnt] = regend[mcnt]
+ = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
+
+ REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
IS_ACTIVE (reg_info[mcnt]) = 0;
MATCHED_SOMETHING (reg_info[mcnt]) = 0;
+ EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
}
- if (regs)
- for (mcnt = 0; mcnt < RE_NREGS; mcnt++)
- regs->start[mcnt] = regs->end[mcnt] = -1;
-
- /* Set up pointers to ends of strings.
- Don't allow the second string to be empty unless both are empty. */
- if (size2 == 0)
+ /* We move `string1' into `string2' if the latter's empty -- but not if
+ `string1' is null. */
+ if (size2 == 0 && string1 != NULL)
{
string2 = string1;
size2 = size1;
@@ -1947,60 +3368,73 @@ re_match_2 (pbufp, string1_arg, size1, string2_arg, size2, pos, regs, mstop)
end2 = string2 + size2;
/* Compute where to stop matching, within the two strings. */
- if (mstop <= size1)
+ if (stop <= size1)
{
- end_match_1 = string1 + mstop;
+ end_match_1 = string1 + stop;
end_match_2 = string2;
}
else
{
end_match_1 = end1;
- end_match_2 = string2 + mstop - size1;
+ end_match_2 = string2 + stop - size1;
}
- /* `p' scans through the pattern as `d' scans through the data. `dend'
- is the end of the input string that `d' points within. `d' is
- advanced into the following input string whenever necessary, but
+ /* `p' scans through the pattern as `d' scans through the data.
+ `dend' is the end of the input string that `d' points within. `d'
+ is advanced into the following input string whenever necessary, but
this happens before fetching; therefore, at the beginning of the
loop, `d' can be pointing at the end of a string, but it cannot
- equal string2. */
-
- if (size1 != 0 && pos <= size1)
- d = string1 + pos, dend = end_match_1;
+ equal `string2'. */
+ if (size1 > 0 && pos <= size1)
+ {
+ d = string1 + pos;
+ dend = end_match_1;
+ }
else
- d = string2 + pos - size1, dend = end_match_2;
-
+ {
+ d = string2 + pos - size1;
+ dend = end_match_2;
+ }
+ DEBUG_PRINT1 ("The compiled pattern is: ");
+ DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
+ DEBUG_PRINT1 ("The string to match is: `");
+ DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
+ DEBUG_PRINT1 ("'\n");
+
/* This loops over pattern commands. It exits by returning from the
- function if match is complete, or it drops through if match fails
- at this starting point in the input data. */
-
- while (1)
+ function if the match is complete, or it drops through if the match
+ fails at this starting point in the input data. */
+ for (;;)
{
- is_a_jump_n = 0;
- /* End of pattern means we might have succeeded. */
+ DEBUG_PRINT2 ("\n0x%x: ", p);
+
if (p == pend)
- {
- /* If not end of string, try backtracking. Otherwise done. */
+ { /* End of pattern means we might have succeeded. */
+ DEBUG_PRINT1 ("end of pattern ... ");
+
+ /* If we haven't matched the entire string, and we want the
+ longest match, try backtracking. */
if (d != end_match_2)
{
- if (stackp != stackb)
- {
- /* More failure points to try. */
-
- unsigned in_same_string =
- IS_IN_FIRST_STRING (best_regend[0])
- == MATCHING_IN_FIRST_STRING;
+ DEBUG_PRINT1 ("backtracking.\n");
+
+ if (!FAIL_STACK_EMPTY ())
+ { /* More failure points to try. */
+ boolean same_str_p = (FIRST_STRING_P (match_end)
+ == MATCHING_IN_FIRST_STRING);
/* If exceeds best match so far, save it. */
- if (! best_regs_set
- || (in_same_string && d > best_regend[0])
- || (! in_same_string && ! MATCHING_IN_FIRST_STRING))
+ if (!best_regs_set
+ || (same_str_p && d > match_end)
+ || (!same_str_p && !MATCHING_IN_FIRST_STRING))
{
- best_regs_set = 1;
- best_regend[0] = d; /* Never use regstart[0]. */
+ best_regs_set = true;
+ match_end = d;
+
+ DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
- for (mcnt = 1; mcnt < RE_NREGS; mcnt++)
+ for (mcnt = 1; mcnt < num_regs; mcnt++)
{
best_regstart[mcnt] = regstart[mcnt];
best_regend[mcnt] = regend[mcnt];
@@ -2008,122 +3442,394 @@ re_match_2 (pbufp, string1_arg, size1, string2_arg, size2, pos, regs, mstop)
}
goto fail;
}
+
/* If no failure points, don't restore garbage. */
else if (best_regs_set)
{
- restore_best_regs:
- /* Restore best match. */
- d = best_regend[0];
+ restore_best_regs:
+ /* Restore best match. It may happen that `dend ==
+ end_match_1' while the restored d is in string2.
+ For example, the pattern `x.*y.*z' against the
+ strings `x-' and `y-z-', if the two strings are
+ not consecutive in memory. */
+ DEBUG_PRINT1 ("Restoring best registers.\n");
- for (mcnt = 0; mcnt < RE_NREGS; mcnt++)
+ d = match_end;
+ dend = ((d >= string1 && d <= end1)
+ ? end_match_1 : end_match_2);
+
+ for (mcnt = 1; mcnt < num_regs; mcnt++)
{
regstart[mcnt] = best_regstart[mcnt];
regend[mcnt] = best_regend[mcnt];
}
}
- }
+ } /* d != end_match_2 */
- /* If caller wants register contents data back, convert it
- to indices. */
- if (regs)
+ DEBUG_PRINT1 ("Accepting match.\n");
+
+ /* If caller wants register contents data back, do it. */
+ if (regs && !bufp->no_sub)
{
- regs->start[0] = pos;
- if (MATCHING_IN_FIRST_STRING)
- regs->end[0] = d - string1;
- else
- regs->end[0] = d - string2 + size1;
- for (mcnt = 1; mcnt < RE_NREGS; mcnt++)
+ /* Have the register data arrays been allocated? */
+ if (bufp->regs_allocated == REGS_UNALLOCATED)
+ { /* No. So allocate them with malloc. We need one
+ extra element beyond `num_regs' for the `-1' marker
+ GNU code uses. */
+ regs->num_regs = MAX (RE_NREGS, num_regs + 1);
+ regs->start = TALLOC (regs->num_regs, regoff_t);
+ regs->end = TALLOC (regs->num_regs, regoff_t);
+ if (regs->start == NULL || regs->end == NULL)
+ return -2;
+ bufp->regs_allocated = REGS_REALLOCATE;
+ }
+ else if (bufp->regs_allocated == REGS_REALLOCATE)
+ { /* Yes. If we need more elements than were already
+ allocated, reallocate them. If we need fewer, just
+ leave it alone. */
+ if (regs->num_regs < num_regs + 1)
+ {
+ regs->num_regs = num_regs + 1;
+ RETALLOC (regs->start, regs->num_regs, regoff_t);
+ RETALLOC (regs->end, regs->num_regs, regoff_t);
+ if (regs->start == NULL || regs->end == NULL)
+ return -2;
+ }
+ }
+ else
{
- if (regend[mcnt] == (unsigned char *) -1)
- {
- regs->start[mcnt] = -1;
- regs->end[mcnt] = -1;
- continue;
- }
- if (IS_IN_FIRST_STRING (regstart[mcnt]))
- regs->start[mcnt] = regstart[mcnt] - string1;
- else
- regs->start[mcnt] = regstart[mcnt] - string2 + size1;
-
- if (IS_IN_FIRST_STRING (regend[mcnt]))
- regs->end[mcnt] = regend[mcnt] - string1;
- else
- regs->end[mcnt] = regend[mcnt] - string2 + size1;
+ /* These braces fend off a "empty body in an else-statement"
+ warning under GCC when assert expands to nothing. */
+ assert (bufp->regs_allocated == REGS_FIXED);
}
- }
- return d - pos - (MATCHING_IN_FIRST_STRING
+
+ /* Convert the pointer data in `regstart' and `regend' to
+ indices. Register zero has to be set differently,
+ since we haven't kept track of any info for it. */
+ if (regs->num_regs > 0)
+ {
+ regs->start[0] = pos;
+ regs->end[0] = (MATCHING_IN_FIRST_STRING ? d - string1
+ : d - string2 + size1);
+ }
+
+ /* Go through the first `min (num_regs, regs->num_regs)'
+ registers, since that is all we initialized. */
+ for (mcnt = 1; mcnt < MIN (num_regs, regs->num_regs); mcnt++)
+ {
+ if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
+ regs->start[mcnt] = regs->end[mcnt] = -1;
+ else
+ {
+ regs->start[mcnt] = POINTER_TO_OFFSET (regstart[mcnt]);
+ regs->end[mcnt] = POINTER_TO_OFFSET (regend[mcnt]);
+ }
+ }
+
+ /* If the regs structure we return has more elements than
+ were in the pattern, set the extra elements to -1. If
+ we (re)allocated the registers, this is the case,
+ because we always allocate enough to have at least one
+ -1 at the end. */
+ for (mcnt = num_regs; mcnt < regs->num_regs; mcnt++)
+ regs->start[mcnt] = regs->end[mcnt] = -1;
+ } /* regs && !bufp->no_sub */
+
+ FREE_VARIABLES ();
+ DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
+ nfailure_points_pushed, nfailure_points_popped,
+ nfailure_points_pushed - nfailure_points_popped);
+ DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
+
+ mcnt = d - pos - (MATCHING_IN_FIRST_STRING
? string1
: string2 - size1);
+
+ DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
+
+ return mcnt;
}
/* Otherwise match next pattern command. */
#ifdef SWITCH_ENUM_BUG
- switch ((int) ((enum regexpcode) *p++))
+ switch ((int) ((re_opcode_t) *p++))
#else
- switch ((enum regexpcode) *p++)
+ switch ((re_opcode_t) *p++)
#endif
{
+ /* Ignore these. Used to ignore the n of succeed_n's which
+ currently have n == 0. */
+ case no_op:
+ DEBUG_PRINT1 ("EXECUTING no_op.\n");
+ break;
+
+
+ /* Match the next n pattern characters exactly. The following
+ byte in the pattern defines n, and the n bytes after that
+ are the characters to match. */
+ case exactn:
+ mcnt = *p++;
+ DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
+
+ /* This is written out as an if-else so we don't waste time
+ testing `translate' inside the loop. */
+ if (translate)
+ {
+ do
+ {
+ PREFETCH ();
+ if (translate[(unsigned char) *d++] != (char) *p++)
+ goto fail;
+ }
+ while (--mcnt);
+ }
+ else
+ {
+ do
+ {
+ PREFETCH ();
+ if (*d++ != (char) *p++) goto fail;
+ }
+ while (--mcnt);
+ }
+ SET_REGS_MATCHED ();
+ break;
+
+
+ /* Match any character except possibly a newline or a null. */
+ case anychar:
+ DEBUG_PRINT1 ("EXECUTING anychar.\n");
+
+ PREFETCH ();
+
+ if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
+ || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
+ goto fail;
+
+ SET_REGS_MATCHED ();
+ DEBUG_PRINT2 (" Matched `%d'.\n", *d);
+ d++;
+ break;
+
+
+ case charset:
+ case charset_not:
+ {
+ register unsigned char c;
+ boolean not = (re_opcode_t) *(p - 1) == charset_not;
+
+ DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
+
+ PREFETCH ();
+ c = TRANSLATE (*d); /* The character to match. */
+
+ /* Cast to `unsigned' instead of `unsigned char' in case the
+ bit list is a full 32 bytes long. */
+ if (c < (unsigned) (*p * BYTEWIDTH)
+ && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
+ not = !not;
+
+ p += 1 + *p;
+
+ if (!not) goto fail;
+
+ SET_REGS_MATCHED ();
+ d++;
+ break;
+ }
+
+
+ /* The beginning of a group is represented by start_memory.
+ The arguments are the register number in the next byte, and the
+ number of groups inner to this one in the next. The text
+ matched within the group is recorded (in the internal
+ registers data structure) under the register number. */
+ case start_memory:
+ DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]);
+
+ /* Find out if this group can match the empty string. */
+ p1 = p; /* To send to group_match_null_string_p. */
+
+ if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
+ REG_MATCH_NULL_STRING_P (reg_info[*p])
+ = group_match_null_string_p (&p1, pend, reg_info);
+
+ /* Save the position in the string where we were the last time
+ we were at this open-group operator in case the group is
+ operated upon by a repetition operator, e.g., with `(a*)*b'
+ against `ab'; then we want to ignore where we are now in
+ the string in case this attempt to match fails. */
+ old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
+ ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
+ : regstart[*p];
+ DEBUG_PRINT2 (" old_regstart: %d\n",
+ POINTER_TO_OFFSET (old_regstart[*p]));
- /* \( [or `(', as appropriate] is represented by start_memory,
- \) by stop_memory. Both of those commands are followed by
- a register number in the next byte. The text matched
- within the \( and \) is recorded under that number. */
- case start_memory:
regstart[*p] = d;
+ DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
+
IS_ACTIVE (reg_info[*p]) = 1;
MATCHED_SOMETHING (reg_info[*p]) = 0;
- p++;
+
+ /* This is the new highest active register. */
+ highest_active_reg = *p;
+
+ /* If nothing was active before, this is the new lowest active
+ register. */
+ if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
+ lowest_active_reg = *p;
+
+ /* Move past the register number and inner group count. */
+ p += 2;
break;
+
+ /* The stop_memory opcode represents the end of a group. Its
+ arguments are the same as start_memory's: the register
+ number, and the number of inner groups. */
case stop_memory:
+ DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]);
+
+ /* We need to save the string position the last time we were at
+ this close-group operator in case the group is operated
+ upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
+ against `aba'; then we want to ignore where we are now in
+ the string in case this attempt to match fails. */
+ old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
+ ? REG_UNSET (regend[*p]) ? d : regend[*p]
+ : regend[*p];
+ DEBUG_PRINT2 (" old_regend: %d\n",
+ POINTER_TO_OFFSET (old_regend[*p]));
+
regend[*p] = d;
- IS_ACTIVE (reg_info[*p]) = 0;
+ DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
- /* If just failed to match something this time around with a sub-
- expression that's in a loop, try to force exit from the loop. */
- if ((! MATCHED_SOMETHING (reg_info[*p])
- || (enum regexpcode) p[-3] == start_memory)
- && (p + 1) != pend)
+ /* This register isn't active anymore. */
+ IS_ACTIVE (reg_info[*p]) = 0;
+
+ /* If this was the only register active, nothing is active
+ anymore. */
+ if (lowest_active_reg == highest_active_reg)
{
- register unsigned char *p2 = p + 1;
+ lowest_active_reg = NO_LOWEST_ACTIVE_REG;
+ highest_active_reg = NO_HIGHEST_ACTIVE_REG;
+ }
+ else
+ { /* We must scan for the new highest active register, since
+ it isn't necessarily one less than now: consider
+ (a(b)c(d(e)f)g). When group 3 ends, after the f), the
+ new highest active register is 1. */
+ unsigned char r = *p - 1;
+ while (r > 0 && !IS_ACTIVE (reg_info[r]))
+ r--;
+
+ /* If we end up at register zero, that means that we saved
+ the registers as the result of an `on_failure_jump', not
+ a `start_memory', and we jumped to past the innermost
+ `stop_memory'. For example, in ((.)*) we save
+ registers 1 and 2 as a result of the *, but when we pop
+ back to the second ), we are at the stop_memory 1.
+ Thus, nothing is active. */
+ if (r == 0)
+ {
+ lowest_active_reg = NO_LOWEST_ACTIVE_REG;
+ highest_active_reg = NO_HIGHEST_ACTIVE_REG;
+ }
+ else
+ highest_active_reg = r;
+ }
+
+ /* If just failed to match something this time around with a
+ group that's operated on by a repetition operator, try to
+ force exit from the ``loop'', and restore the register
+ information for this group that we had before trying this
+ last match. */
+ if ((!MATCHED_SOMETHING (reg_info[*p])
+ || (re_opcode_t) p[-3] == start_memory)
+ && (p + 2) < pend)
+ {
+ boolean is_a_jump_n = false;
+
+ p1 = p + 2;
mcnt = 0;
- switch (*p2++)
+ switch ((re_opcode_t) *p1++)
{
case jump_n:
- is_a_jump_n = 1;
- case finalize_jump:
- case maybe_finalize_jump:
+ is_a_jump_n = true;
+ case pop_failure_jump:
+ case maybe_pop_jump:
case jump:
case dummy_failure_jump:
- EXTRACT_NUMBER_AND_INCR (mcnt, p2);
+ EXTRACT_NUMBER_AND_INCR (mcnt, p1);
if (is_a_jump_n)
- p2 += 2;
+ p1 += 2;
break;
+
+ default:
+ /* do nothing */ ;
}
- p2 += mcnt;
+ p1 += mcnt;
/* If the next operation is a jump backwards in the pattern
- to an on_failure_jump, exit from the loop by forcing a
- failure after pushing on the stack the on_failure_jump's
- jump in the pattern, and d. */
- if (mcnt < 0 && (enum regexpcode) *p2++ == on_failure_jump)
+ to an on_failure_jump right before the start_memory
+ corresponding to this stop_memory, exit from the loop
+ by forcing a failure after pushing on the stack the
+ on_failure_jump's jump in the pattern, and d. */
+ if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
+ && (re_opcode_t) p1[3] == start_memory && p1[4] == *p)
{
- EXTRACT_NUMBER_AND_INCR (mcnt, p2);
- PUSH_FAILURE_POINT (p2 + mcnt, d);
+ /* If this group ever matched anything, then restore
+ what its registers were before trying this last
+ failed match, e.g., with `(a*)*b' against `ab' for
+ regstart[1], and, e.g., with `((a*)*(b*)*)*'
+ against `aba' for regend[3].
+
+ Also restore the registers for inner groups for,
+ e.g., `((a*)(b*))*' against `aba' (register 3 would
+ otherwise get trashed). */
+
+ if (EVER_MATCHED_SOMETHING (reg_info[*p]))
+ {
+ unsigned r;
+
+ EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
+
+ /* Restore this and inner groups' (if any) registers. */
+ for (r = *p; r < *p + *(p + 1); r++)
+ {
+ regstart[r] = old_regstart[r];
+
+ /* xx why this test? */
+ if ((int) old_regend[r] >= (int) regstart[r])
+ regend[r] = old_regend[r];
+ }
+ }
+ p1++;
+ EXTRACT_NUMBER_AND_INCR (mcnt, p1);
+ PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
+
goto fail;
}
}
- p++;
+
+ /* Move past the register number and the inner group count. */
+ p += 2;
break;
+
/* \<digit> has been turned into a `duplicate' command which is
followed by the numeric value of <digit> as the register number. */
case duplicate:
{
- int regno = *p++; /* Get which register to match against */
- register unsigned char *d2, *dend2;
+ register const char *d2, *dend2;
+ int regno = *p++; /* Get which register to match against. */
+ DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
- /* Where in input to try to start matching. */
+ /* Can't back reference a group which we've never matched. */
+ if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
+ goto fail;
+
+ /* Where in input to try to start matching. */
d2 = regstart[regno];
/* Where to stop matching; if both the place to start and
@@ -2131,10 +3837,10 @@ re_match_2 (pbufp, string1_arg, size1, string2_arg, size2, pos, regs, mstop)
set to the place to stop, otherwise, for now have to use
the end of the first string. */
- dend2 = ((IS_IN_FIRST_STRING (regstart[regno])
- == IS_IN_FIRST_STRING (regend[regno]))
+ dend2 = ((FIRST_STRING_P (regstart[regno])
+ == FIRST_STRING_P (regend[regno]))
? regend[regno] : end_match_1);
- while (1)
+ for (;;)
{
/* If necessary, advance to next segment in register
contents. */
@@ -2142,13 +3848,16 @@ re_match_2 (pbufp, string1_arg, size1, string2_arg, size2, pos, regs, mstop)
{
if (dend2 == end_match_2) break;
if (dend2 == regend[regno]) break;
- d2 = string2, dend2 = regend[regno]; /* end of string1 => advance to string2. */
+
+ /* End of string1 => advance to string2. */
+ d2 = string2;
+ dend2 = regend[regno];
}
/* At end of register contents => success */
if (d2 == dend2) break;
/* If necessary, advance to next segment in data. */
- PREFETCH;
+ PREFETCH ();
/* How many characters left in this segment to match. */
mcnt = dend - d;
@@ -2169,188 +3878,323 @@ re_match_2 (pbufp, string1_arg, size1, string2_arg, size2, pos, regs, mstop)
}
break;
- case anychar:
- PREFETCH; /* Fetch a data character. */
- /* Match anything but a newline, maybe even a null. */
- if ((translate ? translate[*d] : *d) == '\n'
- || ((obscure_syntax & RE_DOT_NOT_NULL)
- && (translate ? translate[*d] : *d) == '\000'))
- goto fail;
- SET_REGS_MATCHED;
- d++;
- break;
- case charset:
- case charset_not:
- {
- int not = 0; /* Nonzero for charset_not. */
- register int c;
- if (*(p - 1) == (unsigned char) charset_not)
- not = 1;
-
- PREFETCH; /* Fetch a data character. */
+ /* begline matches the empty string at the beginning of the string
+ (unless `not_bol' is set in `bufp'), and, if
+ `newline_anchor' is set, after newlines. */
+ case begline:
+ DEBUG_PRINT1 ("EXECUTING begline.\n");
+
+ if (AT_STRINGS_BEG (d))
+ {
+ if (!bufp->not_bol) break;
+ }
+ else if (d[-1] == '\n' && bufp->newline_anchor)
+ {
+ break;
+ }
+ /* In all other cases, we fail. */
+ goto fail;
- if (translate)
- c = translate[*d];
- else
- c = *d;
- if (c < *p * BYTEWIDTH
- && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
- not = !not;
+ /* endline is the dual of begline. */
+ case endline:
+ DEBUG_PRINT1 ("EXECUTING endline.\n");
- p += 1 + *p;
+ if (AT_STRINGS_END (d))
+ {
+ if (!bufp->not_eol) break;
+ }
+
+ /* We have to ``prefetch'' the next character. */
+ else if ((d == end1 ? *string2 : *d) == '\n'
+ && bufp->newline_anchor)
+ {
+ break;
+ }
+ goto fail;
- if (!not) goto fail;
- SET_REGS_MATCHED;
- d++;
- break;
- }
- case begline:
- if ((size1 != 0 && d == string1)
- || (size1 == 0 && size2 != 0 && d == string2)
- || (d && d[-1] == '\n')
- || (size1 == 0 && size2 == 0))
+ /* Match at the very beginning of the data. */
+ case begbuf:
+ DEBUG_PRINT1 ("EXECUTING begbuf.\n");
+ if (AT_STRINGS_BEG (d))
break;
- else
- goto fail;
-
- case endline:
- if (d == end2
- || (d == end1 ? (size2 == 0 || *string2 == '\n') : *d == '\n'))
+ goto fail;
+
+
+ /* Match at the very end of the data. */
+ case endbuf:
+ DEBUG_PRINT1 ("EXECUTING endbuf.\n");
+ if (AT_STRINGS_END (d))
break;
- goto fail;
+ goto fail;
- /* `or' constructs are handled by starting each alternative with
- an on_failure_jump that points to the start of the next
- alternative. Each alternative except the last ends with a
- jump to the joining point. (Actually, each jump except for
- the last one really jumps to the following jump, because
- tensioning the jumps is a hassle.) */
- /* The start of a stupid repeat has an on_failure_jump that points
- past the end of the repeat text. This makes a failure point so
- that on failure to match a repetition, matching restarts past
- as many repetitions have been found with no way to fail and
- look for another one. */
+ /* on_failure_keep_string_jump is used to optimize `.*\n'. It
+ pushes NULL as the value for the string on the stack. Then
+ `pop_failure_point' will keep the current value for the
+ string, instead of restoring it. To see why, consider
+ matching `foo\nbar' against `.*\n'. The .* matches the foo;
+ then the . fails against the \n. But the next thing we want
+ to do is match the \n against the \n; if we restored the
+ string value, we would be back at the foo.
+
+ Because this is used only in specific cases, we don't need to
+ check all the things that `on_failure_jump' does, to make
+ sure the right things get saved on the stack. Hence we don't
+ share its code. The only reason to push anything on the
+ stack at all is that otherwise we would have to change
+ `anychar's code to do something besides goto fail in this
+ case; that seems worse than this. */
+ case on_failure_keep_string_jump:
+ DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
+
+ EXTRACT_NUMBER_AND_INCR (mcnt, p);
+ DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
+
+ PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
+ break;
- /* A smart repeat is similar but loops back to the on_failure_jump
- so that each repetition makes another failure point. */
+ /* Uses of on_failure_jump:
+
+ Each alternative starts with an on_failure_jump that points
+ to the beginning of the next alternative. Each alternative
+ except the last ends with a jump that in effect jumps past
+ the rest of the alternatives. (They really jump to the
+ ending jump of the following alternative, because tensioning
+ these jumps is a hassle.)
+
+ Repeats start with an on_failure_jump that points past both
+ the repetition text and either the following jump or
+ pop_failure_jump back to this on_failure_jump. */
case on_failure_jump:
on_failure:
+ DEBUG_PRINT1 ("EXECUTING on_failure_jump");
+
EXTRACT_NUMBER_AND_INCR (mcnt, p);
- PUSH_FAILURE_POINT (p + mcnt, d);
+ DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
+
+ /* If this on_failure_jump comes right before a group (i.e.,
+ the original * applied to a group), save the information
+ for that group and all inner ones, so that if we fail back
+ to this point, the group's information will be correct.
+ For example, in \(a*\)*\1, we need the preceding group,
+ and in \(\(a*\)b*\)\2, we need the inner group. */
+
+ /* We can't use `p' to check ahead because we push
+ a failure point to `p + mcnt' after we do this. */
+ p1 = p;
+
+ /* We need to skip no_op's before we look for the
+ start_memory in case this on_failure_jump is happening as
+ the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
+ against aba. */
+ while (p1 < pend && (re_opcode_t) *p1 == no_op)
+ p1++;
+
+ if (p1 < pend && (re_opcode_t) *p1 == start_memory)
+ {
+ /* We have a new highest active register now. This will
+ get reset at the start_memory we are about to get to,
+ but we will have saved all the registers relevant to
+ this repetition op, as described above. */
+ highest_active_reg = *(p1 + 1) + *(p1 + 2);
+ if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
+ lowest_active_reg = *(p1 + 1);
+ }
+
+ DEBUG_PRINT1 (":\n");
+ PUSH_FAILURE_POINT (p + mcnt, d, -2);
break;
- /* The end of a smart repeat has a maybe_finalize_jump back.
- Change it either to a finalize_jump or an ordinary jump. */
- case maybe_finalize_jump:
+
+ /* A smart repeat ends with `maybe_pop_jump'.
+ We change it to either `pop_failure_jump' or `jump'. */
+ case maybe_pop_jump:
EXTRACT_NUMBER_AND_INCR (mcnt, p);
- {
+ DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
+ {
register unsigned char *p2 = p;
- /* Compare what follows with the beginning of the repeat.
- If we can establish that there is nothing that they would
- both match, we can change to finalize_jump. */
- while (p2 + 1 != pend
- && (*p2 == (unsigned char) stop_memory
- || *p2 == (unsigned char) start_memory))
- p2 += 2; /* Skip over reg number. */
- if (p2 == pend)
- p[-3] = (unsigned char) finalize_jump;
- else if (*p2 == (unsigned char) exactn
- || *p2 == (unsigned char) endline)
+
+ /* Compare the beginning of the repeat with what in the
+ pattern follows its end. If we can establish that there
+ is nothing that they would both match, i.e., that we
+ would have to backtrack because of (as in, e.g., `a*a')
+ then we can change to pop_failure_jump, because we'll
+ never have to backtrack.
+
+ This is not true in the case of alternatives: in
+ `(a|ab)*' we do need to backtrack to the `ab' alternative
+ (e.g., if the string was `ab'). But instead of trying to
+ detect that here, the alternative has put on a dummy
+ failure point which is what we will end up popping. */
+
+ /* Skip over open/close-group commands. */
+ while (p2 + 2 < pend
+ && ((re_opcode_t) *p2 == stop_memory
+ || (re_opcode_t) *p2 == start_memory))
+ p2 += 3; /* Skip over args, too. */
+
+ /* If we're at the end of the pattern, we can change. */
+ if (p2 == pend)
{
- register int c = *p2 == (unsigned char) endline ? '\n' : p2[2];
- register unsigned char *p1 = p + mcnt;
- /* p1[0] ... p1[2] are an on_failure_jump.
- Examine what follows that. */
- if (p1[3] == (unsigned char) exactn && p1[5] != c)
- p[-3] = (unsigned char) finalize_jump;
- else if (p1[3] == (unsigned char) charset
- || p1[3] == (unsigned char) charset_not)
+ /* Consider what happens when matching ":\(.*\)"
+ against ":/". I don't really understand this code
+ yet. */
+ p[-3] = (unsigned char) pop_failure_jump;
+ DEBUG_PRINT1
+ (" End of pattern: change to `pop_failure_jump'.\n");
+ }
+
+ else if ((re_opcode_t) *p2 == exactn
+ || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
+ {
+ register unsigned char c
+ = *p2 == (unsigned char) endline ? '\n' : p2[2];
+ p1 = p + mcnt;
+
+ /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
+ to the `maybe_finalize_jump' of this case. Examine what
+ follows. */
+ if ((re_opcode_t) p1[3] == exactn && p1[5] != c)
+ {
+ p[-3] = (unsigned char) pop_failure_jump;
+ DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
+ c, p1[5]);
+ }
+
+ else if ((re_opcode_t) p1[3] == charset
+ || (re_opcode_t) p1[3] == charset_not)
{
- int not = p1[3] == (unsigned char) charset_not;
- if (c < p1[4] * BYTEWIDTH
+ int not = (re_opcode_t) p1[3] == charset_not;
+
+ if (c < (unsigned char) (p1[4] * BYTEWIDTH)
&& p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
not = !not;
- /* `not' is 1 if c would match. */
- /* That means it is not safe to finalize. */
+
+ /* `not' is equal to 1 if c would match, which means
+ that we can't change to pop_failure_jump. */
if (!not)
- p[-3] = (unsigned char) finalize_jump;
+ {
+ p[-3] = (unsigned char) pop_failure_jump;
+ DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
+ }
}
}
}
p -= 2; /* Point at relative address again. */
- if (p[-1] != (unsigned char) finalize_jump)
+ if ((re_opcode_t) p[-1] != pop_failure_jump)
{
- p[-1] = (unsigned char) jump;
- goto nofinalize;
+ p[-1] = (unsigned char) jump;
+ DEBUG_PRINT1 (" Match => jump.\n");
+ goto unconditional_jump;
}
/* Note fall through. */
- /* The end of a stupid repeat has a finalize_jump back to the
- start, where another failure point will be made which will
- point to after all the repetitions found so far. */
- /* Take off failure points put on by matching on_failure_jump
- because didn't fail. Also remove the register information
- put on by the on_failure_jump. */
- case finalize_jump:
- POP_FAILURE_POINT ();
- /* Note fall through. */
-
- /* Jump without taking off any failure points. */
+ /* The end of a simple repeat has a pop_failure_jump back to
+ its matching on_failure_jump, where the latter will push a
+ failure point. The pop_failure_jump takes off failure
+ points put on by this pop_failure_jump's matching
+ on_failure_jump; we got through the pattern to here from the
+ matching on_failure_jump, so didn't fail. */
+ case pop_failure_jump:
+ {
+ /* We need to pass separate storage for the lowest and
+ highest registers, even though we don't care about the
+ actual values. Otherwise, we will restore only one
+ register from the stack, since lowest will == highest in
+ `pop_failure_point'. */
+ unsigned dummy_low_reg, dummy_high_reg;
+ unsigned char *pdummy;
+ const char *sdummy;
+
+ DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
+ POP_FAILURE_POINT (sdummy, pdummy,
+ dummy_low_reg, dummy_high_reg,
+ reg_dummy, reg_dummy, reg_info_dummy);
+ }
+ /* Note fall through. */
+
+
+ /* Unconditionally jump (without popping any failure points). */
case jump:
- nofinalize:
- EXTRACT_NUMBER_AND_INCR (mcnt, p);
- p += mcnt;
+ unconditional_jump:
+ EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
+ DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
+ p += mcnt; /* Do the jump. */
+ DEBUG_PRINT2 ("(to 0x%x).\n", p);
break;
- case dummy_failure_jump:
- /* Normally, the on_failure_jump pushes a failure point, which
- then gets popped at finalize_jump. We will end up at
- finalize_jump, also, and with a pattern of, say, `a+', we
- are skipping over the on_failure_jump, so we have to push
- something meaningless for finalize_jump to pop. */
- PUSH_FAILURE_POINT (0, 0);
- goto nofinalize;
+
+ /* We need this opcode so we can detect where alternatives end
+ in `group_match_null_string_p' et al. */
+ case jump_past_alt:
+ DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
+ goto unconditional_jump;
- /* Have to succeed matching what follows at least n times. Then
- just handle like an on_failure_jump. */
+ /* Normally, the on_failure_jump pushes a failure point, which
+ then gets popped at pop_failure_jump. We will end up at
+ pop_failure_jump, also, and with a pattern of, say, `a+', we
+ are skipping over the on_failure_jump, so we have to push
+ something meaningless for pop_failure_jump to pop. */
+ case dummy_failure_jump:
+ DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
+ /* It doesn't matter what we push for the string here. What
+ the code at `fail' tests is the value for the pattern. */
+ PUSH_FAILURE_POINT (0, 0, -2);
+ goto unconditional_jump;
+
+
+ /* At the end of an alternative, we need to push a dummy failure
+ point in case we are followed by a `pop_failure_jump', because
+ we don't want the failure point for the alternative to be
+ popped. For example, matching `(a|ab)*' against `aab'
+ requires that we match the `ab' alternative. */
+ case push_dummy_failure:
+ DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
+ /* See comments just above at `dummy_failure_jump' about the
+ two zeroes. */
+ PUSH_FAILURE_POINT (0, 0, -2);
+ break;
+
+ /* Have to succeed matching what follows at least n times.
+ After that, handle like `on_failure_jump'. */
case succeed_n:
EXTRACT_NUMBER (mcnt, p + 2);
+ DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
+
+ assert (mcnt >= 0);
/* Originally, this is how many times we HAVE to succeed. */
- if (mcnt)
+ if (mcnt > 0)
{
mcnt--;
p += 2;
STORE_NUMBER_AND_INCR (p, mcnt);
+ DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p, mcnt);
}
else if (mcnt == 0)
{
- p[2] = unused;
- p[3] = unused;
+ DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2);
+ p[2] = (unsigned char) no_op;
+ p[3] = (unsigned char) no_op;
goto on_failure;
}
- else
- {
- fprintf (stderr, "regex: the succeed_n's n is not set.\n");
- exit (1);
- }
break;
case jump_n:
EXTRACT_NUMBER (mcnt, p + 2);
+ DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
+
/* Originally, this is how many times we CAN jump. */
if (mcnt)
{
mcnt--;
- STORE_NUMBER(p + 2, mcnt);
- goto nofinalize; /* Do the jump without taking off
- any failure points. */
+ STORE_NUMBER (p + 2, mcnt);
+ goto unconditional_jump;
}
/* If don't have to jump any more, skip over the rest of command. */
else
@@ -2359,215 +4203,493 @@ re_match_2 (pbufp, string1_arg, size1, string2_arg, size2, pos, regs, mstop)
case set_number_at:
{
- register unsigned char *p1;
+ DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
EXTRACT_NUMBER_AND_INCR (mcnt, p);
p1 = p + mcnt;
EXTRACT_NUMBER_AND_INCR (mcnt, p);
+ DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
STORE_NUMBER (p1, mcnt);
break;
}
- /* Ignore these. Used to ignore the n of succeed_n's which
- currently have n == 0. */
- case unused:
- break;
-
case wordbound:
- if (AT_WORD_BOUNDARY)
+ DEBUG_PRINT1 ("EXECUTING wordbound.\n");
+ if (AT_WORD_BOUNDARY (d))
break;
- goto fail;
+ goto fail;
case notwordbound:
- if (AT_WORD_BOUNDARY)
+ DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
+ if (AT_WORD_BOUNDARY (d))
goto fail;
- break;
+ break;
case wordbeg:
- if (IS_A_LETTER (d) && (!IS_A_LETTER (d - 1) || AT_STRINGS_BEG))
+ DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
+ if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
break;
- goto fail;
+ goto fail;
case wordend:
- /* Have to check if AT_STRINGS_BEG before looking at d - 1. */
- if (!AT_STRINGS_BEG && IS_A_LETTER (d - 1)
- && (!IS_A_LETTER (d) || AT_STRINGS_END))
+ DEBUG_PRINT1 ("EXECUTING wordend.\n");
+ if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
+ && (!WORDCHAR_P (d) || AT_STRINGS_END (d)))
break;
- goto fail;
+ goto fail;
#ifdef emacs
- case before_dot:
- if (PTR_CHAR_POS (d) >= point)
- goto fail;
- break;
-
+#ifdef emacs19
+ case before_dot:
+ DEBUG_PRINT1 ("EXECUTING before_dot.\n");
+ if (PTR_CHAR_POS ((unsigned char *) d) >= point)
+ goto fail;
+ break;
+
+ case at_dot:
+ DEBUG_PRINT1 ("EXECUTING at_dot.\n");
+ if (PTR_CHAR_POS ((unsigned char *) d) != point)
+ goto fail;
+ break;
+
+ case after_dot:
+ DEBUG_PRINT1 ("EXECUTING after_dot.\n");
+ if (PTR_CHAR_POS ((unsigned char *) d) <= point)
+ goto fail;
+ break;
+#else /* not emacs19 */
case at_dot:
- if (PTR_CHAR_POS (d) != point)
- goto fail;
- break;
-
- case after_dot:
- if (PTR_CHAR_POS (d) <= point)
+ DEBUG_PRINT1 ("EXECUTING at_dot.\n");
+ if (PTR_CHAR_POS ((unsigned char *) d) + 1 != point)
goto fail;
break;
-
- case wordchar:
- mcnt = (int) Sword;
- goto matchsyntax;
+#endif /* not emacs19 */
case syntaxspec:
+ DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
mcnt = *p++;
- matchsyntax:
- PREFETCH;
- if (SYNTAX (*d++) != (enum syntaxcode) mcnt) goto fail;
- SET_REGS_MATCHED;
- break;
-
- case notwordchar:
+ goto matchsyntax;
+
+ case wordchar:
+ DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
mcnt = (int) Sword;
- goto matchnotsyntax;
+ matchsyntax:
+ PREFETCH ();
+ if (SYNTAX (*d++) != (enum syntaxcode) mcnt)
+ goto fail;
+ SET_REGS_MATCHED ();
+ break;
case notsyntaxspec:
+ DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
mcnt = *p++;
- matchnotsyntax:
- PREFETCH;
- if (SYNTAX (*d++) == (enum syntaxcode) mcnt) goto fail;
- SET_REGS_MATCHED;
+ goto matchnotsyntax;
+
+ case notwordchar:
+ DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
+ mcnt = (int) Sword;
+ matchnotsyntax:
+ PREFETCH ();
+ if (SYNTAX (*d++) == (enum syntaxcode) mcnt)
+ goto fail;
+ SET_REGS_MATCHED ();
break;
#else /* not emacs */
-
case wordchar:
- PREFETCH;
- if (!IS_A_LETTER (d))
+ DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
+ PREFETCH ();
+ if (!WORDCHAR_P (d))
goto fail;
- SET_REGS_MATCHED;
+ SET_REGS_MATCHED ();
+ d++;
break;
case notwordchar:
- PREFETCH;
- if (IS_A_LETTER (d))
+ DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
+ PREFETCH ();
+ if (WORDCHAR_P (d))
goto fail;
- SET_REGS_MATCHED;
+ SET_REGS_MATCHED ();
+ d++;
break;
-
#endif /* not emacs */
-
- case begbuf:
- if (AT_STRINGS_BEG)
- break;
- goto fail;
-
- case endbuf:
- if (AT_STRINGS_END)
- break;
- goto fail;
-
- case exactn:
- /* Match the next few pattern characters exactly.
- mcnt is how many characters to match. */
- mcnt = *p++;
- /* This is written out as an if-else so we don't waste time
- testing `translate' inside the loop. */
- if (translate)
- {
- do
- {
- PREFETCH;
- if (translate[*d++] != *p++) goto fail;
- }
- while (--mcnt);
- }
- else
- {
- do
- {
- PREFETCH;
- if (*d++ != *p++) goto fail;
- }
- while (--mcnt);
- }
- SET_REGS_MATCHED;
- break;
+
+ default:
+ abort ();
}
continue; /* Successfully executed one pattern command; keep going. */
- /* Jump here if any matching operation fails. */
+
+ /* We goto here if a matching operation fails. */
fail:
- if (stackp != stackb)
- /* A restart point is known. Restart there and pop it. */
- {
- short last_used_reg, this_reg;
-
- /* If this failure point is from a dummy_failure_point, just
- skip it. */
- if (!stackp[-2])
+ if (!FAIL_STACK_EMPTY ())
+ { /* A restart point is known. Restore to that state. */
+ DEBUG_PRINT1 ("\nFAIL:\n");
+ POP_FAILURE_POINT (d, p,
+ lowest_active_reg, highest_active_reg,
+ regstart, regend, reg_info);
+
+ /* If this failure point is a dummy, try the next one. */
+ if (!p)
+ goto fail;
+
+ /* If we failed to the end of the pattern, don't examine *p. */
+ assert (p <= pend);
+ if (p < pend)
{
- POP_FAILURE_POINT ();
- goto fail;
+ boolean is_a_jump_n = false;
+
+ /* If failed to a backwards jump that's part of a repetition
+ loop, need to pop this failure point and use the next one. */
+ switch ((re_opcode_t) *p)
+ {
+ case jump_n:
+ is_a_jump_n = true;
+ case maybe_pop_jump:
+ case pop_failure_jump:
+ case jump:
+ p1 = p + 1;
+ EXTRACT_NUMBER_AND_INCR (mcnt, p1);
+ p1 += mcnt;
+
+ if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
+ || (!is_a_jump_n
+ && (re_opcode_t) *p1 == on_failure_jump))
+ goto fail;
+ break;
+ default:
+ /* do nothing */ ;
+ }
}
- d = *--stackp;
- p = *--stackp;
if (d >= string1 && d <= end1)
dend = end_match_1;
- /* Restore register info. */
- last_used_reg = (short) *--stackp;
-
- /* Make the ones that weren't saved -1 or 0 again. */
- for (this_reg = RE_NREGS - 1; this_reg > last_used_reg; this_reg--)
- {
- regend[this_reg] = (unsigned char *) -1;
- regstart[this_reg] = (unsigned char *) -1;
- IS_ACTIVE (reg_info[this_reg]) = 0;
- MATCHED_SOMETHING (reg_info[this_reg]) = 0;
- }
-
- /* And restore the rest from the stack. */
- for ( ; this_reg > 0; this_reg--)
- {
- reg_info[this_reg] = *(struct register_info *) *--stackp;
- regend[this_reg] = *--stackp;
- regstart[this_reg] = *--stackp;
- }
- }
+ }
else
break; /* Matching at this starting point really fails. */
- }
+ } /* for (;;) */
if (best_regs_set)
goto restore_best_regs;
+
+ FREE_VARIABLES ();
+
return -1; /* Failure to match. */
-}
+} /* re_match_2 */
+
+/* Subroutine definitions for re_match_2. */
+
+
+/* We are passed P pointing to a register number after a start_memory.
+
+ Return true if the pattern up to the corresponding stop_memory can
+ match the empty string, and false otherwise.
+
+ If we find the matching stop_memory, sets P to point to one past its number.
+ Otherwise, sets P to an undefined byte less than or equal to END.
+
+ We don't handle duplicates properly (yet). */
+
+static boolean
+group_match_null_string_p (p, end, reg_info)
+ unsigned char **p, *end;
+ register_info_type *reg_info;
+{
+ int mcnt;
+ /* Point to after the args to the start_memory. */
+ unsigned char *p1 = *p + 2;
+
+ while (p1 < end)
+ {
+ /* Skip over opcodes that can match nothing, and return true or
+ false, as appropriate, when we get to one that can't, or to the
+ matching stop_memory. */
+
+ switch ((re_opcode_t) *p1)
+ {
+ /* Could be either a loop or a series of alternatives. */
+ case on_failure_jump:
+ p1++;
+ EXTRACT_NUMBER_AND_INCR (mcnt, p1);
+
+ /* If the next operation is not a jump backwards in the
+ pattern. */
+
+ if (mcnt >= 0)
+ {
+ /* Go through the on_failure_jumps of the alternatives,
+ seeing if any of the alternatives cannot match nothing.
+ The last alternative starts with only a jump,
+ whereas the rest start with on_failure_jump and end
+ with a jump, e.g., here is the pattern for `a|b|c':
+
+ /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
+ /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
+ /exactn/1/c
+
+ So, we have to first go through the first (n-1)
+ alternatives and then deal with the last one separately. */
+
+
+ /* Deal with the first (n-1) alternatives, which start
+ with an on_failure_jump (see above) that jumps to right
+ past a jump_past_alt. */
+
+ while ((re_opcode_t) p1[mcnt-3] == jump_past_alt)
+ {
+ /* `mcnt' holds how many bytes long the alternative
+ is, including the ending `jump_past_alt' and
+ its number. */
+
+ if (!alt_match_null_string_p (p1, p1 + mcnt - 3,
+ reg_info))
+ return false;
+
+ /* Move to right after this alternative, including the
+ jump_past_alt. */
+ p1 += mcnt;
+
+ /* Break if it's the beginning of an n-th alternative
+ that doesn't begin with an on_failure_jump. */
+ if ((re_opcode_t) *p1 != on_failure_jump)
+ break;
+
+ /* Still have to check that it's not an n-th
+ alternative that starts with an on_failure_jump. */
+ p1++;
+ EXTRACT_NUMBER_AND_INCR (mcnt, p1);
+ if ((re_opcode_t) p1[mcnt-3] != jump_past_alt)
+ {
+ /* Get to the beginning of the n-th alternative. */
+ p1 -= 3;
+ break;
+ }
+ }
+
+ /* Deal with the last alternative: go back and get number
+ of the `jump_past_alt' just before it. `mcnt' contains
+ the length of the alternative. */
+ EXTRACT_NUMBER (mcnt, p1 - 2);
+
+ if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info))
+ return false;
+
+ p1 += mcnt; /* Get past the n-th alternative. */
+ } /* if mcnt > 0 */
+ break;
+
+
+ case stop_memory:
+ assert (p1[1] == **p);
+ *p = p1 + 2;
+ return true;
+
+
+ default:
+ if (!common_op_match_null_string_p (&p1, end, reg_info))
+ return false;
+ }
+ } /* while p1 < end */
+
+ return false;
+} /* group_match_null_string_p */
+
+
+/* Similar to group_match_null_string_p, but doesn't deal with alternatives:
+ It expects P to be the first byte of a single alternative and END one
+ byte past the last. The alternative can contain groups. */
+
+static boolean
+alt_match_null_string_p (p, end, reg_info)
+ unsigned char *p, *end;
+ register_info_type *reg_info;
+{
+ int mcnt;
+ unsigned char *p1 = p;
+
+ while (p1 < end)
+ {
+ /* Skip over opcodes that can match nothing, and break when we get
+ to one that can't. */
+
+ switch ((re_opcode_t) *p1)
+ {
+ /* It's a loop. */
+ case on_failure_jump:
+ p1++;
+ EXTRACT_NUMBER_AND_INCR (mcnt, p1);
+ p1 += mcnt;
+ break;
+
+ default:
+ if (!common_op_match_null_string_p (&p1, end, reg_info))
+ return false;
+ }
+ } /* while p1 < end */
+
+ return true;
+} /* alt_match_null_string_p */
+/* Deals with the ops common to group_match_null_string_p and
+ alt_match_null_string_p.
+
+ Sets P to one after the op and its arguments, if any. */
+
+static boolean
+common_op_match_null_string_p (p, end, reg_info)
+ unsigned char **p, *end;
+ register_info_type *reg_info;
+{
+ int mcnt;
+ boolean ret;
+ int reg_no;
+ unsigned char *p1 = *p;
+
+ switch ((re_opcode_t) *p1++)
+ {
+ case no_op:
+ case begline:
+ case endline:
+ case begbuf:
+ case endbuf:
+ case wordbeg:
+ case wordend:
+ case wordbound:
+ case notwordbound:
+#ifdef emacs
+ case before_dot:
+ case at_dot:
+ case after_dot:
+#endif
+ break;
+
+ case start_memory:
+ reg_no = *p1;
+ assert (reg_no > 0 && reg_no <= MAX_REGNUM);
+ ret = group_match_null_string_p (&p1, end, reg_info);
+
+ /* Have to set this here in case we're checking a group which
+ contains a group and a back reference to it. */
+
+ if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
+ REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
+
+ if (!ret)
+ return false;
+ break;
+
+ /* If this is an optimized succeed_n for zero times, make the jump. */
+ case jump:
+ EXTRACT_NUMBER_AND_INCR (mcnt, p1);
+ if (mcnt >= 0)
+ p1 += mcnt;
+ else
+ return false;
+ break;
+
+ case succeed_n:
+ /* Get to the number of times to succeed. */
+ p1 += 2;
+ EXTRACT_NUMBER_AND_INCR (mcnt, p1);
+
+ if (mcnt == 0)
+ {
+ p1 -= 4;
+ EXTRACT_NUMBER_AND_INCR (mcnt, p1);
+ p1 += mcnt;
+ }
+ else
+ return false;
+ break;
+
+ case duplicate:
+ if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
+ return false;
+ break;
+
+ case set_number_at:
+ p1 += 4;
+
+ default:
+ /* All other opcodes mean we cannot match the empty string. */
+ return false;
+ }
+
+ *p = p1;
+ return true;
+} /* common_op_match_null_string_p */
+
+
+/* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
+ bytes; nonzero otherwise. */
+
static int
bcmp_translate (s1, s2, len, translate)
unsigned char *s1, *s2;
register int len;
- unsigned char *translate;
+ char *translate;
{
register unsigned char *p1 = s1, *p2 = s2;
while (len)
{
- if (translate [*p1++] != translate [*p2++]) return 1;
+ if (translate[*p1++] != translate[*p2++]) return 1;
len--;
}
return 0;
}
+
+/* Entry points for GNU code. */
+/* re_compile_pattern is the GNU regular expression compiler: it
+ compiles PATTERN (of length SIZE) and puts the result in BUFP.
+ Returns 0 if the pattern was valid, otherwise an error string.
+
+ Assumes the `allocated' (and perhaps `buffer') and `translate' fields
+ are set in BUFP on entry.
+
+ We call regex_compile to do the actual compilation. */
+const char *
+re_compile_pattern (pattern, length, bufp)
+ const char *pattern;
+ int length;
+ struct re_pattern_buffer *bufp;
+{
+ reg_errcode_t ret;
+
+ /* GNU code is written to assume at least RE_NREGS registers will be set
+ (and at least one extra will be -1). */
+ bufp->regs_allocated = REGS_UNALLOCATED;
+
+ /* And GNU code determines whether or not to get register information
+ by passing null for the REGS argument to re_match, etc., not by
+ setting no_sub. */
+ bufp->no_sub = 0;
+
+ /* Match anchors at newline. */
+ bufp->newline_anchor = 1;
+
+ ret = regex_compile (pattern, length, re_syntax_options, bufp);
+
+ return re_error_msg[(int) ret];
+}
-/* Entry points compatible with 4.2 BSD regex library. */
+/* Entry points compatible with 4.2 BSD regex library. We don't define
+ them if this is an Emacs or POSIX compilation. */
-#ifndef emacs
+#if !defined (emacs) && !defined (_POSIX_SOURCE)
+/* BSD has one and only one pattern buffer. */
static struct re_pattern_buffer re_comp_buf;
char *
re_comp (s)
- char *s;
+ const char *s;
{
+ reg_errcode_t ret;
+
if (!s)
{
if (!re_comp_buf.buffer)
@@ -2577,194 +4699,289 @@ re_comp (s)
if (!re_comp_buf.buffer)
{
- if (!(re_comp_buf.buffer = (char *) malloc (200)))
- return "Memory exhausted";
+ re_comp_buf.buffer = (unsigned char *) malloc (200);
+ if (re_comp_buf.buffer == NULL)
+ return "Memory exhausted";
re_comp_buf.allocated = 200;
- if (!(re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH)))
+
+ re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
+ if (re_comp_buf.fastmap == NULL)
return "Memory exhausted";
}
- return re_compile_pattern (s, strlen (s), &re_comp_buf);
+
+ /* Since `re_exec' always passes NULL for the `regs' argument, we
+ don't need to initialize the pattern buffer fields which affect it. */
+
+ /* Match anchors at newlines. */
+ re_comp_buf.newline_anchor = 1;
+
+ ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
+
+ /* Yes, we're discarding `const' here. */
+ return (char *) re_error_msg[(int) ret];
}
+
int
re_exec (s)
- char *s;
+ const char *s;
{
- int len = strlen (s);
- return 0 <= re_search (&re_comp_buf, s, len, 0, len,
- (struct re_registers *) 0);
+ const int len = strlen (s);
+ return
+ 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
}
-#endif /* not emacs */
+#endif /* not emacs and not _POSIX_SOURCE */
+
+/* POSIX.2 functions. Don't define these for Emacs. */
+#ifndef emacs
-
-#ifdef test
+/* regcomp takes a regular expression as a string and compiles it.
-#include <stdio.h>
+ PREG is a regex_t *. We do not expect any fields to be initialized,
+ since POSIX says we shouldn't. Thus, we set
-/* Indexed by a character, gives the upper case equivalent of the
- character. */
-
-char upcase[0400] =
- { 000, 001, 002, 003, 004, 005, 006, 007,
- 010, 011, 012, 013, 014, 015, 016, 017,
- 020, 021, 022, 023, 024, 025, 026, 027,
- 030, 031, 032, 033, 034, 035, 036, 037,
- 040, 041, 042, 043, 044, 045, 046, 047,
- 050, 051, 052, 053, 054, 055, 056, 057,
- 060, 061, 062, 063, 064, 065, 066, 067,
- 070, 071, 072, 073, 074, 075, 076, 077,
- 0100, 0101, 0102, 0103, 0104, 0105, 0106, 0107,
- 0110, 0111, 0112, 0113, 0114, 0115, 0116, 0117,
- 0120, 0121, 0122, 0123, 0124, 0125, 0126, 0127,
- 0130, 0131, 0132, 0133, 0134, 0135, 0136, 0137,
- 0140, 0101, 0102, 0103, 0104, 0105, 0106, 0107,
- 0110, 0111, 0112, 0113, 0114, 0115, 0116, 0117,
- 0120, 0121, 0122, 0123, 0124, 0125, 0126, 0127,
- 0130, 0131, 0132, 0173, 0174, 0175, 0176, 0177,
- 0200, 0201, 0202, 0203, 0204, 0205, 0206, 0207,
- 0210, 0211, 0212, 0213, 0214, 0215, 0216, 0217,
- 0220, 0221, 0222, 0223, 0224, 0225, 0226, 0227,
- 0230, 0231, 0232, 0233, 0234, 0235, 0236, 0237,
- 0240, 0241, 0242, 0243, 0244, 0245, 0246, 0247,
- 0250, 0251, 0252, 0253, 0254, 0255, 0256, 0257,
- 0260, 0261, 0262, 0263, 0264, 0265, 0266, 0267,
- 0270, 0271, 0272, 0273, 0274, 0275, 0276, 0277,
- 0300, 0301, 0302, 0303, 0304, 0305, 0306, 0307,
- 0310, 0311, 0312, 0313, 0314, 0315, 0316, 0317,
- 0320, 0321, 0322, 0323, 0324, 0325, 0326, 0327,
- 0330, 0331, 0332, 0333, 0334, 0335, 0336, 0337,
- 0340, 0341, 0342, 0343, 0344, 0345, 0346, 0347,
- 0350, 0351, 0352, 0353, 0354, 0355, 0356, 0357,
- 0360, 0361, 0362, 0363, 0364, 0365, 0366, 0367,
- 0370, 0371, 0372, 0373, 0374, 0375, 0376, 0377
- };
+ `buffer' to the compiled pattern;
+ `used' to the length of the compiled pattern;
+ `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
+ REG_EXTENDED bit in CFLAGS is set; otherwise, to
+ RE_SYNTAX_POSIX_BASIC;
+ `newline_anchor' to REG_NEWLINE being set in CFLAGS;
+ `fastmap' and `fastmap_accurate' to zero;
+ `re_nsub' to the number of subexpressions in PATTERN.
-#ifdef canned
+ PATTERN is the address of the pattern string.
-#include "tests.h"
+ CFLAGS is a series of bits which affect compilation.
-typedef enum { extended_test, basic_test } test_type;
+ If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
+ use POSIX basic syntax.
-/* Use this to run the tests we've thought of. */
+ If REG_NEWLINE is set, then . and [^...] don't match newline.
+ Also, regexec will try a match beginning after every newline.
-void
-main ()
-{
- test_type t = extended_test;
+ 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.
- if (t == basic_test)
+ It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
+ the return codes and their meanings.) */
+
+int
+regcomp (preg, pattern, cflags)
+ regex_t *preg;
+ const char *pattern;
+ int cflags;
+{
+ reg_errcode_t ret;
+ unsigned syntax
+ = (cflags & REG_EXTENDED) ?
+ RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
+
+ /* regex_compile will allocate the space for the compiled pattern. */
+ preg->buffer = 0;
+ preg->allocated = 0;
+ preg->used = 0;
+
+ /* Don't bother to use a fastmap when searching. This simplifies the
+ REG_NEWLINE case: if we used a fastmap, we'd have to put all the
+ characters after newlines into the fastmap. This way, we just try
+ every character. */
+ preg->fastmap = 0;
+
+ if (cflags & REG_ICASE)
{
- printf ("Running basic tests:\n\n");
- test_posix_basic ();
+ unsigned i;
+
+ preg->translate = (char *) malloc (CHAR_SET_SIZE);
+ if (preg->translate == NULL)
+ 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;
}
- else if (t == extended_test)
- {
- printf ("Running extended tests:\n\n");
- test_posix_extended ();
+ else
+ preg->translate = NULL;
+
+ /* If REG_NEWLINE is set, newlines are treated differently. */
+ if (cflags & REG_NEWLINE)
+ { /* 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;
}
+ else
+ preg->newline_anchor = 0;
+
+ 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 = regex_compile (pattern, strlen (pattern), syntax, preg);
+
+ /* 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;
+
+ return (int) ret;
}
-#else /* not canned */
-/* Use this to run interactive tests. */
+/* 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. */
-void
-main (argc, argv)
- int argc;
- char **argv;
+int
+regexec (preg, string, nmatch, pmatch, eflags)
+ const regex_t *preg;
+ const char *string;
+ size_t nmatch;
+ regmatch_t pmatch[];
+ int eflags;
{
- char pat[80];
- struct re_pattern_buffer buf;
- int i;
- char c;
- char fastmap[(1 << BYTEWIDTH)];
-
- /* Allow a command argument to specify the style of syntax. */
- if (argc > 1)
- obscure_syntax = atoi (argv[1]);
-
- buf.allocated = 40;
- buf.buffer = (char *) malloc (buf.allocated);
- buf.fastmap = fastmap;
- buf.translate = upcase;
-
- while (1)
+ int ret;
+ struct re_registers regs;
+ regex_t private_preg;
+ int len = strlen (string);
+ boolean want_reg_info = !preg->no_sub && nmatch > 0;
+
+ private_preg = *preg;
+
+ private_preg.not_bol = !!(eflags & REG_NOTBOL);
+ private_preg.not_eol = !!(eflags & REG_NOTEOL);
+
+ /* The user has told us exactly how many registers to return
+ information about, via `nmatch'. We have to pass that on to the
+ matching routines. */
+ private_preg.regs_allocated = REGS_FIXED;
+
+ if (want_reg_info)
{
- gets (pat);
+ regs.num_regs = nmatch;
+ regs.start = TALLOC (nmatch, regoff_t);
+ regs.end = TALLOC (nmatch, regoff_t);
+ if (regs.start == NULL || regs.end == NULL)
+ return (int) REG_NOMATCH;
+ }
- if (*pat)
- {
- re_compile_pattern (pat, strlen(pat), &buf);
+ /* Perform the searching operation. */
+ ret = re_search (&private_preg, string, len,
+ /* start: */ 0, /* range: */ len,
+ want_reg_info ? &regs : (struct re_registers *) 0);
+
+ /* Copy the register information to the POSIX structure. */
+ if (want_reg_info)
+ {
+ if (ret >= 0)
+ {
+ unsigned r;
- for (i = 0; i < buf.used; i++)
- printchar (buf.buffer[i]);
+ for (r = 0; r < nmatch; r++)
+ {
+ pmatch[r].rm_so = regs.start[r];
+ pmatch[r].rm_eo = regs.end[r];
+ }
+ }
- putchar ('\n');
+ /* If we needed the temporary register info, free the space now. */
+ free (regs.start);
+ free (regs.end);
+ }
- printf ("%d allocated, %d used.\n", buf.allocated, buf.used);
+ /* We want zero return to mean success, unlike `re_search'. */
+ return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
+}
- re_compile_fastmap (&buf);
- printf ("Allowed by fastmap: ");
- for (i = 0; i < (1 << BYTEWIDTH); i++)
- if (fastmap[i]) printchar (i);
- putchar ('\n');
- }
- gets (pat); /* Now read the string to match against */
+/* Returns a message corresponding to an error code, ERRCODE, returned
+ from either regcomp or regexec. We don't use PREG here. */
- i = re_match (&buf, pat, strlen (pat), 0, 0);
- printf ("Match value %d.\n", i);
- }
-}
+size_t
+regerror (errcode, preg, errbuf, errbuf_size)
+ int errcode;
+ const regex_t *preg;
+ char *errbuf;
+ size_t errbuf_size;
+{
+ const char *msg;
+ size_t msg_size;
-#endif
+ if (errcode < 0
+ || errcode >= (sizeof (re_error_msg) / sizeof (re_error_msg[0])))
+ /* Only error codes returned by the rest of the code should be passed
+ to this routine. If we are given anything else, or if other regex
+ code generates an invalid error code, then the program has a bug.
+ Dump core so we can fix it. */
+ abort ();
+ msg = re_error_msg[errcode];
-#ifdef NOTDEF
-print_buf (bufp)
- struct re_pattern_buffer *bufp;
-{
- int i;
+ /* POSIX doesn't require that we do anything in this case, but why
+ not be nice. */
+ if (! msg)
+ msg = "Success";
- printf ("buf is :\n----------------\n");
- for (i = 0; i < bufp->used; i++)
- printchar (bufp->buffer[i]);
-
- printf ("\n%d allocated, %d used.\n", bufp->allocated, bufp->used);
+ msg_size = strlen (msg) + 1; /* Includes the null. */
- printf ("Allowed by fastmap: ");
- for (i = 0; i < (1 << BYTEWIDTH); i++)
- if (bufp->fastmap[i])
- printchar (i);
- printf ("\nAllowed by translate: ");
- if (bufp->translate)
- for (i = 0; i < (1 << BYTEWIDTH); i++)
- if (bufp->translate[i])
- printchar (i);
- printf ("\nfastmap is%s accurate\n", bufp->fastmap_accurate ? "" : "n't");
- printf ("can %s be null\n----------", bufp->can_be_null ? "" : "not");
-}
-#endif /* NOTDEF */
-
-printchar (c)
- char c;
-{
- if (c < 040 || c >= 0177)
+ if (errbuf_size != 0)
{
- putchar ('\\');
- putchar (((c >> 6) & 3) + '0');
- putchar (((c >> 3) & 7) + '0');
- putchar ((c & 7) + '0');
+ if (msg_size > errbuf_size)
+ {
+ strncpy (errbuf, msg, errbuf_size - 1);
+ errbuf[errbuf_size - 1] = 0;
+ }
+ else
+ strcpy (errbuf, msg);
}
- else
- putchar (c);
+
+ return msg_size;
}
-error (string)
- char *string;
+
+/* Free dynamically allocated space used by PREG. */
+
+void
+regfree (preg)
+ regex_t *preg;
{
- puts (string);
- exit (1);
+ if (preg->buffer != NULL)
+ free (preg->buffer);
+ preg->buffer = NULL;
+
+ preg->allocated = 0;
+ preg->used = 0;
+
+ if (preg->fastmap != NULL)
+ free (preg->fastmap);
+ preg->fastmap = NULL;
+ preg->fastmap_accurate = 0;
+
+ if (preg->translate != NULL)
+ free (preg->translate);
+ preg->translate = NULL;
}
-#endif /* test */
+
+#endif /* not emacs */
+
+/*
+Local variables:
+make-backup-files: t
+version-control: t
+trim-versions-without-asking: nil
+End:
+*/
diff --git a/gnu/usr.bin/diff/regex.h b/gnu/usr.bin/diff/regex.h
index ce79c77e928..408dd210348 100644
--- a/gnu/usr.bin/diff/regex.h
+++ b/gnu/usr.bin/diff/regex.h
@@ -1,10 +1,11 @@
-/* Definitions for data structures callers pass the regex library.
+/* Definitions for data structures and routines for the regular
+ expression library, version 0.12.
- Copyright (C) 1985, 1989-90 Free Software Foundation, Inc.
+ Copyright (C) 1985, 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 1, or (at your option)
+ 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,
@@ -16,242 +17,474 @@
along with this program; if not, write to the Free Software
Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */
+#ifndef __REGEXP_LIBRARY_H__
+#define __REGEXP_LIBRARY_H__
-#ifdef __GNUC__
- #pragma once
+/* POSIX says that <sys/types.h> must be included (by the caller) before
+ <regex.h>. */
+
+#ifdef VMS
+/* VMS doesn't have `size_t' in <sys/types.h>, even though POSIX says it
+ should be there. */
+#include <stddef.h>
#endif
-#ifndef __REGEXP_LIBRARY
-#define __REGEXP_LIBRARY
-/* Define number of parens for which we record the beginnings and ends.
- This affects how much space the `struct re_registers' type takes up. */
-#ifndef RE_NREGS
-#define RE_NREGS 10
-#endif
+/* 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. */
+typedef unsigned reg_syntax_t;
-#define BYTEWIDTH 8
+/* 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)
-/* Maximum number of duplicates an interval can allow. */
-#define RE_DUP_MAX ((1 << 15) - 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 \<digit> matches <digit>.
+ If not set, then \<digit> 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)
+
+/* This global variable defines the particular regexp syntax to use (for
+ some interfaces). When a regexp is compiled, the syntax used is
+ stored in the pattern buffer, so changing this does not affect
+ already-compiled regexps. */
+extern reg_syntax_t re_syntax_options;
+
+/* 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)
+
+/* P1003.2/D11.2, section 4.20.7.1, lines 5078ff. */
+#define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC
+
+#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)
+
+
+/* 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)
+
+
+/* If any error codes are removed, changed, or added, update the
+ `re_error_msg' table in regex.c. */
+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;
+
+/* This data structure represents a compiled pattern. Before calling
+ the pattern compiler, the fields `buffer', `allocated', `fastmap',
+ `translate', and `no_sub' can be set. After the pattern has been
+ compiled, the `re_nsub' field is available. All other fields are
+ private to the regex routines. */
+
+struct re_pattern_buffer
+{
+/* [[[begin pattern_buffer]]] */
+ /* Space that holds the compiled pattern. It is declared as
+ `unsigned char *' because its elements are
+ sometimes used as array indexes. */
+ unsigned char *buffer;
+
+ /* Number of bytes to which `buffer' points. */
+ unsigned long allocated;
+
+ /* Number of bytes actually used in `buffer'. */
+ unsigned long used;
+
+ /* Syntax setting with which the pattern was compiled. */
+ reg_syntax_t syntax;
+
+ /* Pointer to a fastmap, if any, otherwise zero. re_search uses
+ the fastmap, if there is one, to skip over impossible
+ starting points for matches. */
+ char *fastmap;
+
+ /* Either a translate table to apply to all characters before
+ comparing them, or zero for no translation. The translation
+ is applied to a pattern when it is compiled and to a string
+ when it is matched. */
+ char *translate;
+
+ /* Number of subexpressions found by the compiler. */
+ size_t re_nsub;
+
+ /* Zero if this pattern cannot match the empty string, one else.
+ Well, in truth it's used only in `re_search_2', to see
+ whether or not we should use the fastmap, so we don't set
+ this absolutely perfectly; see `re_compile_fastmap' (the
+ `duplicate' case). */
+ unsigned can_be_null : 1;
+
+ /* If REGS_UNALLOCATED, allocate space in the `regs' structure
+ for `max (RE_NREGS, re_nsub + 1)' groups.
+ If REGS_REALLOCATE, reallocate space if necessary.
+ If REGS_FIXED, use what's there. */
+#define REGS_UNALLOCATED 0
+#define REGS_REALLOCATE 1
+#define REGS_FIXED 2
+ unsigned regs_allocated : 2;
+
+ /* Set to zero when `regex_compile' compiles a pattern; set to one
+ by `re_compile_fastmap' if it updates the fastmap. */
+ unsigned fastmap_accurate : 1;
+
+ /* If set, `re_match_2' does not return information about
+ subexpressions. */
+ unsigned no_sub : 1;
+
+ /* If set, a beginning-of-line anchor doesn't match at the
+ beginning of the string. */
+ unsigned not_bol : 1;
+
+ /* Similarly for an end-of-line anchor. */
+ unsigned not_eol : 1;
+
+ /* If true, an anchor at a newline matches. */
+ unsigned newline_anchor : 1;
+
+/* [[[end pattern_buffer]]] */
+};
+
+typedef struct re_pattern_buffer regex_t;
+
+
+/* search.c (search_buffer) in Emacs needs this one opcode value. It is
+ defined both in `regex.c' and here. */
+#define RE_EXACTN_VALUE 1
+
+/* Type for byte offsets within the string. POSIX mandates this. */
+typedef int regoff_t;
-/* This defines the various regexp syntaxes. */
-extern int obscure_syntax;
+/* 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;
+};
-/* The following bits are used in the obscure_syntax variable to choose among
- alternative regexp syntaxes. */
-/* If this bit is set, plain parentheses serve as grouping, and backslash
- parentheses are needed for literal searching.
- If not set, backslash-parentheses are grouping, and plain parentheses
- are for literal searching. */
-#define RE_NO_BK_PARENS 1
+/* 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. */
+#ifndef RE_NREGS
+#define RE_NREGS 30
+#endif
-/* If this bit is set, plain | serves as the `or'-operator, and \| is a
- literal.
- If not set, \| serves as the `or'-operator, and | is a literal. */
-#define RE_NO_BK_VBAR (1 << 1)
-/* If this bit is not set, plain + or ? serves as an operator, and \+, \? are
- literals.
- If set, \+, \? are operators and plain +, ? are literals. */
-#define RE_BK_PLUS_QM (1 << 2)
+/* POSIX specification for registers. Aside from the different names than
+ `re_registers', POSIX uses an array of structures, instead of a
+ structure of arrays. */
+typedef struct
+{
+ 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;
+
+/* Declarations for routines. */
-/* If this bit is set, | binds tighter than ^ or $.
- If not set, the contrary. */
-#define RE_TIGHT_VBAR (1 << 3)
+/* To avoid duplicating every routine declaration -- once with a
+ prototype (if we are ANSI), and once without (if we aren't) -- we
+ use the following macro to declare argument types. This
+ unfortunately clutters up the declarations a bit, but I think it's
+ worth it. */
-/* If this bit is set, then treat newline as an OR operator.
- If not set, treat it as a normal character. */
-#define RE_NEWLINE_OR (1 << 4)
+#if __STDC__
-/* If this bit is set, then special characters may act as normal
- characters in some contexts. Specifically, this applies to:
- ^ -- only special at the beginning, or after ( or |;
- $ -- only special at the end, or before ) or |;
- *, +, ? -- only special when not after the beginning, (, or |.
- If this bit is not set, special characters (such as *, ^, and $)
- always have their special meaning regardless of the surrounding
- context. */
-#define RE_CONTEXT_INDEP_OPS (1 << 5)
+#define _RE_ARGS(args) args
-/* If this bit is not set, then \ before anything inside [ and ] is taken as
- a real \.
- If set, then such a \ escapes the following character. This is a
- special case for awk. */
-#define RE_AWK_CLASS_HACK (1 << 6)
+#else /* not __STDC__ */
-/* If this bit is set, then \{ and \} or { and } serve as interval operators.
- If not set, then \{ and \} and { and } are treated as literals. */
-#define RE_INTERVALS (1 << 7)
+#define _RE_ARGS(args) ()
-/* If this bit is not set, then \{ and \} serve as interval operators and
- { and } are literals.
- If set, then { and } serve as interval operators and \{ and \} are
- literals. */
-#define RE_NO_BK_CURLY_BRACES (1 << 8)
+#endif /* not __STDC__ */
-/* 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 (1 << 9)
-
-/* If this bit is set, then the dot re doesn't match a null byte.
- If not set, it does. */
-#define RE_DOT_NOT_NULL (1 << 10)
-
-/* If this bit is set, then [^...] doesn't match a newline.
- If not set, it does. */
-#define RE_HAT_NOT_NEWLINE (1 << 11)
-
-/* If this bit is set, back references are recognized.
- If not set, they aren't. */
-#define RE_NO_BK_REFS (1 << 12)
-
-/* If this bit is set, back references must refer to a preceding
- subexpression. If not set, a back reference to a nonexistent
- subexpression is treated as literal characters. */
-#define RE_NO_EMPTY_BK_REF (1 << 13)
-
-/* If this bit is set, bracket expressions can't be empty.
- If it is set, they can be empty. */
-#define RE_NO_EMPTY_BRACKETS (1 << 14)
-
-/* If this bit is set, then *, +, ? and { cannot be first in an re or
- immediately after a |, or a (. Furthermore, a | cannot be first or
- last in an re, or immediately follow another | or a (. Also, a ^
- cannot appear in a nonleading position and a $ cannot appear in a
- nontrailing position (outside of bracket expressions, that is). */
-#define RE_CONTEXTUAL_INVALID_OPS (1 << 15)
-
-/* If this bit is set, then +, ? and | aren't recognized as operators.
- If it's not, they are. */
-#define RE_LIMITED_OPS (1 << 16)
-
-/* If this bit is set, then an ending range point has to collate higher
- or equal to the starting range point.
- If it's not set, then when the ending range point collates higher
- than the starting range point, the range is just considered empty. */
-#define RE_NO_EMPTY_RANGES (1 << 17)
-
-/* If this bit is set, then a hyphen (-) can't be an ending range point.
- If it isn't, then it can. */
-#define RE_NO_HYPHEN_RANGE_END (1 << 18)
-
-
-/* Define combinations of bits for the standard possibilities. */
-#define RE_SYNTAX_POSIX_AWK (RE_NO_BK_PARENS | RE_NO_BK_VBAR \
- | RE_CONTEXT_INDEP_OPS)
-#define RE_SYNTAX_AWK (RE_NO_BK_PARENS | RE_NO_BK_VBAR \
- | RE_CONTEXT_INDEP_OPS | RE_AWK_CLASS_HACK)
-#define RE_SYNTAX_EGREP (RE_NO_BK_PARENS | RE_NO_BK_VBAR \
- | RE_CONTEXT_INDEP_OPS | RE_NEWLINE_OR)
-#define RE_SYNTAX_GREP (RE_BK_PLUS_QM | RE_NEWLINE_OR)
-#define RE_SYNTAX_EMACS 0
-#define RE_SYNTAX_POSIX_BASIC (RE_INTERVALS | RE_BK_PLUS_QM \
- | RE_CHAR_CLASSES | RE_DOT_NOT_NULL \
- | RE_HAT_NOT_NEWLINE | RE_NO_EMPTY_BK_REF \
- | RE_NO_EMPTY_BRACKETS | RE_LIMITED_OPS \
- | RE_NO_EMPTY_RANGES | RE_NO_HYPHEN_RANGE_END)
-
-#define RE_SYNTAX_POSIX_EXTENDED (RE_INTERVALS | RE_NO_BK_CURLY_BRACES \
- | RE_NO_BK_VBAR | RE_NO_BK_PARENS \
- | RE_HAT_NOT_NEWLINE | RE_CHAR_CLASSES \
- | RE_NO_EMPTY_BRACKETS | RE_CONTEXTUAL_INVALID_OPS \
- | RE_NO_BK_REFS | RE_NO_EMPTY_RANGES \
- | RE_NO_HYPHEN_RANGE_END)
-
-
-/* This data structure is used to represent a compiled pattern. */
+/* Sets the current default syntax to SYNTAX, and return the old syntax.
+ You can also simply assign to the `re_syntax_options' variable. */
+extern reg_syntax_t re_set_syntax _RE_ARGS ((reg_syntax_t syntax));
-struct re_pattern_buffer
- {
- char *buffer; /* Space holding the compiled pattern commands. */
- long allocated; /* Size of space that `buffer' points to. */
- long used; /* Length of portion of buffer actually occupied */
- char *fastmap; /* Pointer to fastmap, if any, or zero if none. */
- /* re_search uses the fastmap, if there is one,
- to skip over totally implausible characters. */
- char *translate; /* Translate table to apply to all characters before
- comparing, or zero for no translation.
- The translation is applied to a pattern when it is
- compiled and to data when it is matched. */
- char fastmap_accurate;
- /* Set to zero when a new pattern is stored,
- set to one when the fastmap is updated from it. */
- char can_be_null; /* Set to one by compiling fastmap
- if this pattern might match the null string.
- It does not necessarily match the null string
- in that case, but if this is zero, it cannot.
- 2 as value means can match null string
- but at end of range or before a character
- listed in the fastmap. */
- };
-
-
-/* search.c (search_buffer) needs this one value. It is defined both in
- regex.c and here. */
-#define RE_EXACTN_VALUE 1
+/* Compile the regular expression PATTERN, with length LENGTH
+ and syntax given by the global `re_syntax_options', into the buffer
+ BUFFER. Return NULL if successful, and an error string if not. */
+extern const char *re_compile_pattern
+ _RE_ARGS ((const char *pattern, int length,
+ struct re_pattern_buffer *buffer));
-/* Structure to store register contents data in.
+/* Compile a fastmap for the compiled pattern in BUFFER; used to
+ accelerate searches. Return 0 if successful and -2 if was an
+ internal error. */
+extern int re_compile_fastmap _RE_ARGS ((struct re_pattern_buffer *buffer));
- Pass the address of such a structure as an argument to re_match, etc.,
- if you want this information back.
- For i from 1 to RE_NREGS - 1, start[i] records the starting index in
- the string of where the ith subexpression matched, and end[i] records
- one after the ending index. start[0] and end[0] are analogous, for
- the entire pattern. */
+/* Search in the string STRING (with length LENGTH) for the pattern
+ compiled into BUFFER. Start searching at position START, for RANGE
+ characters. Return the starting position of the match, -1 for no
+ match, or -2 for an internal error. Also return register
+ information in REGS (if REGS and BUFFER->no_sub are nonzero). */
+extern int re_search
+ _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string,
+ int length, int start, int range, struct re_registers *regs));
-struct re_registers
- {
- int start[RE_NREGS];
- int end[RE_NREGS];
- };
+/* Like `re_search', but search in the concatenation of STRING1 and
+ STRING2. Also, stop searching at index START + STOP. */
+extern int re_search_2
+ _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string1,
+ int length1, const char *string2, int length2,
+ int start, int range, struct re_registers *regs, int stop));
-
-#ifdef __STDC__
-
-extern char *re_compile_pattern (char *, int, struct re_pattern_buffer *);
-/* Is this really advertised? */
-extern void re_compile_fastmap (struct re_pattern_buffer *);
-extern int re_search (struct re_pattern_buffer *, char*, int, int, int,
- struct re_registers *);
-extern int re_search_2 (struct re_pattern_buffer *, char *, int,
- char *, int, int, int,
- struct re_registers *, int);
-extern int re_match (struct re_pattern_buffer *, char *, int, int,
- struct re_registers *);
-extern int re_match_2 (struct re_pattern_buffer *, char *, int,
- char *, int, int, struct re_registers *, int);
-/* 4.2 bsd compatibility. */
-extern char *re_comp (char *);
-extern int re_exec (char *);
+/* Like `re_search', but return how many characters in STRING the regexp
+ in BUFFER matched, starting at position START. */
+extern int re_match
+ _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string,
+ int length, int start, struct re_registers *regs));
-#else /* !__STDC__ */
-extern char *re_compile_pattern ();
-/* Is this really advertised? */
-extern void re_compile_fastmap ();
-extern int re_search (), re_search_2 ();
-extern int re_match (), re_match_2 ();
+/* Relates to `re_match' as `re_search_2' relates to `re_search'. */
+extern int re_match_2
+ _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string1,
+ int length1, const char *string2, int length2,
+ int start, struct re_registers *regs, int stop));
-/* 4.2 bsd compatibility. */
-extern char *re_comp ();
-extern int re_exec ();
-#endif /* __STDC__ */
+/* Set REGS to hold NUM_REGS registers, storing them in STARTS and
+ ENDS. Subsequent matches using BUFFER and REGS will use this memory
+ for recording register information. STARTS and ENDS must be
+ allocated with malloc, and must each be at least `NUM_REGS * sizeof
+ (regoff_t)' bytes long.
+ If NUM_REGS == 0, then subsequent matches should allocate their own
+ register data.
-#ifdef SYNTAX_TABLE
-extern char *re_syntax_table;
-#endif
+ Unless this function is called, the first search or match using
+ PATTERN_BUFFER will allocate its own register data, without
+ freeing the old data. */
+extern void re_set_registers
+ _RE_ARGS ((struct re_pattern_buffer *buffer, struct re_registers *regs,
+ unsigned num_regs, regoff_t *starts, regoff_t *ends));
-#endif /* !__REGEXP_LIBRARY */
+/* 4.2 bsd compatibility. */
+extern char *re_comp _RE_ARGS ((const char *));
+extern int re_exec _RE_ARGS ((const char *));
+
+/* POSIX compatibility. */
+extern int regcomp _RE_ARGS ((regex_t *preg, const char *pattern, int cflags));
+extern int regexec
+ _RE_ARGS ((const regex_t *preg, const char *string, size_t nmatch,
+ regmatch_t pmatch[], int eflags));
+extern size_t regerror
+ _RE_ARGS ((int errcode, const regex_t *preg, char *errbuf,
+ size_t errbuf_size));
+extern void regfree _RE_ARGS ((regex_t *preg));
+
+#endif /* not __REGEXP_LIBRARY_H__ */
+
+/*
+Local variables:
+make-backup-files: t
+version-control: t
+trim-versions-without-asking: nil
+End:
+*/
diff --git a/gnu/usr.bin/diff/side.c b/gnu/usr.bin/diff/side.c
new file mode 100644
index 00000000000..a3d6d8bcd0e
--- /dev/null
+++ b/gnu/usr.bin/diff/side.c
@@ -0,0 +1,283 @@
+/* sdiff-format output routines for GNU DIFF.
+ Copyright (C) 1991, 1992, 1993 Free Software Foundation, Inc.
+
+This file is part of GNU DIFF.
+
+GNU DIFF is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY. No author or distributor
+accepts responsibility to anyone for the consequences of using it
+or for whether it serves any particular purpose or works at all,
+unless he says so in writing. Refer to the GNU DIFF General Public
+License for full details.
+
+Everyone is granted permission to copy, modify and redistribute
+GNU DIFF, but only under the conditions described in the
+GNU DIFF General Public License. A copy of this license is
+supposed to have been given to you along with GNU DIFF so you
+can know your rights and responsibilities. It should be in a
+file named COPYING. Among other things, the copyright notice
+and this notice must be preserved on all copies. */
+
+
+#include "diff.h"
+
+
+static void print_sdiff_hunk ();
+static void print_sdiff_common_lines ();
+static void print_1sdiff_line ();
+
+/* Next line number to be printed in the two input files. */
+static int next0, next1;
+
+/* Print the edit-script SCRIPT as a sdiff style output. */
+
+void
+print_sdiff_script (script)
+ struct change *script;
+{
+ begin_output ();
+
+ next0 = next1 = - files[0].prefix_lines;
+ print_script (script, find_change, print_sdiff_hunk);
+
+ print_sdiff_common_lines (files[0].valid_lines, files[1].valid_lines);
+}
+
+/* Tab from column FROM to column TO, where FROM <= TO. Yield TO. */
+
+static unsigned
+tab_from_to (from, to)
+ unsigned from, to;
+{
+ FILE *out = outfile;
+ unsigned tab;
+
+ if (! tab_expand_flag)
+ for (tab = from + TAB_WIDTH - from % TAB_WIDTH; tab <= to; tab += TAB_WIDTH)
+ {
+ putc ('\t', out);
+ from = tab;
+ }
+ while (from++ < to)
+ putc (' ', out);
+ return to;
+}
+
+/*
+ * Print the text for half an sdiff line. This means truncate to width
+ * observing tabs, and trim a trailing newline. Returns the last column
+ * written (not the number of chars).
+ */
+static unsigned
+print_half_line (line, indent, out_bound)
+ const char * const *line;
+ unsigned indent, out_bound;
+{
+ FILE *out = outfile;
+ register unsigned in_position = 0, out_position = 0;
+ register const char
+ *text_pointer = line[0],
+ *text_limit = line[1];
+
+ while (text_pointer < text_limit)
+ {
+ register unsigned char c = *text_pointer++;
+
+ switch (c)
+ {
+ case '\t':
+ {
+ unsigned spaces = TAB_WIDTH - in_position % TAB_WIDTH;
+ if (in_position == out_position)
+ {
+ unsigned tabstop = out_position + spaces;
+ if (tab_expand_flag)
+ {
+ if (out_bound < tabstop)
+ tabstop = out_bound;
+ for (; out_position < tabstop; out_position++)
+ putc (' ', out);
+ }
+ else
+ if (tabstop < out_bound)
+ {
+ out_position = tabstop;
+ putc (c, out);
+ }
+ }
+ in_position += spaces;
+ }
+ break;
+
+ case '\r':
+ {
+ putc (c, out);
+ tab_from_to (0, indent);
+ in_position = out_position = 0;
+ }
+ break;
+
+ case '\b':
+ if (in_position != 0 && --in_position < out_bound)
+ if (out_position <= in_position)
+ /* Add spaces to make up for suppressed tab past out_bound. */
+ for (; out_position < in_position; out_position++)
+ putc (' ', out);
+ else
+ {
+ out_position = in_position;
+ putc (c, out);
+ }
+ break;
+
+ case '\f':
+ case '\v':
+ if (in_position < out_bound)
+ putc (c, out);
+ break;
+
+ default:
+ {
+ register unsigned p = in_position;
+ if (textchar[c])
+ in_position++;
+ if (p < out_bound)
+ {
+ out_position = in_position;
+ putc (c, out);
+ }
+ }
+ break;
+
+ case '\n':
+ return out_position;
+ }
+ }
+
+ return out_position;
+}
+
+/*
+ * Print side by side lines with a separator in the middle.
+ * NULL parameters are taken to indicate whitespace text.
+ * Blank lines that can easily be caught are reduced to a single newline.
+ */
+
+static void
+print_1sdiff_line (left, sep, right)
+ const char * const *left;
+ int sep;
+ const char * const *right;
+{
+ FILE *out = outfile;
+ unsigned hw = sdiff_half_width, c2o = sdiff_column2_offset;
+ unsigned col = 0;
+ int put_newline = 0;
+
+ if (left)
+ {
+ if (left[1][-1] == '\n')
+ put_newline = 1;
+ col = print_half_line (left, 0, hw);
+ }
+
+ if (sep != ' ')
+ {
+ col = tab_from_to (col, (hw + c2o - 1) / 2) + 1;
+ if (sep == '|' && put_newline != (right[1][-1] == '\n'))
+ sep = put_newline ? '/' : '\\';
+ putc (sep, out);
+ }
+
+ if (right)
+ {
+ if (right[1][-1] == '\n')
+ put_newline = 1;
+ if (**right != '\n')
+ {
+ col = tab_from_to (col, c2o);
+ print_half_line (right, col, hw);
+ }
+ }
+
+ if (put_newline)
+ putc ('\n', out);
+}
+
+/* Print lines common to both files in side-by-side format. */
+static void
+print_sdiff_common_lines (limit0, limit1)
+ int limit0, limit1;
+{
+ int i0 = next0, i1 = next1;
+
+ if (! sdiff_skip_common_lines && (i0 != limit0 || i1 != limit1))
+ {
+ if (sdiff_help_sdiff)
+ fprintf (outfile, "i%d,%d\n", limit0 - i0, limit1 - i1);
+
+ if (! sdiff_left_only)
+ {
+ while (i0 != limit0 && i1 != limit1)
+ print_1sdiff_line (&files[0].linbuf[i0++], ' ', &files[1].linbuf[i1++]);
+ while (i1 != limit1)
+ print_1sdiff_line (0, ')', &files[1].linbuf[i1++]);
+ }
+ while (i0 != limit0)
+ print_1sdiff_line (&files[0].linbuf[i0++], '(', 0);
+ }
+
+ next0 = limit0;
+ next1 = limit1;
+}
+
+/* Print a hunk of an sdiff diff.
+ This is a contiguous portion of a complete edit script,
+ describing changes in consecutive lines. */
+
+static void
+print_sdiff_hunk (hunk)
+ struct change *hunk;
+{
+ int first0, last0, first1, last1, deletes, inserts;
+ register int i, j;
+
+ /* Determine range of line numbers involved in each file. */
+ analyze_hunk (hunk, &first0, &last0, &first1, &last1, &deletes, &inserts);
+ if (!deletes && !inserts)
+ return;
+
+ /* Print out lines up to this change. */
+ print_sdiff_common_lines (first0, first1);
+
+ if (sdiff_help_sdiff)
+ fprintf (outfile, "c%d,%d\n", last0 - first0 + 1, last1 - first1 + 1);
+
+ /* Print ``xxx | xxx '' lines */
+ if (inserts && deletes)
+ {
+ for (i = first0, j = first1; i <= last0 && j <= last1; ++i, ++j)
+ print_1sdiff_line (&files[0].linbuf[i], '|', &files[1].linbuf[j]);
+ deletes = i <= last0;
+ inserts = j <= last1;
+ next0 = first0 = i;
+ next1 = first1 = j;
+ }
+
+
+ /* Print `` > xxx '' lines */
+ if (inserts)
+ {
+ for (j = first1; j <= last1; ++j)
+ print_1sdiff_line (0, '>', &files[1].linbuf[j]);
+ next1 = j;
+ }
+
+ /* Print ``xxx < '' lines */
+ if (deletes)
+ {
+ for (i = first0; i <= last0; ++i)
+ print_1sdiff_line (&files[0].linbuf[i], '<', 0);
+ next0 = i;
+ }
+}
diff --git a/gnu/usr.bin/diff/system.h b/gnu/usr.bin/diff/system.h
new file mode 100644
index 00000000000..b17d39a08af
--- /dev/null
+++ b/gnu/usr.bin/diff/system.h
@@ -0,0 +1,159 @@
+/* System dependent declarations.
+ Copyright (C) 1988, 1989, 1992, 1993 Free Software Foundation, Inc.
+
+This file is part of GNU DIFF.
+
+GNU DIFF 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 DIFF 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 DIFF; see the file COPYING. If not, write to
+the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */
+
+#include <sys/types.h>
+#include <sys/stat.h>
+
+#ifndef S_ISDIR
+#define S_ISDIR(mode) (((mode) & S_IFMT) == S_IFDIR)
+#endif
+#ifndef S_ISREG
+#define S_ISREG(mode) (((mode) & S_IFMT) == S_IFREG)
+#endif
+
+#if HAVE_UNISTD_H
+#include <unistd.h>
+#endif
+
+#if HAVE_TIME_H
+#include <time.h>
+#else
+#include <sys/time.h>
+#endif
+
+#if HAVE_FCNTL_H
+#include <fcntl.h>
+#else
+#include <sys/file.h>
+#endif
+
+#if !HAVE_DUP2
+#define dup2(f,t) (close (t), fcntl (f,F_DUPFD,t))
+#endif
+
+#ifndef O_RDONLY
+#define O_RDONLY 0
+#endif
+
+#if HAVE_SYS_WAIT_H
+#ifndef _POSIX_VERSION
+/* Prevent the NeXT prototype using union wait from causing problems. */
+#define wait system_wait
+#endif
+#include <sys/wait.h>
+#ifndef _POSIX_VERSION
+#undef wait
+#endif
+#endif /* HAVE_SYS_WAIT_H */
+
+#ifndef WEXITSTATUS
+#define WEXITSTATUS(stat_val) ((unsigned)(stat_val) >> 8)
+#undef WIFEXITED /* Avoid 4.3BSD incompatibility with Posix. */
+#endif
+#ifndef WIFEXITED
+#define WIFEXITED(stat_val) (((stat_val) & 255) == 0)
+#endif
+
+#if HAVE_ST_BLKSIZE
+#define STAT_BLOCKSIZE(s) (s).st_blksize
+#else
+#define STAT_BLOCKSIZE(s) (S_ISREG ((s).st_mode) ? 8192 : 4096)
+#endif
+
+#if DIRENT || defined (_POSIX_VERSION)
+#include <dirent.h>
+#ifdef direct
+#undef direct
+#endif
+#define direct dirent
+#else /* ! (DIRENT || defined (_POSIX_VERSION)) */
+#if SYSNDIR
+#include <sys/ndir.h>
+#else
+#if SYSDIR
+#include <sys/dir.h>
+#else
+#include <ndir.h>
+#endif
+#endif
+#endif /* ! (DIRENT || defined (_POSIX_VERSION)) */
+
+#if HAVE_VFORK_H
+#include <vfork.h>
+#endif
+
+#if HAVE_STRING_H || STDC_HEADERS
+#include <string.h>
+#ifndef index
+#define index strchr
+#endif
+#ifndef rindex
+#define rindex strrchr
+#endif
+#ifndef bcopy
+#define bcopy(s,d,n) memcpy (d,s,n)
+#endif
+#ifndef bcmp
+#define bcmp(s1,s2,n) memcmp (s1,s2,n)
+#endif
+#ifndef bzero
+#define bzero(s,n) memset (s,0,n)
+#endif
+#else
+#include <strings.h>
+#endif
+#if !HAVE_MEMCHR && !STDC_HEADERS
+char *memchr ();
+#endif
+
+#if STDC_HEADERS
+#include <stdlib.h>
+#include <limits.h>
+#else
+char *getenv ();
+char *malloc ();
+char *realloc ();
+#if __STDC__ || __GNUC__
+#include "limits.h"
+#else
+#define INT_MAX 2147483647
+#define CHAR_BIT 8
+#endif
+#endif
+
+#include <errno.h>
+#if !STDC_HEADERS
+extern int errno;
+#endif
+
+#ifdef TRUE
+#undef TRUE
+#endif
+#ifdef FALSE
+#undef FALSE
+#endif
+#define TRUE 1
+#define FALSE 0
+
+#if !__STDC__
+#define volatile
+#endif
+
+#define min(a,b) ((a) <= (b) ? (a) : (b))
+#define max(a,b) ((a) >= (b) ? (a) : (b))
diff --git a/gnu/usr.bin/diff/util.c b/gnu/usr.bin/diff/util.c
index cba32536384..e72fd4d2370 100644
--- a/gnu/usr.bin/diff/util.c
+++ b/gnu/usr.bin/diff/util.c
@@ -1,11 +1,11 @@
/* Support routines for GNU DIFF.
- Copyright (C) 1988, 1989 Free Software Foundation, Inc.
+ Copyright (C) 1988, 1989, 1992 Free Software Foundation, Inc.
This file is part of GNU DIFF.
GNU DIFF 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 1, or (at your option)
+the Free Software Foundation; either version 2, or (at your option)
any later version.
GNU DIFF is distributed in the hope that it will be useful,
@@ -26,7 +26,9 @@ void
perror_with_name (text)
char *text;
{
+ int e = errno;
fprintf (stderr, "%s: ", program);
+ errno = e;
perror (text);
}
@@ -36,8 +38,10 @@ void
pfatal_with_name (text)
char *text;
{
+ int e = errno;
print_message_queue ();
fprintf (stderr, "%s: ", program);
+ errno = e;
perror (text);
exit (2);
}
@@ -59,11 +63,11 @@ error (format, arg, arg1)
/* Print an error message containing the string TEXT, then exit. */
void
-fatal (message)
- char *message;
+fatal (m)
+ char *m;
{
print_message_queue ();
- error (message, "");
+ error ("%s", m, 0);
exit (2);
}
@@ -90,7 +94,11 @@ message (format, arg1, arg2)
new->next = 0;
}
else
- printf (format, arg1, arg2);
+ {
+ if (sdiff_help_sdiff)
+ putchar (' ');
+ printf (format, arg1, arg2);
+ }
}
/* Output all the messages that were saved up by calls to `message'. */
@@ -111,23 +119,39 @@ print_message_queue ()
we fork off a `pr' and make OUTFILE a pipe to it.
`pr' then outputs to our stdout. */
+static char *current_name0;
+static char *current_name1;
+static int current_depth;
+
void
setup_output (name0, name1, depth)
char *name0, *name1;
int depth;
{
+ current_name0 = name0;
+ current_name1 = name1;
+ current_depth = depth;
+ outfile = 0;
+}
+
+void
+begin_output ()
+{
char *name;
+ if (outfile != 0)
+ return;
+
/* Construct the header of this piece of diff. */
- name = (char *) xmalloc (strlen (name0) + strlen (name1)
+ name = (char *) xmalloc (strlen (current_name0) + strlen (current_name1)
+ strlen (switch_string) + 15);
strcpy (name, "diff");
strcat (name, switch_string);
strcat (name, " ");
- strcat (name, name0);
+ strcat (name, current_name0);
strcat (name, " ");
- strcat (name, name1);
+ strcat (name, current_name1);
if (paginate_flag)
{
@@ -172,11 +196,26 @@ setup_output (name0, name1, depth)
/* If handling multiple files (because scanning a directory),
print which files the following output is about. */
- if (depth > 0)
+ if (current_depth > 0)
printf ("%s\n", name);
}
free (name);
+
+ /* A special header is needed at the beginning of context output. */
+ switch (output_style)
+ {
+ case OUTPUT_CONTEXT:
+ print_context_header (files, 0);
+ break;
+
+ case OUTPUT_UNIFIED:
+ print_context_header (files, 1);
+ break;
+
+ default:
+ break;
+ }
}
/* Call after the end of output of diffs for one file.
@@ -185,51 +224,33 @@ setup_output (name0, name1, depth)
void
finish_output ()
{
- if (outfile != stdout)
+ if (outfile != 0 && outfile != stdout)
{
fclose (outfile);
wait (0);
}
+
+ outfile = 0;
}
/* Compare two lines (typically one from each input file)
according to the command line options.
- Each line is described by a `struct line_def'.
Return 1 if the lines differ, like `bcmp'. */
int
-line_cmp (s1, s2)
- struct line_def *s1, *s2;
+line_cmp (s1, len1, s2, len2)
+ const char *s1, *s2;
+ int len1, len2;
{
- register char *t1, *t2;
- register char end_char = line_end_char;
- int savechar;
+ register const unsigned char *t1, *t2;
+ register unsigned char end_char = line_end_char;
/* Check first for exact identity.
If that is true, return 0 immediately.
This detects the common case of exact identity
faster than complete comparison would. */
- t1 = s1->text;
- t2 = s2->text;
-
- /* Alter the character following line 2 so it doesn't
- match that following line 1.
- (We used to alter the character after line 1,
- but that caused trouble if line 2 directly follows line 1.) */
- savechar = s2->text[s2->length];
- s2->text[s2->length] = s1->text[s1->length] + 1;
-
- /* Now find the first mismatch; this won't go past the
- character we just changed. */
- while (*t1++ == *t2++);
-
- /* Undo the alteration. */
- s2->text[s2->length] = savechar;
-
- /* If the comparison stopped at the alteration,
- the two lines are identical. */
- if (t2 == s2->text + s2->length + 1)
+ if (len1 == len2 && bcmp (s1, s2, len1) == 0)
return 0;
/* Not exactly identical, but perhaps they match anyway
@@ -237,21 +258,21 @@ line_cmp (s1, s2)
if (ignore_case_flag || ignore_space_change_flag || ignore_all_space_flag)
{
- t1 = s1->text;
- t2 = s2->text;
+ t1 = (const unsigned char *) s1;
+ t2 = (const unsigned char *) s2;
while (1)
{
- register char c1 = *t1++;
- register char c2 = *t2++;
+ register unsigned char c1 = *t1++;
+ register unsigned char c2 = *t2++;
/* Ignore horizontal whitespace if -b or -w is specified. */
if (ignore_all_space_flag)
{
- /* For -w, just skip past any spaces or tabs. */
- while (c1 == ' ' || c1 == '\t') c1 = *t1++;
- while (c2 == ' ' || c2 == '\t') c2 = *t2++;
+ /* For -w, just skip past any white space. */
+ while (Is_space (c1)) c1 = *t1++;
+ while (Is_space (c2)) c2 = *t2++;
}
else if (ignore_space_change_flag)
{
@@ -378,59 +399,88 @@ print_script (script, hunkfun, printfun)
void
print_1_line (line_flag, line)
- char *line_flag;
- struct line_def *line;
+ const char *line_flag;
+ const char * const *line;
{
- int length = line->length; /* must be nonzero */
- const char *text = line->text; /* Help the compiler. */
+ const char *text = line[0], *limit = line[1]; /* Help the compiler. */
FILE *out = outfile; /* Help the compiler some more. */
+ const char *flag_format = 0;
/* If -T was specified, use a Tab between the line-flag and the text.
Otherwise use a Space (as Unix diff does).
Print neither space nor tab if line-flags are empty. */
if (line_flag != NULL && line_flag[0] != 0)
- fprintf (out, tab_align_flag ? "%s\t" : "%s ", line_flag);
+ {
+ flag_format = tab_align_flag ? "%s\t" : "%s ";
+ fprintf (out, flag_format, line_flag);
+ }
- /* Now output the contents of the line.
- If -t was specified, expand tabs to spaces.
- Otherwise output verbatim. */
+ output_1_line (text, limit, flag_format, line_flag);
- if (tab_expand_flag)
+ if ((line_flag == NULL || line_flag[0] != 0) && limit[-1] != '\n'
+ && line_end_char == '\n')
+ fprintf (out, "\n\\ No newline at end of file\n");
+}
+
+/* Output a line from TEXT up to LIMIT. Without -t, output verbatim.
+ With -t, expand white space characters to spaces, and if FLAG_FORMAT
+ is nonzero, output it with argument LINE_FLAG after every
+ internal carriage return, so that tab stops continue to line up. */
+
+void
+output_1_line (text, limit, flag_format, line_flag)
+ const char *text, *limit, *flag_format, *line_flag;
+{
+ if (!tab_expand_flag)
+ fwrite (text, sizeof (char), limit - text, outfile);
+ else
{
- register int column = 0;
- register int i;
- for (i = 0; i < line->length; i++)
- {
- register char c = line->text[i];
- switch (c)
+ register FILE *out = outfile;
+ register char c;
+ register const char *t = text;
+ register unsigned column = 0;
+
+ while (t < limit)
+ switch ((c = *t++))
+ {
+ case '\t':
{
- case '\t':
- column++;
- while (column & 7)
- {
- putc (' ', out);
- column++;
- }
- c = ' ';
- break;
- case '\b':
- column--;
- break;
- default:
- column++;
- break;
+ unsigned spaces = TAB_WIDTH - column % TAB_WIDTH;
+ column += spaces;
+ do
+ putc (' ', out);
+ while (--spaces);
}
- putc (c, out);
- }
+ break;
+
+ case '\r':
+ putc (c, out);
+ if (flag_format && t < limit && *t != '\n')
+ fprintf (out, flag_format, line_flag);
+ column = 0;
+ break;
+
+ case '\b':
+ if (column == 0)
+ continue;
+ column--;
+ putc (c, out);
+ break;
+
+ default:
+ if (textchar[(unsigned char) c])
+ column++;
+ /* fall into */
+ case '\f':
+ case '\v':
+ putc (c, out);
+ break;
+ }
}
- else
- fwrite (text, sizeof (char), length, out);
- if ((line_flag == NULL || line_flag[0] != 0) && text[length - 1] != '\n'
- && line_end_char == '\n')
- fprintf (out, "\n\\ No newline at end of file\n");
}
+int
change_letter (inserts, deletes)
int inserts, deletes;
{
@@ -446,18 +496,15 @@ change_letter (inserts, deletes)
into an actual line number in the input file.
The internal line number is LNUM. FILE points to the data on the file.
- Internal line numbers count from 0 within the current chunk.
- Actual line numbers count from 1 within the entire file;
- in addition, they include lines ignored for comparison purposes.
-
- The `ltran' feature is no longer in use. */
+ Internal line numbers count from 0 starting after the prefix.
+ Actual line numbers count from 1 within the entire file. */
int
translate_line_number (file, lnum)
struct file_data *file;
int lnum;
{
- return lnum + 1;
+ return lnum + file->prefix_lines + 1;
}
void
@@ -515,7 +562,7 @@ analyze_hunk (hunk, first0, last0, first1, last1, deletes, inserts)
{
int f0, l0, f1, l1, show_from, show_to;
int i;
- int nontrivial = !(ignore_blank_lines_flag || ignore_regexp);
+ int nontrivial = !(ignore_blank_lines_flag || ignore_regexp_list);
struct change *next;
show_from = show_to = 0;
@@ -531,22 +578,36 @@ analyze_hunk (hunk, first0, last0, first1, last1, deletes, inserts)
show_to += next->inserted;
for (i = next->line0; i <= l0 && ! nontrivial; i++)
- if ((!ignore_blank_lines_flag || files[0].linbuf[i].length > 1)
- && (!ignore_regexp
- || 0 > re_search (&ignore_regexp_compiled,
- files[0].linbuf[i].text,
- files[0].linbuf[i].length, 0,
- files[0].linbuf[i].length, 0)))
- nontrivial = 1;
+ if (!ignore_blank_lines_flag || files[0].linbuf[i][0] != '\n')
+ {
+ struct regexp_list *r;
+ const char *line = files[0].linbuf[i];
+ int len = files[0].linbuf[i + 1] - line;
+
+ for (r = ignore_regexp_list; r; r = r->next)
+ if (0 <= re_search (&r->buf, line, len, 0, len, 0))
+ break; /* Found a match. Ignore this line. */
+ /* If we got all the way through the regexp list without
+ finding a match, then it's nontrivial. */
+ if (r == NULL)
+ nontrivial = 1;
+ }
for (i = next->line1; i <= l1 && ! nontrivial; i++)
- if ((!ignore_blank_lines_flag || files[1].linbuf[i].length > 1)
- && (!ignore_regexp
- || 0 > re_search (&ignore_regexp_compiled,
- files[1].linbuf[i].text,
- files[1].linbuf[i].length, 0,
- files[1].linbuf[i].length, 0)))
- nontrivial = 1;
+ if (!ignore_blank_lines_flag || files[1].linbuf[i][0] != '\n')
+ {
+ struct regexp_list *r;
+ const char *line = files[1].linbuf[i];
+ int len = files[1].linbuf[i + 1] - line;
+
+ for (r = ignore_regexp_list; r; r = r->next)
+ if (0 <= re_search (&r->buf, line, len, 0, len, 0))
+ break; /* Found a match. Ignore this line. */
+ /* If we got all the way through the regexp list without
+ finding a match, then it's nontrivial. */
+ if (r == NULL)
+ nontrivial = 1;
+ }
}
*first0 = f0;
@@ -615,6 +676,7 @@ concat (s1, s2, s3)
return new;
}
+void
debug_script (sp)
struct change *sp;
{
@@ -624,3 +686,18 @@ debug_script (sp)
sp->line0, sp->line1, sp->deleted, sp->inserted);
fflush (stderr);
}
+
+#if !HAVE_MEMCHR
+char *
+memchr (s, c, n)
+ char *s;
+ int c;
+ size_t n;
+{
+ unsigned char *p = (unsigned char *) s, *lim = p + n;
+ for (; p < lim; p++)
+ if (*p == c)
+ return (char *) p;
+ return 0;
+}
+#endif
diff --git a/gnu/usr.bin/diff/version.c b/gnu/usr.bin/diff/version.c
index 26fa21d3ab6..cb9d3b9c45c 100644
--- a/gnu/usr.bin/diff/version.c
+++ b/gnu/usr.bin/diff/version.c
@@ -1,3 +1,3 @@
/* Version number of GNU diff. */
-char *version_string = "1.15";
+char *version_string = "2.3";