/[pcre]/code/trunk/ChangeLog
ViewVC logotype

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 435 - (show annotations) (download)
Sat Sep 5 10:20:44 2009 UTC (5 years ago) by ph10
File size: 171156 byte(s)
Further updates to partial matching.

1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.00 ??-???-??
5 ----------------------
6
7 1. The table for translating pcre_compile() error codes into POSIX error codes
8 was out-of-date, and there was no check on the pcre_compile() error code
9 being within the table. This could lead to an OK return being given in
10 error.
11
12 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
13 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
14 in a Windows environment.
15
16 3. The pcregrep --count option prints the count for each file even when it is
17 zero, as does GNU grep. However, pcregrep was also printing all files when
18 --files-with-matches was added. Now, when both options are given, it prints
19 counts only for those files that have at least one match. (GNU grep just
20 prints the file name in this circumstance, but including the count seems
21 more useful - otherwise, why use --count?) Also ensured that the
22 combination -clh just lists non-zero counts, with no names.
23
24 4. The long form of the pcregrep -F option was incorrectly implemented as
25 --fixed_strings instead of --fixed-strings. This is an incompatible change,
26 but it seems right to fix it, and I didn't think it was worth preserving
27 the old behaviour.
28
29 5. The command line items --regex=pattern and --regexp=pattern were not
30 recognized by pcregrep, which required --regex pattern or --regexp pattern
31 (with a space rather than an '='). The man page documented the '=' forms,
32 which are compatible with GNU grep; these now work.
33
34 6. No libpcreposix.pc file was created for pkg-config; there was just
35 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
36
37 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
38 when UCP support is not needed, by modifying the Python script that
39 generates it from Unicode data files. This should not matter if the module
40 is correctly used as a library, but I received one complaint about 50K of
41 unwanted data. My guess is that the person linked everything into his
42 program rather than using a library. Anyway, it does no harm.
43
44 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
45 was a minimum greater than 1 for a wide character in a possessive
46 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
47 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
48 character. Chaos in the form of incorrect output or a compiling loop could
49 result.
50
51 9. The restrictions on what a pattern can contain when partial matching is
52 requested for pcre_exec() have been removed. All patterns can now be
53 partially matched by this function. In addition, if there are at least two
54 slots in the offset vector, the offset of the earliest inspected character
55 for the match and the offset of the end of the subject are set in them when
56 PCRE_ERROR_PARTIAL is returned.
57
58 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
59 synonymous with PCRE_PARTIAL, for backwards compatibility, and
60 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
61 and may be more useful for multi-segment matching, especially with
62 pcre_exec().
63
64 11. Partial matching with pcre_exec() is now more intuitive. A partial match
65 used to be given if ever the end of the subject was reached; now it is
66 given only if matching could not proceed because another character was
67 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
68 string "Z", which now yields "no match" instead of "partial match". In the
69 case of pcre_dfa_exec(), "no match" is given if every matching path for the
70 final character ended with (*FAIL).
71
72 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
73 if the pattern had a "must contain" character that was already found in the
74 earlier partial match, unless partial matching was again requested. For
75 example, with the pattern /dog.(body)?/, the "must contain" character is
76 "g". If the first part-match was for the string "dog", restarting with
77 "sbody" failed. This bug has been fixed.
78
79 13. The string returned by pcre_dfa_exec() after a partial match has been
80 changed so that it starts at the first inspected character rather than the
81 first character of the match. This makes a difference only if the pattern
82 starts with a lookbehind assertion or \b or \B (\K is not supported by
83 pcre_dfa_exec()). It's an incompatible change, but it makes the two
84 matching functions compatible, and I think it's the right thing to do.
85
86 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
87 so that the demonstration program is easily available in environments where
88 PCRE has not been installed from source.
89
90 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
91 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
92 library.
93
94 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
95 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
96 is not the first non-POSIX option to be added. Clearly some people find
97 these options useful.
98
99 17. If a caller to the POSIX matching function regexec() passes a non-zero
100 value for \fInmatch\fP with a NULL value for \fIpmatch\fP, the value of
101 \fInmatch\fP is forced to zero.
102
103
104 Version 7.9 11-Apr-09
105 ---------------------
106
107 1. When building with support for bzlib/zlib (pcregrep) and/or readline
108 (pcretest), all targets were linked against these libraries. This included
109 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
110 libraries. This caused unwanted dependencies to be created. This problem
111 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
112 pcretest is linked with readline.
113
114 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
115 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
116 moved outside it again, because FALSE and TRUE are already defined in AIX,
117 but BOOL is not.
118
119 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
120 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
121
122 4. The pcregrep documentation said spaces were inserted as well as colons (or
123 hyphens) following file names and line numbers when outputting matching
124 lines. This is not true; no spaces are inserted. I have also clarified the
125 wording for the --colour (or --color) option.
126
127 5. In pcregrep, when --colour was used with -o, the list of matching strings
128 was not coloured; this is different to GNU grep, so I have changed it to be
129 the same.
130
131 6. When --colo(u)r was used in pcregrep, only the first matching substring in
132 each matching line was coloured. Now it goes on to look for further matches
133 of any of the test patterns, which is the same behaviour as GNU grep.
134
135 7. A pattern that could match an empty string could cause pcregrep to loop; it
136 doesn't make sense to accept an empty string match in pcregrep, so I have
137 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
138 seems to be how GNU grep behaves.
139
140 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
141 start or after a newline", because the conditional assertion was not being
142 correctly handled. The rule now is that both the assertion and what follows
143 in the first alternative must satisfy the test.
144
145 9. If auto-callout was enabled in a pattern with a conditional group whose
146 condition was an assertion, PCRE could crash during matching, both with
147 pcre_exec() and pcre_dfa_exec().
148
149 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
150 used for matching.
151
152 11. Unicode property support in character classes was not working for
153 characters (bytes) greater than 127 when not in UTF-8 mode.
154
155 12. Added the -M command line option to pcretest.
156
157 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
158
159 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
160
161 16. Added comments and documentation about mis-use of no_arg in the C++
162 wrapper.
163
164 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
165 from Martin Jerabek that uses macro names for all relevant character and
166 string constants.
167
168 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
169 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
170 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
171 these, but not everybody uses configure.
172
173 19. A conditional group that had only one branch was not being correctly
174 recognized as an item that could match an empty string. This meant that an
175 enclosing group might also not be so recognized, causing infinite looping
176 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
177 with the subject "ab", where knowledge that the repeated group can match
178 nothing is needed in order to break the loop.
179
180 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
181 exec(), but without supplying a callout function, matching went wrong.
182
183 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
184 leak if the size of the offset vector was greater than 30. When the vector
185 is smaller, the saved offsets during recursion go onto a local stack
186 vector, but for larger vectors malloc() is used. It was failing to free
187 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
188 error, in fact).
189
190 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
191 heapframe that is used only when UTF-8 support is enabled. This caused no
192 problem, but was untidy.
193
194 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
195 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
196 included within another project.
197
198 24. Steven Van Ingelgem's patches to add more options to the CMake support,
199 slightly modified by me:
200
201 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
202 not building pcregrep.
203
204 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
205 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
206
207 25. Forward references, both numeric and by name, in patterns that made use of
208 duplicate group numbers, could behave incorrectly or give incorrect errors,
209 because when scanning forward to find the reference group, PCRE was not
210 taking into account the duplicate group numbers. A pattern such as
211 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
212
213 26. Changed a few more instances of "const unsigned char *" to USPTR, making
214 the feature of a custom pointer more persuasive (as requested by a user).
215
216 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
217 pcretest.c, inside #ifndefs, because it seems they are sometimes already
218 pre-defined.
219
220 28. Added support for (*UTF8) at the start of a pattern.
221
222 29. Arrange for flags added by the "release type" setting in CMake to be shown
223 in the configuration summary.
224
225
226 Version 7.8 05-Sep-08
227 ---------------------
228
229 1. Replaced UCP searching code with optimized version as implemented for Ad
230 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
231 stage table and inline lookup instead of a function, giving speed ups of 2
232 to 5 times on some simple patterns that I tested. Permission was given to
233 distribute the MultiStage2.py script that generates the tables (it's not in
234 the tarball, but is in the Subversion repository).
235
236 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
237 scripts.
238
239 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
240 a group with a zero qualifier. The result of the study could be incorrect,
241 or the function might crash, depending on the pattern.
242
243 4. Caseless matching was not working for non-ASCII characters in back
244 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
245 It now works when Unicode Property Support is available.
246
247 5. In pcretest, an escape such as \x{de} in the data was always generating
248 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
249 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
250 truncation.
251
252 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
253
254 7. Added two (int) casts to pcregrep when printing the difference of two
255 pointers, in case they are 64-bit values.
256
257 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
258 test 2 if it fails.
259
260 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
261 and a #define of that name to empty if it is not externally set. This is to
262 allow users of MSVC to set it if necessary.
263
264 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
265 the convenience functions in the pcre_get.c source file.
266
267 11. An option change at the start of a pattern that had top-level alternatives
268 could cause overwriting and/or a crash. This command provoked a crash in
269 some environments:
270
271 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
272
273 This potential security problem was recorded as CVE-2008-2371.
274
275 12. For a pattern where the match had to start at the beginning or immediately
276 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
277 pcre_dfa_exec() could read past the end of the passed subject if there was
278 no match. To help with detecting such bugs (e.g. with valgrind), I modified
279 pcretest so that it places the subject at the end of its malloc-ed buffer.
280
281 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
282 exec() might read past the end of the data buffer in UTF-8 mode.
283
284 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
285 the data contained the byte 0x85 as part of a UTF-8 character within its
286 first line. This applied both to normal and DFA matching.
287
288 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
289 /^[^d]*?$/8 failed to match "abc".
290
291 16. Added a missing copyright notice to pcrecpp_internal.h.
292
293 17. Make it more clear in the documentation that values returned from
294 pcre_exec() in ovector are byte offsets, not character counts.
295
296 18. Tidied a few places to stop certain compilers from issuing warnings.
297
298 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
299 supplied by Stefan Weber. I made a further small update for 7.8 because
300 there is a change of source arrangements: the pcre_searchfuncs.c module is
301 replaced by pcre_ucd.c.
302
303
304 Version 7.7 07-May-08
305 ---------------------
306
307 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
308 a string to a long long, pretend we don't even have a long long." This is
309 done by checking for the strtoq, strtoll, and _strtoi64 functions.
310
311 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
312 pre-7.6 versions, which defined a global no_arg variable instead of putting
313 it in the RE class. (See also #8 below.)
314
315 3. Remove a line of dead code, identified by coverity and reported by Nuno
316 Lopes.
317
318 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
319
320 (1) The include/exclude patterns were being applied to the whole pathnames
321 of files, instead of just to the final components.
322
323 (2) If there was more than one level of directory, the subdirectories were
324 skipped unless they satisfied the include/exclude conditions. This is
325 inconsistent with GNU grep (and could even be seen as contrary to the
326 pcregrep specification - which I improved to make it absolutely clear).
327 The action now is always to scan all levels of directory, and just
328 apply the include/exclude patterns to regular files.
329
330 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
331 --exclude_dir in the tests to avoid scanning .svn directories.
332
333 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
334 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
335 doesn't support NULs in patterns.
336
337 7. Added some missing "const"s to declarations of static tables in
338 pcre_compile.c and pcre_dfa_exec.c.
339
340 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
341 caused by fix #2 above. (Subsequently also a second patch to fix the
342 first patch. And a third patch - this was a messy problem.)
343
344 9. Applied Craig's patch to remove the use of push_back().
345
346 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
347 matching function regexec().
348
349 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
350 which, however, unlike Perl's \g{...}, are subroutine calls, not back
351 references. PCRE supports relative numbers with this syntax (I don't think
352 Oniguruma does).
353
354 12. Previously, a group with a zero repeat such as (...){0} was completely
355 omitted from the compiled regex. However, this means that if the group
356 was called as a subroutine from elsewhere in the pattern, things went wrong
357 (an internal error was given). Such groups are now left in the compiled
358 pattern, with a new opcode that causes them to be skipped at execution
359 time.
360
361 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
362 to the way PCRE behaves:
363
364 (a) A lone ] character is dis-allowed (Perl treats it as data).
365
366 (b) A back reference to an unmatched subpattern matches an empty string
367 (Perl fails the current match path).
368
369 (c) A data ] in a character class must be notated as \] because if the
370 first data character in a class is ], it defines an empty class. (In
371 Perl it is not possible to have an empty class.) The empty class []
372 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
373 The negative empty class [^] matches any one character, independently
374 of the DOTALL setting.
375
376 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
377 non-existent subpattern following a character class starting with ']' and
378 containing () gave an internal compiling error instead of "reference to
379 non-existent subpattern". Fortunately, when the pattern did exist, the
380 compiled code was correct. (When scanning forwards to check for the
381 existencd of the subpattern, it was treating the data ']' as terminating
382 the class, so got the count wrong. When actually compiling, the reference
383 was subsequently set up correctly.)
384
385 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
386 it was being rejected as not supported by pcre_dfa_exec(), even though
387 other assertions are supported. I have made pcre_dfa_exec() support
388 (*FAIL).
389
390 16. The implementation of 13c above involved the invention of a new opcode,
391 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
392 cannot be changed at match time, I realized I could make a small
393 improvement to matching performance by compiling OP_ALLANY instead of
394 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
395 on the OP_ANY path.
396
397 17. Compiling pcretest on Windows with readline support failed without the
398 following two fixes: (1) Make the unistd.h include conditional on
399 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
400
401 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
402 ncurses library to be included for pcretest when ReadLine support is
403 requested, but also to allow for it to be overridden. This patch came from
404 Daniel Bergström.
405
406 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
407 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
408 any errors with the current Unicode tables. Thanks to Peter Kankowski for
409 spotting this.
410
411
412 Version 7.6 28-Jan-08
413 ---------------------
414
415 1. A character class containing a very large number of characters with
416 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
417 overflow.
418
419 2. Patch to cut out the "long long" test in pcrecpp_unittest when
420 HAVE_LONG_LONG is not defined.
421
422 3. Applied Christian Ehrlicher's patch to update the CMake build files to
423 bring them up to date and include new features. This patch includes:
424
425 - Fixed PH's badly added libz and libbz2 support.
426 - Fixed a problem with static linking.
427 - Added pcredemo. [But later removed - see 7 below.]
428 - Fixed dftables problem and added an option.
429 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
430 HAVE_LONG_LONG.
431 - Added readline support for pcretest.
432 - Added an listing of the option settings after cmake has run.
433
434 4. A user submitted a patch to Makefile that makes it easy to create
435 "pcre.dll" under mingw when using Configure/Make. I added stuff to
436 Makefile.am that cause it to include this special target, without
437 affecting anything else. Note that the same mingw target plus all
438 the other distribution libraries and programs are now supported
439 when configuring with CMake (see 6 below) instead of with
440 Configure/Make.
441
442 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
443 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
444 exported in the Windows port". It has not yet been confirmed that the patch
445 solves the problem, but it does no harm.
446
447 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
448 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
449 with CMake, and also correct the comment about stack recursion.
450
451 7. Remove the automatic building of pcredemo from the ./configure system and
452 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
453 of a program that users should build themselves after PCRE is installed, so
454 building it automatically is not really right. What is more, it gave
455 trouble in some build environments.
456
457 8. Further tidies to CMakeLists.txt from Sheri and Christian.
458
459
460 Version 7.5 10-Jan-08
461 ---------------------
462
463 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
464 values in parens when parsing an RE using the C++ wrapper."
465
466 2. Negative specials like \S did not work in character classes in UTF-8 mode.
467 Characters greater than 255 were excluded from the class instead of being
468 included.
469
470 3. The same bug as (2) above applied to negated POSIX classes such as
471 [:^space:].
472
473 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
474 defined or documented. It seems to have been a typo for PCRE_STATIC, so
475 I have changed it.
476
477 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
478 first named subpattern) and a construct such as (?&a) would reference the
479 first named subpattern whose name started with "a" (in other words, the
480 length check was missing). Both these problems are fixed. "Subpattern name
481 expected" is now given for (?&) (a zero-length name), and this patch also
482 makes it give the same error for \k'' (previously it complained that that
483 was a reference to a non-existent subpattern).
484
485 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
486 this is right because (?- can be followed by option settings as well as by
487 digits. I have, however, made the messages clearer.
488
489 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
490 than the number used in the conditional) now cause a compile-time error.
491 This is actually not compatible with Perl, which accepts such patterns, but
492 treats the conditional as always being FALSE (as PCRE used to), but it
493 seems to me that giving a diagnostic is better.
494
495 8. Change "alphameric" to the more common word "alphanumeric" in comments
496 and messages.
497
498 9. Fix two occurrences of "backslash" in comments that should have been
499 "backspace".
500
501 10. Remove two redundant lines of code that can never be obeyed (their function
502 was moved elsewhere).
503
504 11. The program that makes PCRE's Unicode character property table had a bug
505 which caused it to generate incorrect table entries for sequences of
506 characters that have the same character type, but are in different scripts.
507 It amalgamated them into a single range, with the script of the first of
508 them. In other words, some characters were in the wrong script. There were
509 thirteen such cases, affecting characters in the following ranges:
510
511 U+002b0 - U+002c1
512 U+0060c - U+0060d
513 U+0061e - U+00612
514 U+0064b - U+0065e
515 U+0074d - U+0076d
516 U+01800 - U+01805
517 U+01d00 - U+01d77
518 U+01d9b - U+01dbf
519 U+0200b - U+0200f
520 U+030fc - U+030fe
521 U+03260 - U+0327f
522 U+0fb46 - U+0fbb1
523 U+10450 - U+1049d
524
525 12. The -o option (show only the matching part of a line) for pcregrep was not
526 compatible with GNU grep in that, if there was more than one match in a
527 line, it showed only the first of them. It now behaves in the same way as
528 GNU grep.
529
530 13. If the -o and -v options were combined for pcregrep, it printed a blank
531 line for every non-matching line. GNU grep prints nothing, and pcregrep now
532 does the same. The return code can be used to tell if there were any
533 non-matching lines.
534
535 14. Added --file-offsets and --line-offsets to pcregrep.
536
537 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
538 infinitely looping recursion. The bug was that positive lookaheads were not
539 being skipped when checking for a possible empty match (negative lookaheads
540 and both kinds of lookbehind were skipped).
541
542 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
543 inclusion of <windows.h> to before rather than after the definition of
544 INVALID_FILE_ATTRIBUTES (patch from David Byron).
545
546 17. Specifying a possessive quantifier with a specific limit for a Unicode
547 character property caused pcre_compile() to compile bad code, which led at
548 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
549 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
550 caused the error; without that there was no problem.
551
552 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
553
554 19. Added --enable-pcretest-libreadline.
555
556 20. In pcrecpp.cc, the variable 'count' was incremented twice in
557 RE::GlobalReplace(). As a result, the number of replacements returned was
558 double what it should be. I removed one of the increments, but Craig sent a
559 later patch that removed the other one (the right fix) and added unit tests
560 that check the return values (which was not done before).
561
562 21. Several CMake things:
563
564 (1) Arranged that, when cmake is used on Unix, the libraries end up with
565 the names libpcre and libpcreposix, not just pcre and pcreposix.
566
567 (2) The above change means that pcretest and pcregrep are now correctly
568 linked with the newly-built libraries, not previously installed ones.
569
570 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
571
572 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
573 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
574 UTF-8 newline character). The key issue is that the pattern starts .*;
575 this means that the match must be either at the beginning, or after a
576 newline. The bug was in the code for advancing after a failed match and
577 checking that the new position followed a newline. It was not taking
578 account of UTF-8 characters correctly.
579
580 23. PCRE was behaving differently from Perl in the way it recognized POSIX
581 character classes. PCRE was not treating the sequence [:...:] as a
582 character class unless the ... were all letters. Perl, however, seems to
583 allow any characters between [: and :], though of course it rejects as
584 unknown any "names" that contain non-letters, because all the known class
585 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
586 for example, whereas PCRE did not - it did not recognize a POSIX character
587 class. This seemed a bit dangerous, so the code has been changed to be
588 closer to Perl. The behaviour is not identical to Perl, because PCRE will
589 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
590 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
591 Perl does, and where it didn't before.
592
593 24. Rewrite so as to remove the single use of %n from pcregrep because in some
594 Windows environments %n is disabled by default.
595
596
597 Version 7.4 21-Sep-07
598 ---------------------
599
600 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
601 means that a class such as [\s] counted as "explicit reference to CR or
602 LF". That isn't really right - the whole point of the change was to try to
603 help when there was an actual mention of one of the two characters. So now
604 the change happens only if \r or \n (or a literal CR or LF) character is
605 encountered.
606
607 2. The 32-bit options word was also used for 6 internal flags, but the numbers
608 of both had grown to the point where there were only 3 bits left.
609 Fortunately, there was spare space in the data structure, and so I have
610 moved the internal flags into a new 16-bit field to free up more option
611 bits.
612
613 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
614 but did not set the internal JCHANGED flag - either of these is enough to
615 control the way the "get" function works - but the PCRE_INFO_JCHANGED
616 facility is supposed to tell if (?J) was ever used, so now (?J) at the
617 start sets both bits.
618
619 4. Added options (at build time, compile time, exec time) to change \R from
620 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
621
622 5. doc/pcresyntax.html was missing from the distribution.
623
624 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
625 compatibility, even though it is no longer used.
626
627 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
628 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
629 windows.h file is present (where different names are used). [This was
630 reversed later after testing - see 16 below.]
631
632 8. Changed all #include <config.h> to #include "config.h". There were also
633 some further <pcre.h> cases that I changed to "pcre.h".
634
635 9. When pcregrep was used with the --colour option, it missed the line ending
636 sequence off the lines that it output.
637
638 10. It was pointed out to me that arrays of string pointers cause lots of
639 relocations when a shared library is dynamically loaded. A technique of
640 using a single long string with a table of offsets can drastically reduce
641 these. I have refactored PCRE in four places to do this. The result is
642 dramatic:
643
644 Originally: 290
645 After changing UCP table: 187
646 After changing error message table: 43
647 After changing table of "verbs" 36
648 After changing table of Posix names 22
649
650 Thanks to the folks working on Gregex for glib for this insight.
651
652 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
653 unicode-properties was also set.
654
655 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
656
657 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
658 checked only for CRLF.
659
660 14. Added casts to pcretest.c to avoid compiler warnings.
661
662 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
663
664 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
665 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
666 entirely. This removes changes made in 7 above.
667
668 17. The CMake files have been updated, and there is now more information about
669 building with CMake in the NON-UNIX-USE document.
670
671
672 Version 7.3 28-Aug-07
673 ---------------------
674
675 1. In the rejigging of the build system that eventually resulted in 7.1, the
676 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
677 brackets there is not right, since it causes compilers to look for an
678 installed pcre.h, not the version that is in the source that is being
679 compiled (which of course may be different). I have changed it back to:
680
681 #include "pcre.h"
682
683 I have a vague recollection that the change was concerned with compiling in
684 different directories, but in the new build system, that is taken care of
685 by the VPATH setting the Makefile.
686
687 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
688 when the subject happened to end in the byte 0x85 (e.g. if the last
689 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
690 characters but of course it shouldn't be taken as a newline when it is part
691 of another character. The bug was that, for an unlimited repeat of . in
692 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
693 characters when looking for a newline.
694
695 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
696
697 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
698 in debug output.
699
700 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
701 long printing in the pcrecpp unittest when running under MinGW.
702
703 6. ESC_K was left out of the EBCDIC table.
704
705 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
706 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
707 limit also applies to "virtual nesting" when a pattern is recursive, and in
708 this case 1000 isn't so big. I have been able to remove this limit at the
709 expense of backing off one optimization in certain circumstances. Normally,
710 when pcre_exec() would call its internal match() function recursively and
711 immediately return the result unconditionally, it uses a "tail recursion"
712 feature to save stack. However, when a subpattern that can match an empty
713 string has an unlimited repetition quantifier, it no longer makes this
714 optimization. That gives it a stack frame in which to save the data for
715 checking that an empty string has been matched. Previously this was taken
716 from the 1000-entry workspace that had been reserved. So now there is no
717 explicit limit, but more stack is used.
718
719 8. Applied Daniel's patches to solve problems with the import/export magic
720 syntax that is required for Windows, and which was going wrong for the
721 pcreposix and pcrecpp parts of the library. These were overlooked when this
722 problem was solved for the main library.
723
724 9. There were some crude static tests to avoid integer overflow when computing
725 the size of patterns that contain repeated groups with explicit upper
726 limits. As the maximum quantifier is 65535, the maximum group length was
727 set at 30,000 so that the product of these two numbers did not overflow a
728 32-bit integer. However, it turns out that people want to use groups that
729 are longer than 30,000 bytes (though not repeat them that many times).
730 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
731 made it possible to implement the integer overflow checks in a much more
732 dynamic way, which I have now done. The artificial limitation on group
733 length has been removed - we now have only the limit on the total length of
734 the compiled pattern, which depends on the LINK_SIZE setting.
735
736 10. Fixed a bug in the documentation for get/copy named substring when
737 duplicate names are permitted. If none of the named substrings are set, the
738 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
739 empty string.
740
741 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
742 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
743 because the ] is interpreted as the first data character and the
744 terminating ] is not found. PCRE has been made compatible with Perl in this
745 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
746 cause memory overwriting.
747
748 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
749 string has been matched (to stop an infinite loop). It was not recognizing
750 a conditional subpattern that could match an empty string if that
751 subpattern was within another subpattern. For example, it looped when
752 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
753 condition was not nested. This bug has been fixed.
754
755 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
756 past the start of the subject in the presence of bytes with the top bit
757 set, for example "\x8aBCD".
758
759 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
760 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
761
762 14. Optimized (?!) to (*FAIL).
763
764 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
765 This restricts code points to be within the range 0 to 0x10FFFF, excluding
766 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
767 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
768 does: it's just the validity check that is more restrictive.
769
770 16. Inserted checks for integer overflows during escape sequence (backslash)
771 processing, and also fixed erroneous offset values for syntax errors during
772 backslash processing.
773
774 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
775 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
776
777 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
778 caused an overrun.
779
780 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
781 something other than just ASCII characters) inside a group that had an
782 unlimited repeat caused a loop at compile time (while checking to see
783 whether the group could match an empty string).
784
785 20. Debugging a pattern containing \p or \P could cause a crash. For example,
786 [\P{Any}] did so. (Error in the code for printing property names.)
787
788 21. An orphan \E inside a character class could cause a crash.
789
790 22. A repeated capturing bracket such as (A)? could cause a wild memory
791 reference during compilation.
792
793 23. There are several functions in pcre_compile() that scan along a compiled
794 expression for various reasons (e.g. to see if it's fixed length for look
795 behind). There were bugs in these functions when a repeated \p or \P was
796 present in the pattern. These operators have additional parameters compared
797 with \d, etc, and these were not being taken into account when moving along
798 the compiled data. Specifically:
799
800 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
801 length.
802
803 (b) An item such as \pL+ within a repeated group could cause crashes or
804 loops.
805
806 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
807 "reference to non-existent subpattern" error.
808
809 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
810
811 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
812 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
813
814 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
815
816 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
817 character were causing crashes (broken optimization).
818
819 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
820 \p or \P) caused a compile-time loop.
821
822 28. More problems have arisen in unanchored patterns when CRLF is a valid line
823 break. For example, the unstudied pattern [\r\n]A does not match the string
824 "\r\nA" because change 7.0/46 below moves the current point on by two
825 characters after failing to match at the start. However, the pattern \nA
826 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
827 the same is true. There doesn't seem any very clean way out of this, but
828 what I have chosen to do makes the common cases work: PCRE now takes note
829 of whether there can be an explicit match for \r or \n anywhere in the
830 pattern, and if so, 7.0/46 no longer applies. As part of this change,
831 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
832 pattern has explicit CR or LF references.
833
834 29. Added (*CR) etc for changing newline setting at start of pattern.
835
836
837 Version 7.2 19-Jun-07
838 ---------------------
839
840 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
841 which is apparently normally available under Windows.
842
843 2. Re-jig the pcregrep tests with different newline settings in an attempt
844 to make them independent of the local environment's newline setting.
845
846 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
847
848 4. Some of the "internals" tests were previously cut out when the link size
849 was not 2, because the output contained actual offsets. The recent new
850 "Z" feature of pcretest means that these can be cut out, making the tests
851 usable with all link sizes.
852
853 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
854 stack recursion. This gives a massive performance boost under BSD, but just
855 a small improvement under Linux. However, it saves one field in the frame
856 in all cases.
857
858 6. Added more features from the forthcoming Perl 5.10:
859
860 (a) (?-n) (where n is a string of digits) is a relative subroutine or
861 recursion call. It refers to the nth most recently opened parentheses.
862
863 (b) (?+n) is also a relative subroutine call; it refers to the nth next
864 to be opened parentheses.
865
866 (c) Conditions that refer to capturing parentheses can be specified
867 relatively, for example, (?(-2)... or (?(+3)...
868
869 (d) \K resets the start of the current match so that everything before
870 is not part of it.
871
872 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
873
874 (f) \g{name} is another synonym - part of Perl 5.10's unification of
875 reference syntax.
876
877 (g) (?| introduces a group in which the numbering of parentheses in each
878 alternative starts with the same number.
879
880 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
881
882 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
883 PCRE_INFO_JCHANGED.
884
885 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
886 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
887 for detecting groups that can match an empty string.
888
889 9. A pattern with a very large number of alternatives (more than several
890 hundred) was running out of internal workspace during the pre-compile
891 phase, where pcre_compile() figures out how much memory will be needed. A
892 bit of new cunning has reduced the workspace needed for groups with
893 alternatives. The 1000-alternative test pattern now uses 12 bytes of
894 workspace instead of running out of the 4096 that are available.
895
896 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
897
898 11. Applied patch from Google to remove an optimization that didn't quite work.
899 The report of the bug said:
900
901 pcrecpp::RE("a*").FullMatch("aaa") matches, while
902 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
903 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
904
905 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
906 it matched the wrong number of bytes.
907
908
909 Version 7.1 24-Apr-07
910 ---------------------
911
912 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
913 that is more "standard", making use of automake and other Autotools. There
914 is some re-arrangement of the files and adjustment of comments consequent
915 on this.
916
917 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
918 for recursive directory scanning broke on some systems because the files
919 are not scanned in any specific order and on different systems the order
920 was different. A call to "sort" has been inserted into RunGrepTest for the
921 approprate test as a short-term fix. In the longer term there may be an
922 alternative.
923
924 3. I had an email from Eric Raymond about problems translating some of PCRE's
925 man pages to HTML (despite the fact that I distribute HTML pages, some
926 people do their own conversions for various reasons). The problems
927 concerned the use of low-level troff macros .br and .in. I have therefore
928 removed all such uses from the man pages (some were redundant, some could
929 be replaced by .nf/.fi pairs). The 132html script that I use to generate
930 HTML has been updated to handle .nf/.fi and to complain if it encounters
931 .br or .in.
932
933 4. Updated comments in configure.ac that get placed in config.h.in and also
934 arranged for config.h to be included in the distribution, with the name
935 config.h.generic, for the benefit of those who have to compile without
936 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
937
938 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
939 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
940 makevp.bat for the current PCRE, using the additional files
941 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
942
943 6. A Windows user reported a minor discrepancy with test 2, which turned out
944 to be caused by a trailing space on an input line that had got lost in his
945 copy. The trailing space was an accident, so I've just removed it.
946
947 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
948 that is needed.
949
950 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
951 as "const" (a) because they are and (b) because it helps the PHP
952 maintainers who have recently made a script to detect big data structures
953 in the php code that should be moved to the .rodata section. I remembered
954 to update Builducptable as well, so it won't revert if ucptable.h is ever
955 re-created.
956
957 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
958 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
959 order to be able to cut out the UTF-8 tables in the latter when UTF-8
960 support is not required. This saves 1.5-2K of code, which is important in
961 some applications.
962
963 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
964 so as not to refer to the tables, even though these functions will never be
965 called when UTF-8 support is disabled. Otherwise there are problems with a
966 shared library.
967
968 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
969
970 (a) It was defining its arguments as char * instead of void *.
971
972 (b) It was assuming that all moves were upwards in memory; this was true
973 a long time ago when I wrote it, but is no longer the case.
974
975 The emulated memove() is provided for those environments that have neither
976 memmove() nor bcopy(). I didn't think anyone used it these days, but that
977 is clearly not the case, as these two bugs were recently reported.
978
979 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
980 and Detrail to create the HTML documentation, the .txt form of the man
981 pages, and it removes trailing spaces from listed files. It also creates
982 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
983 case, it wraps all the #defines with #ifndefs. This script should be run
984 before "make dist".
985
986 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
987 with Unicode property support.
988
989 (a) For a maximizing quantifier, if the two different cases of the
990 character were of different lengths in their UTF-8 codings (there are
991 some cases like this - I found 11), and the matching function had to
992 back up over a mixture of the two cases, it incorrectly assumed they
993 were both the same length.
994
995 (b) When PCRE was configured to use the heap rather than the stack for
996 recursion during matching, it was not correctly preserving the data for
997 the other case of a UTF-8 character when checking ahead for a match
998 while processing a minimizing repeat. If the check also involved
999 matching a wide character, but failed, corruption could cause an
1000 erroneous result when trying to check for a repeat of the original
1001 character.
1002
1003 13. Some tidying changes to the testing mechanism:
1004
1005 (a) The RunTest script now detects the internal link size and whether there
1006 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
1007 values substituted by "configure". (The RunGrepTest script already did
1008 this for UTF-8.) The configure.ac script no longer substitutes the
1009 relevant variables.
1010
1011 (b) The debugging options /B and /D in pcretest show the compiled bytecode
1012 with length and offset values. This means that the output is different
1013 for different internal link sizes. Test 2 is skipped for link sizes
1014 other than 2 because of this, bypassing the problem. Unfortunately,
1015 there was also a test in test 3 (the locale tests) that used /B and
1016 failed for link sizes other than 2. Rather than cut the whole test out,
1017 I have added a new /Z option to pcretest that replaces the length and
1018 offset values with spaces. This is now used to make test 3 independent
1019 of link size. (Test 2 will be tidied up later.)
1020
1021 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
1022 segmentation fault instead of returning the appropriate error message.
1023
1024 15. In multiline mode when the newline sequence was set to "any", the pattern
1025 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
1026 This doesn't seem right; it now treats the CRLF combination as the line
1027 ending, and so does not match in that case. It's only a pattern such as ^$
1028 that would hit this one: something like ^ABC$ would have failed after \r
1029 and then tried again after \r\n.
1030
1031 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
1032 in an attempt to make files that differ only in their line terminators
1033 compare equal. This works on Linux.
1034
1035 17. Under certain error circumstances pcregrep might try to free random memory
1036 as it exited. This is now fixed, thanks to valgrind.
1037
1038 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
1039 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
1040 was because its rules for how to advance for /g after matching an empty
1041 string at the end of a line did not allow for this case. They now check for
1042 it specially.
1043
1044 20. pcretest is supposed to handle patterns and data of any length, by
1045 extending its buffers when necessary. It was getting this wrong when the
1046 buffer for a data line had to be extended.
1047
1048 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
1049 CRLF as a newline sequence.
1050
1051 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
1052 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
1053 I have nevertheless tidied it up.
1054
1055 23. Added some casts to kill warnings from HP-UX ia64 compiler.
1056
1057 24. Added a man page for pcre-config.
1058
1059
1060 Version 7.0 19-Dec-06
1061 ---------------------
1062
1063 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
1064 moving to gcc 4.1.1.
1065
1066 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
1067 sys/time.h, which is documented as needed for this function. It doesn't
1068 seem to matter on Linux, but it showed up on some releases of OS X.
1069
1070 3. It seems that there are systems where bytes whose values are greater than
1071 127 match isprint() in the "C" locale. The "C" locale should be the
1072 default when a C program starts up. In most systems, only ASCII printing
1073 characters match isprint(). This difference caused the output from pcretest
1074 to vary, making some of the tests fail. I have changed pcretest so that:
1075
1076 (a) When it is outputting text in the compiled version of a pattern, bytes
1077 other than 32-126 are always shown as hex escapes.
1078
1079 (b) When it is outputting text that is a matched part of a subject string,
1080 it does the same, unless a different locale has been set for the match
1081 (using the /L modifier). In this case, it uses isprint() to decide.
1082
1083 4. Fixed a major bug that caused incorrect computation of the amount of memory
1084 required for a compiled pattern when options that changed within the
1085 pattern affected the logic of the preliminary scan that determines the
1086 length. The relevant options are -x, and -i in UTF-8 mode. The result was
1087 that the computed length was too small. The symptoms of this bug were
1088 either the PCRE error "internal error: code overflow" from pcre_compile(),
1089 or a glibc crash with a message such as "pcretest: free(): invalid next
1090 size (fast)". Examples of patterns that provoked this bug (shown in
1091 pcretest format) are:
1092
1093 /(?-x: )/x
1094 /(?x)(?-x: \s*#\s*)/
1095 /((?i)[\x{c0}])/8
1096 /(?i:[\x{c0}])/8
1097
1098 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
1099 is now done differently.
1100
1101 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
1102 wrapper classes; (b) implement a new function in the C++ scanner that is
1103 more efficient than the old way of doing things because it avoids levels of
1104 recursion in the regex matching; (c) add a paragraph to the documentation
1105 for the FullMatch() function.
1106
1107 6. The escape sequence \n was being treated as whatever was defined as
1108 "newline". Not only was this contrary to the documentation, which states
1109 that \n is character 10 (hex 0A), but it also went horribly wrong when
1110 "newline" was defined as CRLF. This has been fixed.
1111
1112 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
1113 was being set to -1 for the "end of line" case (supposedly a value that no
1114 character can have). Though this value is never used (the check for end of
1115 line is "zero bytes in current character"), it caused compiler complaints.
1116 I've changed it to 0xffffffff.
1117
1118 8. In pcre_version.c, the version string was being built by a sequence of
1119 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
1120 string (as it is for production releases) called a macro with an empty
1121 argument. The C standard says the result of this is undefined. The gcc
1122 compiler treats it as an empty string (which was what was wanted) but it is
1123 reported that Visual C gives an error. The source has been hacked around to
1124 avoid this problem.
1125
1126 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
1127 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
1128 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
1129 of them did).
1130
1131 10. Originally, pcretest opened its input and output without "b"; then I was
1132 told that "b" was needed in some environments, so it was added for release
1133 5.0 to both the input and output. (It makes no difference on Unix-like
1134 systems.) Later I was told that it is wrong for the input on Windows. I've
1135 now abstracted the modes into two macros, to make it easier to fiddle with
1136 them, and removed "b" from the input mode under Windows.
1137
1138 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
1139
1140 12. Added -help and --help to pcretest as an official way of being reminded
1141 of the options.
1142
1143 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
1144 and pcrecpp.cc because they annoy compilers at high warning levels.
1145
1146 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
1147
1148 15. Fixed an occurrence of == in configure.ac that should have been = (shell
1149 scripts are not C programs :-) and which was not noticed because it works
1150 on Linux.
1151
1152 16. pcretest is supposed to handle any length of pattern and data line (as one
1153 line or as a continued sequence of lines) by extending its input buffer if
1154 necessary. This feature was broken for very long pattern lines, leading to
1155 a string of junk being passed to pcre_compile() if the pattern was longer
1156 than about 50K.
1157
1158 17. I have done a major re-factoring of the way pcre_compile() computes the
1159 amount of memory needed for a compiled pattern. Previously, there was code
1160 that made a preliminary scan of the pattern in order to do this. That was
1161 OK when PCRE was new, but as the facilities have expanded, it has become
1162 harder and harder to keep it in step with the real compile phase, and there
1163 have been a number of bugs (see for example, 4 above). I have now found a
1164 cunning way of running the real compile function in a "fake" mode that
1165 enables it to compute how much memory it would need, while actually only
1166 ever using a few hundred bytes of working memory and without too many
1167 tests of the mode. This should make future maintenance and development
1168 easier. A side effect of this work is that the limit of 200 on the nesting
1169 depth of parentheses has been removed (though this was never a serious
1170 limitation, I suspect). However, there is a downside: pcre_compile() now
1171 runs more slowly than before (30% or more, depending on the pattern). I
1172 hope this isn't a big issue. There is no effect on runtime performance.
1173
1174 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
1175 newline (only possible for the last line of a file) and it was a
1176 pattern that set a locale (followed by /Lsomething), pcretest crashed.
1177
1178 19. Added additional timing features to pcretest. (1) The -tm option now times
1179 matching only, not compiling. (2) Both -t and -tm can be followed, as a
1180 separate command line item, by a number that specifies the number of
1181 repeats to use when timing. The default is 50000; this gives better
1182 precision, but takes uncomfortably long for very large patterns.
1183
1184 20. Extended pcre_study() to be more clever in cases where a branch of a
1185 subpattern has no definite first character. For example, (a*|b*)[cd] would
1186 previously give no result from pcre_study(). Now it recognizes that the
1187 first character must be a, b, c, or d.
1188
1189 21. There was an incorrect error "recursive call could loop indefinitely" if
1190 a subpattern (or the entire pattern) that was being tested for matching an
1191 empty string contained only one non-empty item after a nested subpattern.
1192 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
1193 incorrectly, because the \d was being skipped in the check.
1194
1195 22. The pcretest program now has a new pattern option /B and a command line
1196 option -b, which is equivalent to adding /B to every pattern. This causes
1197 it to show the compiled bytecode, without the additional information that
1198 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
1199 is the same as /B/I).
1200
1201 23. A new optimization is now able automatically to treat some sequences such
1202 as a*b as a*+b. More specifically, if something simple (such as a character
1203 or a simple class like \d) has an unlimited quantifier, and is followed by
1204 something that cannot possibly match the quantified thing, the quantifier
1205 is automatically "possessified".
1206
1207 24. A recursive reference to a subpattern whose number was greater than 39
1208 went wrong under certain circumstances in UTF-8 mode. This bug could also
1209 have affected the operation of pcre_study().
1210
1211 25. Realized that a little bit of performance could be had by replacing
1212 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
1213
1214 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
1215
1216 27. Possessive quantifiers such as a++ were previously implemented by turning
1217 them into atomic groups such as ($>a+). Now they have their own opcodes,
1218 which improves performance. This includes the automatically created ones
1219 from 23 above.
1220
1221 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
1222 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
1223 the first matched character to be a colon. This applied both to named and
1224 numbered groups.
1225
1226 29. The ucpinternal.h header file was missing its idempotency #ifdef.
1227
1228 30. I was sent a "project" file called libpcre.a.dev which I understand makes
1229 building PCRE on Windows easier, so I have included it in the distribution.
1230
1231 31. There is now a check in pcretest against a ridiculously large number being
1232 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
1233 loop, the loop is abandoned.
1234
1235 32. Forward references to subpatterns in conditions such as (?(2)...) where
1236 subpattern 2 is defined later cause pcre_compile() to search forwards in
1237 the pattern for the relevant set of parentheses. This search went wrong
1238 when there were unescaped parentheses in a character class, parentheses
1239 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
1240
1241 33. "Subroutine" calls and backreferences were previously restricted to
1242 referencing subpatterns earlier in the regex. This restriction has now
1243 been removed.
1244
1245 34. Added a number of extra features that are going to be in Perl 5.10. On the
1246 whole, these are just syntactic alternatives for features that PCRE had
1247 previously implemented using the Python syntax or my own invention. The
1248 other formats are all retained for compatibility.
1249
1250 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
1251 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
1252 also .NET compatible.
1253
1254 (b) A recursion or subroutine call to a named group can now be defined as
1255 (?&name) as well as (?P>name).
1256
1257 (c) A backreference to a named group can now be defined as \k<name> or
1258 \k'name' as well as (?P=name). The new forms, as well as being in Perl
1259 5.10, are also .NET compatible.
1260
1261 (d) A conditional reference to a named group can now use the syntax
1262 (?(<name>) or (?('name') as well as (?(name).
1263
1264 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
1265 groups (named and numbered) that are never evaluated inline, but can be
1266 called as "subroutines" from elsewhere. In effect, the DEFINE condition
1267 is always false. There may be only one alternative in such a group.
1268
1269 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
1270 as the simple (?(R). The condition is true only if the most recent
1271 recursion is that of the given number or name. It does not search out
1272 through the entire recursion stack.
1273
1274 (g) The escape \gN or \g{N} has been added, where N is a positive or
1275 negative number, specifying an absolute or relative reference.
1276
1277 35. Tidied to get rid of some further signed/unsigned compiler warnings and
1278 some "unreachable code" warnings.
1279
1280 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
1281 things, this adds five new scripts.
1282
1283 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
1284 There were also incompatibilities regarding the handling of \Q..\E inside
1285 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
1286 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
1287
1288 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
1289 matches an empty string, and forcibly breaks the loop. There were bugs in
1290 this code in non-simple cases. For a pattern such as ^(a()*)* matched
1291 against aaaa the result was just "a" rather than "aaaa", for example. Two
1292 separate and independent bugs (that affected different cases) have been
1293 fixed.
1294
1295 39. Refactored the code to abolish the use of different opcodes for small
1296 capturing bracket numbers. This is a tidy that I avoided doing when I
1297 removed the limit on the number of capturing brackets for 3.5 back in 2001.
1298 The new approach is not only tidier, it makes it possible to reduce the
1299 memory needed to fix the previous bug (38).
1300
1301 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
1302 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
1303 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
1304 mode.
1305
1306 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
1307 report.
1308
1309 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
1310 copy construction and assignment in the C++ wrapper.
1311
1312 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
1313 couple of bugs that could have given wrong results in the "--newline=crlf"
1314 case.
1315
1316 44. Added a number of casts and did some reorganization of signed/unsigned int
1317 variables following suggestions from Dair Grant. Also renamed the variable
1318 "this" as "item" because it is a C++ keyword.
1319
1320 45. Arranged for dftables to add
1321
1322 #include "pcre_internal.h"
1323
1324 to pcre_chartables.c because without it, gcc 4.x may remove the array
1325 definition from the final binary if PCRE is built into a static library and
1326 dead code stripping is activated.
1327
1328 46. For an unanchored pattern, if a match attempt fails at the start of a
1329 newline sequence, and the newline setting is CRLF or ANY, and the next two
1330 characters are CRLF, advance by two characters instead of one.
1331
1332
1333 Version 6.7 04-Jul-06
1334 ---------------------
1335
1336 1. In order to handle tests when input lines are enormously long, pcretest has
1337 been re-factored so that it automatically extends its buffers when
1338 necessary. The code is crude, but this _is_ just a test program. The
1339 default size has been increased from 32K to 50K.
1340
1341 2. The code in pcre_study() was using the value of the re argument before
1342 testing it for NULL. (Of course, in any sensible call of the function, it
1343 won't be NULL.)
1344
1345 3. The memmove() emulation function in pcre_internal.h, which is used on
1346 systems that lack both memmove() and bcopy() - that is, hardly ever -
1347 was missing a "static" storage class specifier.
1348
1349 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
1350 containing an extended class (one that cannot be represented by a bitmap
1351 because it contains high-valued characters or Unicode property items, e.g.
1352 [\pZ]). Almost always one would set UTF-8 mode when processing such a
1353 pattern, but PCRE should not loop if you do not (it no longer does).
1354 [Detail: two cases were found: (a) a repeated subpattern containing an
1355 extended class; (b) a recursive reference to a subpattern that followed a
1356 previous extended class. It wasn't skipping over the extended class
1357 correctly when UTF-8 mode was not set.]
1358
1359 5. A negated single-character class was not being recognized as fixed-length
1360 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
1361 compile error "lookbehind assertion is not fixed length".
1362
1363 6. The RunPerlTest auxiliary script was showing an unexpected difference
1364 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
1365 write a Perl script that can interpret lines of an input file either as
1366 byte characters or as UTF-8, which is what "perltest" was being required to
1367 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
1368 can't do is switch easily at run time between having the "use utf8;" pragma
1369 or not. In the end, I fudged it by using the RunPerlTest script to insert
1370 "use utf8;" explicitly for the UTF-8 tests.
1371
1372 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
1373 the end of the subject string, contrary to the documentation and to what
1374 Perl does. This was true of both matching functions. Now it matches only at
1375 the start of the subject and immediately after *internal* newlines.
1376
1377 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
1378 a pointer to an int instead of a pointer to an unsigned long int. This
1379 caused problems on 64-bit systems.
1380
1381 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
1382 instance of the 'standard' template library not being so standard".
1383
1384 10. There was no check on the number of named subpatterns nor the maximum
1385 length of a subpattern name. The product of these values is used to compute
1386 the size of the memory block for a compiled pattern. By supplying a very
1387 long subpattern name and a large number of named subpatterns, the size
1388 computation could be caused to overflow. This is now prevented by limiting
1389 the length of names to 32 characters, and the number of named subpatterns
1390 to 10,000.
1391
1392 11. Subpatterns that are repeated with specific counts have to be replicated in
1393 the compiled pattern. The size of memory for this was computed from the
1394 length of the subpattern and the repeat count. The latter is limited to
1395 65535, but there was no limit on the former, meaning that integer overflow
1396 could in principle occur. The compiled length of a repeated subpattern is
1397 now limited to 30,000 bytes in order to prevent this.
1398
1399 12. Added the optional facility to have named substrings with the same name.
1400
1401 13. Added the ability to use a named substring as a condition, using the
1402 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
1403 are numbers (not recommended). Forward references are permitted.
1404
1405 14. Added forward references in named backreferences (if you see what I mean).
1406
1407 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
1408 pattern could run off the end of the subject. For example, the pattern
1409 "(?s)(.{1,5})"8 did this with the subject "ab".
1410
1411 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
1412 PCRE_CASELESS was set when matching characters that were quantified with ?
1413 or *.
1414
1415 17. A character class other than a single negated character that had a minimum
1416 but no maximum quantifier - for example [ab]{6,} - was not handled
1417 correctly by pce_dfa_exec(). It would match only one character.
1418
1419 18. A valid (though odd) pattern that looked like a POSIX character
1420 class but used an invalid character after [ (for example [[,abc,]]) caused
1421 pcre_compile() to give the error "Failed: internal error: code overflow" or
1422 in some cases to crash with a glibc free() error. This could even happen if
1423 the pattern terminated after [[ but there just happened to be a sequence of
1424 letters, a binary zero, and a closing ] in the memory that followed.
1425
1426 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
1427 over the years. Originally (before any Unicode support), just the bottom 8
1428 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
1429 output from "man perlunicode" includes this:
1430
1431 The regular expression compiler produces polymorphic opcodes. That
1432 is, the pattern adapts to the data and automatically switches to
1433 the Unicode character scheme when presented with Unicode data--or
1434 instead uses a traditional byte scheme when presented with byte
1435 data.
1436
1437 Sadly, a wide octal escape does not cause a switch, and in a string with
1438 no other multibyte characters, these octal escapes are treated as before.
1439 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
1440 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
1441 Unicode string.
1442
1443 I have not perpetrated such confusion in PCRE. Up till now, it took just
1444 the bottom 8 bits, as in old Perl. I have now made octal escapes with
1445 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
1446 translate to the appropriate multibyte character.
1447
1448 29. Applied some refactoring to reduce the number of warnings from Microsoft
1449 and Borland compilers. This has included removing the fudge introduced
1450 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
1451 a warning about an unused variable.
1452
1453 21. PCRE has not included VT (character 0x0b) in the set of whitespace
1454 characters since release 4.0, because Perl (from release 5.004) does not.
1455 [Or at least, is documented not to: some releases seem to be in conflict
1456 with the documentation.] However, when a pattern was studied with
1457 pcre_study() and all its branches started with \s, PCRE still included VT
1458 as a possible starting character. Of course, this did no harm; it just
1459 caused an unnecessary match attempt.
1460
1461 22. Removed a now-redundant internal flag bit that recorded the fact that case
1462 dependency changed within the pattern. This was once needed for "required
1463 byte" processing, but is no longer used. This recovers a now-scarce options
1464 bit. Also moved the least significant internal flag bit to the most-
1465 significant bit of the word, which was not previously used (hangover from
1466 the days when it was an int rather than a uint) to free up another bit for
1467 the future.
1468
1469 23. Added support for CRLF line endings as well as CR and LF. As well as the
1470 default being selectable at build time, it can now be changed at runtime
1471 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
1472 specify that it is scanning data with non-default line endings.
1473
1474 24. Changed the definition of CXXLINK to make it agree with the definition of
1475 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
1476
1477 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
1478 recursions. This makes a big different to stack usage for some patterns.
1479
1480 26. If a subpattern containing a named recursion or subroutine reference such
1481 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
1482 the space required for the compiled pattern went wrong and gave too small a
1483 value. Depending on the environment, this could lead to "Failed: internal
1484 error: code overflow at offset 49" or "glibc detected double free or
1485 corruption" errors.
1486
1487 27. Applied patches from Google (a) to support the new newline modes and (b) to
1488 advance over multibyte UTF-8 characters in GlobalReplace.
1489
1490 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
1491 difference for some implementation of PCRE in some Windows version.
1492
1493 29. Added some extra testing facilities to pcretest:
1494
1495 \q<number> in a data line sets the "match limit" value
1496 \Q<number> in a data line sets the "match recursion limt" value
1497 -S <number> sets the stack size, where <number> is in megabytes
1498
1499 The -S option isn't available for Windows.
1500
1501
1502 Version 6.6 06-Feb-06
1503 ---------------------
1504
1505 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
1506 in pcreposix.h. I have copied the definition from pcre.h.
1507
1508 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
1509 because pcre.h is no longer a built file.
1510
1511 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
1512 not normally included in the compiled code.
1513
1514
1515 Version 6.5 01-Feb-06
1516 ---------------------
1517
1518 1. When using the partial match feature with pcre_dfa_exec(), it was not
1519 anchoring the second and subsequent partial matches at the new starting
1520 point. This could lead to incorrect results. For example, with the pattern
1521 /1234/, partially matching against "123" and then "a4" gave a match.
1522
1523 2. Changes to pcregrep:
1524
1525 (a) All non-match returns from pcre_exec() were being treated as failures
1526 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
1527 error message is output. Some extra information is given for the
1528 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
1529 probably the only errors that are likely to be caused by users (by
1530 specifying a regex that has nested indefinite repeats, for instance).
1531 If there are more than 20 of these errors, pcregrep is abandoned.
1532
1533 (b) A binary zero was treated as data while matching, but terminated the
1534 output line if it was written out. This has been fixed: binary zeroes
1535 are now no different to any other data bytes.
1536
1537 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
1538 used to set a locale for matching. The --locale=xxxx long option has
1539 been added (no short equivalent) to specify a locale explicitly on the
1540 pcregrep command, overriding the environment variables.
1541
1542 (d) When -B was used with -n, some line numbers in the output were one less
1543 than they should have been.
1544
1545 (e) Added the -o (--only-matching) option.
1546
1547 (f) If -A or -C was used with -c (count only), some lines of context were
1548 accidentally printed for the final match.
1549
1550 (g) Added the -H (--with-filename) option.
1551
1552 (h) The combination of options -rh failed to suppress file names for files
1553 that were found from directory arguments.
1554
1555 (i) Added the -D (--devices) and -d (--directories) options.
1556
1557 (j) Added the -F (--fixed-strings) option.
1558
1559 (k) Allow "-" to be used as a file name for -f as well as for a data file.
1560
1561 (l) Added the --colo(u)r option.
1562
1563 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
1564 is not present by default.
1565
1566 3. A nasty bug was discovered in the handling of recursive patterns, that is,
1567 items such as (?R) or (?1), when the recursion could match a number of
1568 alternatives. If it matched one of the alternatives, but subsequently,
1569 outside the recursion, there was a failure, the code tried to back up into
1570 the recursion. However, because of the way PCRE is implemented, this is not
1571 possible, and the result was an incorrect result from the match.
1572
1573 In order to prevent this happening, the specification of recursion has
1574 been changed so that all such subpatterns are automatically treated as
1575 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
1576
1577 4. I had overlooked the fact that, in some locales, there are characters for
1578 which isalpha() is true but neither isupper() nor islower() are true. In
1579 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
1580 and ordfeminine) are like this. This affected the treatment of \w and \W
1581 when they appeared in character classes, but not when they appeared outside
1582 a character class. The bit map for "word" characters is now created
1583 separately from the results of isalnum() instead of just taking it from the
1584 upper, lower, and digit maps. (Plus the underscore character, of course.)
1585
1586 5. The above bug also affected the handling of POSIX character classes such as
1587 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
1588 permanent tables. Instead, the bit maps for such a class were previously
1589 created as the appropriate unions of the upper, lower, and digit bitmaps.
1590 Now they are created by subtraction from the [[:word:]] class, which has
1591 its own bitmap.
1592
1593 6. The [[:blank:]] character class matches horizontal, but not vertical space.
1594 It is created by subtracting the vertical space characters (\x09, \x0a,
1595 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
1596 subtraction was done in the overall bitmap for a character class, meaning
1597 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
1598 be recognized. This bug has been fixed.
1599
1600 7. Patches from the folks at Google:
1601
1602 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
1603 real life, but is still worth protecting against".
1604
1605 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
1606 regular expressions".
1607
1608 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
1609 have it.
1610
1611 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
1612 "configure" and the latter not, in order to fix a problem somebody had
1613 with compiling the Arg class on HP-UX.
1614
1615 (e) Improve the error-handling of the C++ wrapper a little bit.
1616
1617 (f) New tests for checking recursion limiting.
1618
1619 8. The pcre_memmove() function, which is used only if the environment does not
1620 have a standard memmove() function (and is therefore rarely compiled),
1621 contained two bugs: (a) use of int instead of size_t, and (b) it was not
1622 returning a result (though PCRE never actually uses the result).
1623
1624 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
1625 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
1626 returned instead of calling malloc() with an overflowing number that would
1627 most likely cause subsequent chaos.
1628
1629 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
1630
1631 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
1632 with this option is matched, the nmatch and pmatch options of regexec() are
1633 ignored.
1634
1635 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
1636 provided in case anyone wants to the the POSIX interface with UTF-8
1637 strings.
1638
1639 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
1640 C++ linking (needed for some HP-UX environments).
1641
1642 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
1643 (unused parameter) and in the pcre_printint() function (omitted "default"
1644 switch label when the default is to do nothing).
1645
1646 15. Added some code to make it possible, when PCRE is compiled as a C++
1647 library, to replace subject pointers for pcre_exec() with a smart pointer
1648 class, thus making it possible to process discontinuous strings.
1649
1650 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
1651 much the same function. They were added by different people who were trying
1652 to make PCRE easy to compile on non-Unix systems. It has been suggested
1653 that PCRE_EXPORT be abolished now that there is more automatic apparatus
1654 for compiling on Windows systems. I have therefore replaced it with
1655 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
1656 defaults to "extern" for C or "extern C" for C++, which works fine on
1657 Unix-like systems. It is now possible to override the value of PCRE_DATA_
1658 SCOPE with something explicit in config.h. In addition:
1659
1660 (a) pcreposix.h still had just "extern" instead of either of these macros;
1661 I have replaced it with PCRE_DATA_SCOPE.
1662
1663 (b) Functions such as _pcre_xclass(), which are internal to the library,
1664 but external in the C sense, all had PCRE_EXPORT in their definitions.
1665 This is apparently wrong for the Windows case, so I have removed it.
1666 (It makes no difference on Unix-like systems.)
1667
1668 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
1669 of recursive calls to match(). This is different to MATCH_LIMIT because
1670 that limits the total number of calls to match(), not all of which increase
1671 the depth of recursion. Limiting the recursion depth limits the amount of
1672 stack (or heap if NO_RECURSE is set) that is used. The default can be set
1673 when PCRE is compiled, and changed at run time. A patch from Google adds
1674 this functionality to the C++ interface.
1675
1676 18. Changes to the handling of Unicode character properties:
1677
1678 (a) Updated the table to Unicode 4.1.0.
1679
1680 (b) Recognize characters that are not in the table as "Cn" (undefined).
1681
1682 (c) I revised the way the table is implemented to a much improved format
1683 which includes recognition of ranges. It now supports the ranges that
1684 are defined in UnicodeData.txt, and it also amalgamates other
1685 characters into ranges. This has reduced the number of entries in the
1686 table from around 16,000 to around 3,000, thus reducing its size
1687 considerably. I realized I did not need to use a tree structure after
1688 all - a binary chop search is just as efficient. Having reduced the
1689 number of entries, I extended their size from 6 bytes to 8 bytes to
1690 allow for more data.
1691
1692 (d) Added support for Unicode script names via properties such as \p{Han}.
1693
1694 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
1695 matching that character.
1696
1697 20. When matching a repeated Unicode property with a minimum greater than zero,
1698 (for example \pL{2,}), PCRE could look past the end of the subject if it
1699 reached it while seeking the minimum number of characters. This could
1700 happen only if some of the characters were more than one byte long, because
1701 there is a check for at least the minimum number of bytes.
1702
1703 21. Refactored the implementation of \p and \P so as to be more general, to
1704 allow for more different types of property in future. This has changed the
1705 compiled form incompatibly. Anybody with saved compiled patterns that use
1706 \p or \P will have to recompile them.
1707
1708 22. Added "Any" and "L&" to the supported property types.
1709
1710 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
1711 but give a compile time error if the value is greater than 0xff.
1712
1713 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
1714 accidentally not being installed or uninstalled.
1715
1716 25. The pcre.h file was built from pcre.h.in, but the only changes that were
1717 made were to insert the current release number. This seemed silly, because
1718 it made things harder for people building PCRE on systems that don't run
1719 "configure". I have turned pcre.h into a distributed file, no longer built
1720 by "configure", with the version identification directly included. There is
1721 no longer a pcre.h.in file.
1722
1723 However, this change necessitated a change to the pcre-config script as
1724 well. It is built from pcre-config.in, and one of the substitutions was the
1725 release number. I have updated configure.ac so that ./configure now finds
1726 the release number by grepping pcre.h.
1727
1728 26. Added the ability to run the tests under valgrind.
1729
1730
1731 Version 6.4 05-Sep-05
1732 ---------------------
1733
1734 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
1735 "--" to be printed when multiple files were scanned, even when none of the
1736 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
1737 consider it to be a bug, and have restored the previous behaviour.
1738
1739 2. A couple of code tidies to get rid of compiler warnings.
1740
1741 3. The pcretest program used to cheat by referring to symbols in the library
1742 whose names begin with _pcre_. These are internal symbols that are not
1743 really supposed to be visible externally, and in some environments it is
1744 possible to suppress them. The cheating is now confined to including
1745 certain files from the library's source, which is a bit cleaner.
1746
1747 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
1748 file's purpose clearer.
1749
1750 5. Reorganized pcre_ucp_findchar().
1751
1752
1753 Version 6.3 15-Aug-05
1754 ---------------------
1755
1756 1. The file libpcre.pc.in did not have general read permission in the tarball.
1757
1758 2. There were some problems when building without C++ support:
1759
1760 (a) If C++ support was not built, "make install" and "make test" still
1761 tried to test it.
1762
1763 (b) There were problems when the value of CXX was explicitly set. Some
1764 changes have been made to try to fix these, and ...
1765
1766 (c) --disable-cpp can now be used to explicitly disable C++ support.
1767
1768 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
1769 backslash in a target when C++ was disabled. This confuses some
1770 versions of "make", apparently. Using an intermediate variable solves
1771 this. (Same for CPP_LOBJ.)
1772
1773 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
1774 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
1775 necessary on certain architectures.
1776
1777 4. Added a setting of -export-symbols-regex to the link command to remove
1778 those symbols that are exported in the C sense, but actually are local
1779 within the library, and not documented. Their names all begin with
1780 "_pcre_". This is not a perfect job, because (a) we have to except some
1781 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
1782 available (and never for static libraries). I have made a note to try to
1783 find a way round (a) in the future.
1784
1785
1786 Version 6.2 01-Aug-05
1787 ---------------------
1788
1789 1. There was no test for integer overflow of quantifier values. A construction
1790 such as {1111111111111111} would give undefined results. What is worse, if
1791 a minimum quantifier for a parenthesized subpattern overflowed and became
1792 negative, the calculation of the memory size went wrong. This could have
1793 led to memory overwriting.
1794
1795 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
1796
1797 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
1798 operating environments where this matters.
1799
1800 4. Applied Giuseppe Maxia's patch to add additional features for controlling
1801 PCRE options from within the C++ wrapper.
1802
1803 5. Named capturing subpatterns were not being correctly counted when a pattern
1804 was compiled. This caused two problems: (a) If there were more than 100
1805 such subpatterns, the calculation of the memory needed for the whole
1806 compiled pattern went wrong, leading to an overflow error. (b) Numerical
1807 back references of the form \12, where the number was greater than 9, were
1808 not recognized as back references, even though there were sufficient
1809 previous subpatterns.
1810
1811 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
1812 versions of gcc, e.g. 2.95.4.
1813
1814
1815 Version 6.1 21-Jun-05
1816 ---------------------
1817
1818 1. There was one reference to the variable "posix" in pcretest.c that was not
1819 surrounded by "#if !defined NOPOSIX".
1820
1821 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
1822 the cross-check on the old pcre_info() function, for the benefit of the
1823 cut-down version of PCRE that is currently imported into Exim.
1824
1825 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
1826 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
1827 patterns that start (?i) but I don't think that matters. The use of (?i) is
1828 just an example; this all applies to the other options as well.
1829
1830 4. Since libtool seems to echo the compile commands it is issuing, the output
1831 from "make" can be reduced a bit by putting "@" in front of each libtool
1832 compile command.
1833
1834 5. Patch from the folks at Google for configure.in to be a bit more thorough
1835 in checking for a suitable C++ installation before trying to compile the
1836 C++ stuff. This should fix a reported problem when a compiler was present,
1837 but no suitable headers.
1838
1839 6. The man pages all had just "PCRE" as their title. I have changed them to
1840 be the relevant file name. I have also arranged that these names are
1841 retained in the file doc/pcre.txt, which is a concatenation in text format
1842 of all the man pages except the little individual ones for each function.
1843
1844 7. The NON-UNIX-USE file had not been updated for the different set of source
1845 files that come with release 6. I also added a few comments about the C++
1846 wrapper.
1847
1848
1849 Version 6.0 07-Jun-05
1850 ---------------------
1851
1852 1. Some minor internal re-organization to help with my DFA experiments.
1853
1854 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
1855 didn't matter for the library itself when fully configured, but did matter
1856 when compiling without UCP support, or within Exim, where the ucp files are
1857 not imported.
1858
1859 3. Refactoring of the library code to split up the various functions into
1860 different source modules. The addition of the new DFA matching code (see
1861 below) to a single monolithic source would have made it really too
1862 unwieldy, quite apart from causing all the code to be include in a
1863 statically linked application, when only some functions are used. This is
1864 relevant even without the DFA addition now that patterns can be compiled in
1865 one application and matched in another.
1866
1867 The downside of splitting up is that there have to be some external
1868 functions and data tables that are used internally in different modules of
1869 the library but which are not part of the API. These have all had their
1870 names changed to start with "_pcre_" so that they are unlikely to clash
1871 with other external names.
1872
1873 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
1874 a different (DFA) algorithm. Although it is slower than the original
1875 function, it does have some advantages for certain types of matching
1876 problem.
1877
1878 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
1879 including restarting after a partial match.
1880
1881 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
1882 defined when compiling for Windows was sent to me. I have put it into the
1883 code, though I have no means of testing or verifying it.
1884
1885 7. Added the pcre_refcount() auxiliary function.
1886
1887 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
1888 match before or at the first newline in the subject string. In pcretest,
1889 the /f option on a pattern can be used to set this.
1890
1891 9. A repeated \w when used in UTF-8 mode with characters greater than 256
1892 would behave wrongly. This has been present in PCRE since release 4.0.
1893
1894 10. A number of changes to the pcregrep command:
1895
1896 (a) Refactored how -x works; insert ^(...)$ instead of setting
1897 PCRE_ANCHORED and checking the length, in preparation for adding
1898 something similar for -w.
1899
1900 (b) Added the -w (match as a word) option.
1901
1902 (c) Refactored the way lines are read and buffered so as to have more
1903 than one at a time available.
1904
1905 (d) Implemented a pcregrep test script.
1906
1907 (e) Added the -M (multiline match) option. This allows patterns to match
1908 over several lines of the subject. The buffering ensures that at least
1909 8K, or the rest of the document (whichever is the shorter) is available
1910 for matching (and similarly the previous 8K for lookbehind assertions).
1911
1912 (f) Changed the --help output so that it now says
1913
1914 -w, --word-regex(p)
1915
1916 instead of two lines, one with "regex" and the other with "regexp"
1917 because that confused at least one person since the short forms are the
1918 same. (This required a bit of code, as the output is generated
1919 automatically from a table. It wasn't just a text change.)
1920
1921 (g) -- can be used to terminate pcregrep options if the next thing isn't an
1922 option but starts with a hyphen. Could be a pattern or a path name
1923 starting with a hyphen, for instance.
1924
1925 (h) "-" can be given as a file name to represent stdin.
1926
1927 (i) When file names are being printed, "(standard input)" is used for
1928 the standard input, for compatibility with GNU grep. Previously
1929 "<stdin>" was used.
1930
1931 (j) The option --label=xxx can be used to supply a name to be used for
1932 stdin when file names are being printed. There is no short form.
1933
1934 (k) Re-factored the options decoding logic because we are going to add
1935 two more options that take data. Such options can now be given in four
1936 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
1937
1938 (l) Added the -A, -B, and -C options for requesting that lines of context
1939 around matches be printed.
1940
1941 (m) Added the -L option to print the names of files that do not contain
1942 any matching lines, that is, the complement of -l.
1943
1944 (n) The return code is 2 if any file cannot be opened, but pcregrep does
1945 continue to scan other files.
1946
1947 (o) The -s option was incorrectly implemented. For compatibility with other
1948 greps, it now suppresses the error message for a non-existent or non-
1949 accessible file (but not the return code). There is a new option called
1950 -q that suppresses the output of matching lines, which was what -s was
1951 previously doing.
1952
1953 (p) Added --include and --exclude options to specify files for inclusion
1954 and exclusion when recursing.
1955
1956 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
1957 Hopefully, it now does.
1958
1959 12. Missing cast in pcre_study().
1960
1961 13. Added an "uninstall" target to the makefile.
1962
1963 14. Replaced "extern" in the function prototypes in Makefile.in with
1964 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
1965 world, but is set differently for Windows.
1966
1967 15. Added a second compiling function called pcre_compile2(). The only
1968 difference is that it has an extra argument, which is a pointer to an
1969 integer error code. When there is a compile-time failure, this is set
1970 non-zero, in addition to the error test pointer being set to point to an
1971 error message. The new argument may be NULL if no error number is required
1972 (but then you may as well call pcre_compile(), which is now just a
1973 wrapper). This facility is provided because some applications need a
1974 numeric error indication, but it has also enabled me to tidy up the way
1975 compile-time errors are handled in the POSIX wrapper.
1976
1977 16. Added VPATH=.libs to the makefile; this should help when building with one
1978 prefix path and installing with another. (Or so I'm told by someone who
1979 knows more about this stuff than I do.)
1980
1981 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
1982 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
1983 match everything, including newlines. This is not POSIX-compatible, but
1984 somebody wanted the feature. From pcretest it can be activated by using
1985 both the P and the s flags.
1986
1987 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
1988
1989 19. libpcre.pc was being incorrectly installed as executable.
1990
1991 20. A couple of places in pcretest check for end-of-line by looking for '\n';
1992 it now also looks for '\r' so that it will work unmodified on Windows.
1993
1994 21. Added Google's contributed C++ wrapper to the distribution.
1995
1996 22. Added some untidy missing memory free() calls in pcretest, to keep
1997 Electric Fence happy when testing.
1998
1999
2000
2001 Version 5.0 13-Sep-04
2002 ---------------------
2003
2004 1. Internal change: literal characters are no longer packed up into items
2005 containing multiple characters in a single byte-string. Each character
2006 is now matched using a separate opcode. However, there may be more than one
2007 byte in the character in UTF-8 mode.
2008
2009 2. The pcre_callout_block structure has two new fields: pattern_position and
2010 next_item_length. These contain the offset in the pattern to the next match
2011 item, and its length, respectively.
2012
2013 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
2014 insertion of callouts before each pattern item. Added the /C option to
2015 pcretest to make use of this.
2016
2017 4. On the advice of a Windows user, the lines
2018
2019 #if defined(_WIN32) || defined(WIN32)
2020 _setmode( _fileno( stdout ), 0x8000 );
2021 #endif /* defined(_WIN32) || defined(WIN32) */
2022
2023 have been added to the source of pcretest. This apparently does useful
2024 magic in relation to line terminators.
2025
2026 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
2027 for the benefit of those environments where the "b" makes a difference.
2028
2029 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
2030 to know about it. I have put a hack into configure.in that adds in code
2031 to set GCC=yes if CC=icc. This seems to end up at a point in the
2032 generated configure script that is early enough to affect the setting of
2033 compiler options, which is what is needed, but I have no means of testing
2034 whether it really works. (The user who reported this had patched the
2035 generated configure script, which of course I cannot do.)
2036
2037 LATER: After change 22 below (new libtool files), the configure script
2038 seems to know about icc (and also ecc). Therefore, I have commented out
2039 this hack in configure.in.
2040
2041 7. Added support for pkg-config (2 patches were sent in).
2042
2043 8. Negated POSIX character classes that used a combination of internal tables
2044 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
2045 [[:^ascii]]. Typically, they would match almost any characters. The other
2046 POSIX classes were not broken in this way.
2047
2048 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
2049 to find the match, as PCRE was deluded into thinking that the match had to
2050 start at the start point or following a newline. The same bug applied to
2051 patterns with negative forward assertions or any backward assertions
2052 preceding ".*" at the start, unless the pattern required a fixed first
2053 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
2054
2055 10. In UTF-8 mode, when moving forwards in the subject after a failed match
2056 starting at the last subject character, bytes beyond the end of the subject
2057 string were read.
2058
2059 11. Renamed the variable "class" as "classbits" to make life easier for C++
2060 users. (Previously there was a macro definition, but it apparently wasn't
2061 enough.)
2062
2063 12. Added the new field "tables" to the extra data so that tables can be passed
2064 in at exec time, or the internal tables can be re-selected. This allows
2065 a compiled regex to be saved and re-used at a later time by a different
2066 program that might have everything at different addresses.
2067
2068 13. Modified the pcre-config script so that, when run on Solaris, it shows a
2069 -R library as well as a -L library.
2070
2071 14. The debugging options of pcretest (-d on the command line or D on a
2072 pattern) showed incorrect output for anything following an extended class
2073 that contained multibyte characters and which was followed by a quantifier.
2074
2075 15. Added optional support for general category Unicode character properties
2076 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
2077 support. It adds about 90K to the size of the library. The meanings of the
2078 inbuilt class escapes such as \d and \s have NOT been changed.
2079
2080 16. Updated pcredemo.c to include calls to free() to release the memory for the
2081 compiled pattern.
2082
2083 17. The generated file chartables.c was being created in the source directory
2084 instead of in the building directory. This caused the build to fail if the
2085 source directory was different from the building directory, and was
2086 read-only.
2087
2088 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
2089 file. No doubt somebody will tell me if they don't make sense... Also added
2090 Dan Mooney's comments about building on OpenVMS.
2091
2092 19. Added support for partial matching via the PCRE_PARTIAL option for
2093 pcre_exec() and the \P data escape in pcretest.
2094
2095 20. Extended pcretest with 3 new pattern features:
2096
2097 (i) A pattern option of the form ">rest-of-line" causes pcretest to
2098 write the compiled pattern to the file whose name is "rest-of-line".
2099 This is a straight binary dump of the data, with the saved pointer to
2100 the character tables forced to be NULL. The study data, if any, is
2101 written too. After writing, pcretest reads a new pattern.
2102
2103 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
2104 compiled pattern from the given file. There must not be any
2105 occurrences of "<" in the file name (pretty unlikely); if there are,
2106 pcretest will instead treat the initial "<" as a pattern delimiter.
2107 After reading in the pattern, pcretest goes on to read data lines as
2108 usual.
2109
2110 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
2111 and 16-bit fields in a compiled pattern, to simulate a pattern that
2112 was compiled on a host of opposite endianness.
2113
2114 21. The pcre-exec() function can now cope with patterns that were compiled on
2115 hosts of opposite endianness, with this restriction:
2116
2117 As for any compiled expression that is saved and used later, the tables
2118 pointer field cannot be preserved; the extra_data field in the arguments
2119 to pcre_exec() should be used to pass in a tables address if a value
2120 other than the default internal tables were used at compile time.
2121
2122 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
2123 now diagnosed as an error. Previously, most of the time, a negative number
2124 would have been treated as zero, but if in addition "ovector" was passed as
2125 NULL, a crash could occur.
2126
2127 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
2128 new versions from the libtool 1.5 distribution (the last one is a copy of
2129 a file called libtool.m4). This seems to have fixed the need to patch
2130 "configure" to support Darwin 1.3 (which I used to do). However, I still
2131 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
2132 workstation).
2133
2134 24. Changed the PCRE licence to be the more standard "BSD" licence.
2135
2136
2137 Version 4.5 01-Dec-03
2138 ---------------------
2139
2140 1. There has been some re-arrangement of the code for the match() function so
2141 that it can be compiled in a version that does not call itself recursively.
2142 Instead, it keeps those local variables that need separate instances for
2143 each "recursion" in a frame on the heap, and gets/frees frames whenever it
2144 needs to "recurse". Keeping track of where control must go is done by means
2145 of setjmp/longjmp. The whole thing is implemented by a set of macros that
2146 hide most of the details from the main code, and operates only if
2147 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
2148 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
2149 operating.
2150
2151 To make it easier for callers to provide specially tailored get/free
2152 functions for this usage, two new functions, pcre_stack_malloc, and
2153 pcre_stack_free, are used. They are always called in strict stacking order,
2154 and the size of block requested is always the same.
2155
2156 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
2157 PCRE has been compiled to use the stack or the heap for recursion. The
2158 -C option of pcretest uses this to show which version is compiled.
2159
2160 A new data escape \S, is added to pcretest; it causes the amounts of store
2161 obtained and freed by both kinds of malloc/free at match time to be added
2162 to the output.
2163
2164 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
2165 what's available on my current Linux desktop machine.
2166
2167 3. When matching a UTF-8 string, the test for a valid string at the start has
2168 been extended. If start_offset is not zero, PCRE now checks that it points
2169 to a byte that is the start of a UTF-8 character. If not, it returns
2170 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
2171 this is necessary because there may be backward assertions in the pattern.
2172 When matching the same subject several times, it may save resources to use
2173 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
2174
2175 4. The code for checking the validity of UTF-8 strings has been tightened so
2176 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
2177 containing "overlong sequences".
2178
2179 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
2180 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
2181 should have been "&", but it just so happened that all the cases this let
2182 through by mistake were picked up later in the function.
2183
2184 6. I had used a variable called "isblank" - this is a C99 function, causing
2185 some compilers to warn. To avoid this, I renamed it (as "blankclass").
2186
2187 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
2188 prompting; (b) run "./pcretest /dev/null" at the start of the test script
2189 so the version is shown; (c) stop "make test" echoing "./RunTest".
2190
2191 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
2192
2193 9. The prototype for memmove() for systems that don't have it was using
2194 size_t, but the inclusion of the header that defines size_t was later. I've
2195 moved the #includes for the C headers earlier to avoid this.
2196
2197 10. Added some adjustments to the code to make it easier to compiler on certain
2198 special systems:
2199
2200 (a) Some "const" qualifiers were missing.
2201 (b) Added the macro EXPORT before all exported functions; by default this
2202 is defined to be empty.
2203 (c) Changed the dftables auxiliary program (that builds chartables.c) so
2204 that it reads its output file name as an argument instead of writing
2205 to the standard output and assuming this can be redirected.
2206
2207 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
2208 class containing characters with values greater than 255, PCRE compilation
2209 went into a loop.
2210
2211 12. A recursive reference to a subpattern that was within another subpattern
2212 that had a minimum quantifier of zero caused PCRE to crash. For example,
2213 (x(y(?2))z)? provoked this bug with a subject that got as far as the
2214 recursion. If the recursively-called subpattern itself had a zero repeat,
2215 that was OK.
2216
2217 13. In pcretest, the buffer for reading a data line was set at 30K, but the
2218 buffer into which it was copied (for escape processing) was still set at
2219 1024, so long lines caused crashes.
2220
2221 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
2222 "internal error: code overflow...". This applied to any character class
2223 that was followed by a possessive quantifier.
2224
2225 15. Modified the Makefile to add libpcre.la as a prerequisite for
2226 libpcreposix.la because I was told this is needed for a parallel build to
2227 work.
2228
2229 16. If a pattern that contained .* following optional items at the start was
2230 studied, the wrong optimizing data was generated, leading to matching
2231 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
2232 matching string must start with a or b or c. The correct conclusion for
2233 this pattern is that a match can start with any character.
2234
2235
2236 Version 4.4 13-Aug-03
2237 ---------------------
2238
2239 1. In UTF-8 mode, a character class containing characters with values between
2240 127 and 255 was not handled correctly if the compiled pattern was studied.
2241 In fixing this, I have also improved the studying algorithm for such
2242 classes (slightly).
2243
2244 2. Three internal functions had redundant arguments passed to them. Removal
2245 might give a very teeny performance improvement.
2246
2247 3. Documentation bug: the value of the capture_top field in a callout is *one
2248 more than* the number of the hightest numbered captured substring.
2249
2250 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
2251 in incorrectly linking with a previously installed version. They now link
2252 explicitly with libpcre.la.
2253
2254 5. configure.in no longer needs to recognize Cygwin specially.
2255
2256 6. A problem in pcre.in for Windows platforms is fixed.
2257
2258 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
2259 pcretest, it used to include the size of the study block as part of its
2260 output. Unfortunately, the structure contains a field that has a different
2261 size on different hardware architectures. This meant that the tests that
2262 showed this size failed. As the block is currently always of a fixed size,
2263 this information isn't actually particularly useful in pcretest output, so
2264 I have just removed it.
2265
2266 8. Three pre-processor statements accidentally did not start in column 1.
2267 Sadly, there are *still* compilers around that complain, even though
2268 standard C has not required this for well over a decade. Sigh.
2269
2270 9. In pcretest, the code for checking callouts passed small integers in the
2271 callout_data field, which is a void * field. However, some picky compilers
2272 complained about the casts involved for this on 64-bit systems. Now
2273 pcretest passes the address of the small integer instead, which should get
2274 rid of the warnings.
2275
2276 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
2277 both compile and run time, and gives an error if an invalid UTF-8 sequence
2278 is found. There is a option for disabling this check in cases where the
2279 string is known to be correct and/or the maximum performance is wanted.
2280
2281 11. In response to a bug report, I changed one line in Makefile.in from
2282
2283 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
2284 to
2285 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
2286
2287 to look similar to other lines, but I have no way of telling whether this
2288 is the right thing to do, as I do not use Windows. No doubt I'll get told
2289 if it's wrong...
2290
2291
2292 Version 4.3 21-May-03
2293 ---------------------
2294
2295 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
2296 Makefile.
2297
2298 2. Some refactoring to improve the quality of the code:
2299
2300 (i) The utf8_table... variables are now declared "const".
2301
2302 (ii) The code for \cx, which used the "case flipping" table to upper case
2303 lower case letters, now just substracts 32. This is ASCII-specific,
2304 but the whole concept of \cx is ASCII-specific, so it seems
2305 reasonable.
2306
2307 (iii) PCRE was using its character types table to recognize decimal and
2308 hexadecimal digits in the pattern. This is silly, because it handles
2309 only 0-9, a-f, and A-F, but the character types table is locale-
2310 specific, which means strange things might happen. A private
2311 table is now used for this - though it costs 256 bytes, a table is
2312 much faster than multiple explicit tests. Of course, the standard
2313 character types table is still used for matching digits in subject
2314 strings against \d.
2315
2316 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
2317 ending in _t are). So I've renamed it as ESC_tee.
2318
2319 3. The first argument for regexec() in the POSIX wrapper should have been
2320 defined as "const".
2321
2322 4. Changed pcretest to use malloc() for its buffers so that they can be
2323 Electric Fenced for debugging.
2324
2325 5. There were several places in the code where, in UTF-8 mode, PCRE would try
2326 to read one or more bytes before the start of the subject string. Often this
2327 had no effect on PCRE's behaviour, but in some circumstances it could
2328 provoke a segmentation fault.
2329
2330 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
2331 to try to read one or more bytes before the start of the subject string.
2332
2333 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
2334 UTF-8 support could misbehave in various ways if the subject string
2335 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
2336 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
2337 back over UTF-8 characters.)
2338
2339
2340 Version 4.2 14-Apr-03
2341 ---------------------
2342
2343 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
2344
2345 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
2346 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
2347 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
2348 [WIN_PREFIX]: new variable, "cyg" for Cygwin
2349 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
2350 and BUILD_EXEEXT
2351 Note: automatic setting of the BUILD variables is not yet working
2352 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
2353 compile-time but not at link-time
2354 [LINK]: use for linking executables only
2355 make different versions for Windows and non-Windows
2356 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
2357 libraries
2358 [LINK_FOR_BUILD]: new variable
2359 [OBJEXT]: use throughout
2360 [EXEEXT]: use throughout
2361 <winshared>: new target
2362 <wininstall>: new target
2363 <dftables.o>: use native compiler
2364 <dftables>: use native linker
2365 <install>: handle Windows platform correctly
2366 <clean>: ditto
2367 <check>: ditto
2368 copy DLL to top builddir before testing
2369
2370 As part of these changes, -no-undefined was removed again. This was reported
2371 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
2372 in any case.
2373
2374 3. Some tidies to get rid of compiler warnings:
2375
2376 . In the match_data structure, match_limit was an unsigned long int, whereas
2377 match_call_count was an int. I've made them both unsigned long ints.
2378
2379 . In pcretest the fact that a const uschar * doesn't automatically cast to
2380 a void * provoked a warning.
2381
2382 . Turning on some more compiler warnings threw up some "shadow" variables
2383 and a few more missing casts.
2384
2385 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
2386 option, a class that contained a single character with a value between 128
2387 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
2388
2389 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
2390 option, a class that contained several characters, but with at least one
2391 whose value was between 128 and 255 caused PCRE to crash.
2392
2393
2394 Version 4.1 12-Mar-03
2395 ---------------------
2396
2397 1. Compiling with gcc -pedantic found a couple of places where casts were
2398 needed, and a string in dftables.c that was longer than standard compilers are
2399 required to support.
2400
2401 2. Compiling with Sun's compiler found a few more places where the code could
2402 be tidied up in order to avoid warnings.
2403
2404 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
2405 first of these names is deprecated in the latest Autoconf in favour of the name
2406 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
2407 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
2408 analogy I have changed it to CFLAGS_FOR_BUILD.
2409
2410 4. Added -no-undefined to the linking command in the Makefile, because this is
2411 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
2412 linking step for the pcreposix library.
2413
2414 5. PCRE was failing to diagnose the case of two named groups with the same
2415 name.
2416
2417 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
2418 literal character that is needed in the subject for a match, and scans along to
2419 ensure that it is present before embarking on the full matching process. This
2420 saves time in cases of nested unlimited repeats that are never going to match.
2421 Problem: the scan can take a lot of time if the subject is very long (e.g.
2422 megabytes), thus penalizing straightforward matches. It is now done only if the
2423 amount of subject to be scanned is less than 1000 bytes.
2424
2425 7. A lesser problem with the same optimization is that it was recording the
2426 first character of an anchored pattern as "needed", thus provoking a search
2427 right along the subject, even when the first match of the pattern was going to
2428 fail. The "needed" character is now not set for anchored patterns, unless it
2429 follows something in the pattern that is of non-fixed length. Thus, it still
2430 fulfils its original purpose of finding quick non-matches in cases of nested
2431 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
2432
2433
2434 Version 4.0 17-Feb-03
2435 ---------------------
2436
2437 1. If a comment in an extended regex that started immediately after a meta-item
2438 extended to the end of string, PCRE compiled incorrect data. This could lead to
2439 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
2440
2441 2. Moved to autoconf 2.53 and libtool 1.4.2.
2442
2443 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
2444 the special perltest8 script is no longer needed - all the tests can be run
2445 from a single perltest script.
2446
2447 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
2448 by \s. It has now been removed in PCRE. This means it isn't recognized as
2449 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
2450 class [:space:] *does* include VT, thereby creating a mess.
2451
2452 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
2453 space and tab.
2454
2455 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
2456 its new features into the main test script, reducing the number of scripts.
2457
2458 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
2459 were backward compatible, and made the (?i) apply to the whole pattern, as if
2460 /i were given. Now it behaves more logically, and applies the option setting
2461 only to what follows. PCRE has been changed to follow suit. However, if it
2462 finds options settings right at the start of the pattern, it extracts them into
2463 the global options, as before. Thus, they show up in the info data.
2464
2465 8. Added support for the \Q...\E escape sequence. Characters in between are
2466 treated as literals. This is slightly different from Perl in that $ and @ are
2467 also handled as literals inside the quotes. In Perl, they will cause variable
2468 interpolation. Note the following examples:
2469
2470 Pattern PCRE matches Perl matches
2471
2472 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
2473 \Qabc\$xyz\E abc\$xyz abc\$xyz
2474 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2475
2476 For compatibility with Perl, \Q...\E sequences are recognized inside character
2477 classes as well as outside them.
2478
2479 9. Re-organized 3 code statements in pcretest to avoid "overflow in
2480 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
2481 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
2482 signed/unsigned warnings.
2483
2484 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
2485 option for pcretest, so I've replaced it by a simple function that does just
2486 that job.
2487
2488 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
2489 "pcregrep -".
2490
2491 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
2492 Java package. This provides some syntactic sugar for simple cases of what my
2493 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
2494 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
2495 item, you can use this simplified notation. Note that only makes sense with
2496 greedy quantifiers. Consequently, the use of the possessive quantifier forces
2497 greediness, whatever the setting of the PCRE_UNGREEDY option.
2498
2499 13. A change of greediness default within a pattern was not taking effect at
2500 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
2501 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
2502 was abstracted outside.
2503
2504 14. PCRE now supports the \G assertion. It is true when the current matching
2505 position is at the start point of the match. This differs from \A when the
2506 starting offset is non-zero. Used with the /g option of pcretest (or similar
2507 code), it works in the same way as it does for Perl's /g option. If all
2508 alternatives of a regex begin with \G, the expression is anchored to the start
2509 match position, and the "anchored" flag is set in the compiled expression.
2510
2511 15. Some bugs concerning the handling of certain option changes within patterns
2512 have been fixed. These applied to options other than (?ims). For example,
2513 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
2514 been the other way round. Some of this was related to change 7 above.
2515
2516 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
2517 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
2518 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
2519 POSIX classes only within a class (e.g. /[[:alpha:]]/).
2520
2521 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
2522 mode. Unlike ".", it always matches newline, whatever the setting of
2523 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
2524 assertions. Perl allows it, but it doesn't (in general) work because it can't
2525 calculate the length of the lookbehind. At least, that's the case for Perl
2526 5.8.0 - I've been told they are going to document that it doesn't work in
2527 future.
2528
2529 18. Added an error diagnosis for escapes that PCRE does not support: these are
2530 \L, \l, \N, \P, \p, \U, \u, and \X.
2531
2532 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
2533 reading past the end of the pattern in cases such as /[abcd/.
2534
2535 20. PCRE was getting more memory than necessary for patterns with classes that
2536 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
2537
2538 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
2539 compiling PCRE for use with Virtual Pascal.
2540
2541 22. Small fix to the Makefile to make it work properly if the build is done
2542 outside the source tree.
2543
2544 23. Added a new extension: a condition to go with recursion. If a conditional
2545 subpattern starts with (?(R) the "true" branch is used if recursion has
2546 happened, whereas the "false" branch is used only at the top level.
2547
2548 24. When there was a very long string of literal characters (over 255 bytes
2549 without UTF support, over 250 bytes with UTF support), the computation of how
2550 much memory was required could be incorrect, leading to segfaults or other
2551 strange effects.
2552
2553 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
2554 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
2555 there was a subsequent back reference to those brackets. This meant that, for
2556 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
2557 possible to check for precisely this case. All we can do is abandon the
2558 optimization if .* occurs inside capturing brackets when there are any back
2559 references whatsoever. (See below for a better fix that came later.)
2560
2561 26. The handling of the optimization for finding the first character of a
2562 non-anchored pattern, and for finding a character that is required later in the
2563 match were failing in some cases. This didn't break the matching; it just
2564 failed to optimize when it could. The way this is done has been re-implemented.
2565
2566 27. Fixed typo in error message for invalid (?R item (it said "(?p").
2567
2568 28. Added a new feature that provides some of the functionality that Perl
2569 provides with (?{...}). The facility is termed a "callout". The way it is done
2570 in PCRE is for the caller to provide an optional function, by setting
2571 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
2572 global variable. By default it is unset, which disables all calling out. To get
2573 the function called, the regex must include (?C) at appropriate points. This
2574 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
2575 This provides a means of identifying different callout points. When PCRE
2576 reaches such a point in the regex, if pcre_callout has been set, the external
2577 function is called. It is provided with data in a structure called
2578 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
2579 matching continues; if it returns a non-zero value, the match at the current
2580 point fails. However, backtracking will occur if possible. [This was changed
2581 later and other features added - see item 49 below.]
2582
2583 29. pcretest is upgraded to test the callout functionality. It provides a
2584 callout function that displays information. By default, it shows the start of
2585 the match and the current position in the text. There are some new data escapes
2586 to vary what happens:
2587
2588 \C+ in addition, show current contents of captured substrings
2589 \C- do not supply a callout function
2590 \C!n return 1 when callout number n is reached
2591 \C!n!m return 1 when callout number n is reached for the mth time
2592
2593 30. If pcregrep was called with the -l option and just a single file name, it
2594 output "<stdin>" if a match was found, instead of the file name.
2595
2596 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
2597 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
2598 pcre_exec(). This saves a malloc/free per call. The default value of
2599 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
2600 when configuring.
2601
2602 32. The default maximum size of a compiled pattern is 64K. There have been a
2603 few cases of people hitting this limit. The code now uses macros to handle the
2604 storing of links as offsets within the compiled pattern. It defaults to 2-byte
2605 links, but this can be changed to 3 or 4 bytes by --with-link-size when
2606 configuring. Tests 2 and 5 work only with 2-byte links because they output
2607 debugging information about compiled patterns.
2608
2609 33. Internal code re-arrangements:
2610
2611 (a) Moved the debugging function for printing out a compiled regex into
2612 its own source file (printint.c) and used #include to pull it into
2613 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
2614 separate copies.
2615
2616 (b) Defined the list of op-code names for debugging as a macro in
2617 internal.h so that it is next to the definition of the opcodes.
2618
2619 (c) Defined a table of op-code lengths for simpler skipping along compiled
2620 code. This is again a macro in internal.h so that it is next to the
2621 definition of the opcodes.
2622
2623 34. Added support for recursive calls to individual subpatterns, along the
2624 lines of Robin Houston's patch (but implemented somewhat differently).
2625
2626 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
2627 allow it to read and process whole directories in Win32. This code was
2628 contributed by Lionel Fourquaux; it has not been tested by me.
2629
2630 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
2631 used to name a group. Names consist of alphanumerics and underscores, and must
2632 be unique. Back references use the syntax (?P=name) and recursive calls use
2633 (?P>name) which is a PCRE extension to the Python extension. Groups still have
2634 numbers. The function pcre_fullinfo() can be used after compilation to extract
2635 a name/number map. There are three relevant calls:
2636
2637 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
2638 PCRE_INFO_NAMECOUNT yields the number of entries
2639 PCRE_INFO_NAMETABLE yields a pointer to the map.
2640
2641 The map is a vector of fixed-size entries. The size of each entry depends on
2642 the length of the longest name used. The first two bytes of each entry are the
2643 group number, most significant byte first. There follows the corresponding
2644 name, zero terminated. The names are in alphabetical order.
2645
2646 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
2647 case instead of 255. Making it the same both with and without UTF-8 support
2648 means that the same test output works with both.
2649
2650 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
2651 calling malloc() with a zero argument.
2652
2653 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
2654 optimization. I've improved things by keeping a bitmap of backreferences with
2655 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
2656 fact referenced, the optimization can be applied. It is unlikely that a
2657 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
2658 the match to follow \n) will appear inside brackets with a number greater than
2659 31, but if it does, any back reference > 31 suppresses the optimization.
2660
2661 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
2662 of disabling numbered capturing parentheses. Any opening parenthesis that is
2663 not followed by ? behaves as if it were followed by ?: but named parentheses
2664 can still be used for capturing (and they will acquire numbers in the usual
2665 way).
2666
2667 41. Redesigned the return codes from the match() function into yes/no/error so
2668 that errors can be passed back from deep inside the nested calls. A malloc
2669 failure while inside a recursive subpattern call now causes the
2670 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
2671
2672 42. It is now possible to set a limit on the number of times the match()
2673 function is called in a call to pcre_exec(). This facility makes it possible to
2674 limit the amount of recursion and backtracking, though not in a directly
2675 obvious way, because the match() function is used in a number of different
2676 circumstances. The count starts from zero for each position in the subject
2677 string (for non-anchored patterns). The default limit is, for compatibility, a
2678 large number, namely 10 000 000. You can change this in two ways:
2679
2680 (a) When configuring PCRE before making, you can use --with-match-limit=n
2681 to set a default value for the compiled library.
2682
2683 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
2684 a different value is set. See 45 below.
2685
2686 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
2687
2688 43. Added a new function pcre_config(int, void *) to enable run-time extraction
2689 of things that can be changed at compile time. The first argument specifies
2690 what is wanted and the second points to where the information is to be placed.
2691 The current list of available information is:
2692
2693 PCRE_CONFIG_UTF8
2694
2695 The output is an integer that is set to one if UTF-8 support is available;
2696 otherwise it is set to zero.
2697
2698 PCRE_CONFIG_NEWLINE
2699
2700 The output is an integer that it set to the value of the code that is used for
2701 newline. It is either LF (10) or CR (13).
2702
2703 PCRE_CONFIG_LINK_SIZE
2704
2705 The output is an integer that contains the number of bytes used for internal
2706 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
2707
2708 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
2709
2710 The output is an integer that contains the threshold above which the POSIX
2711 interface uses malloc() for output vectors. See item 31 above.
2712
2713 PCRE_CONFIG_MATCH_LIMIT
2714
2715 The output is an unsigned integer that contains the default limit of the number
2716 of match() calls in a pcre_exec() execution. See 42 above.
2717
2718 44. pcretest has been upgraded by the addition of the -C option. This causes it
2719 to extract all the available output from the new pcre_config() function, and to
2720 output it. The program then exits immediately.
2721
2722 45. A need has arisen to pass over additional data with calls to pcre_exec() in
2723 order to support additional features. One way would have been to define
2724 pcre_exec2() (for example) with extra arguments, but this would not have been
2725 extensible, and would also have required all calls to the original function to
2726 be mapped to the new one. Instead, I have chosen to extend the mechanism that
2727 is used for passing in "extra" data from pcre_study().
2728
2729 The pcre_extra structure is now exposed and defined in pcre.h. It currently
2730 contains the following fields:
2731
2732 flags a bitmap indicating which of the following fields are set
2733 study_data opaque data from pcre_study()
2734 match_limit a way of specifying a limit on match() calls for a specific
2735 call to pcre_exec()
2736 callout_data data for callouts (see 49 below)
2737
2738 The flag bits are also defined in pcre.h, and are
2739
2740 PCRE_EXTRA_STUDY_DATA
2741 PCRE_EXTRA_MATCH_LIMIT
2742 PCRE_EXTRA_CALLOUT_DATA
2743
2744 The pcre_study() function now returns one of these new pcre_extra blocks, with
2745 the actual study data pointed to by the study_data field, and the
2746 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
2747 before. That is, this change is entirely upwards-compatible and requires no
2748 change to existing code.
2749
2750 If you want to pass in additional data to pcre_exec(), you can either place it
2751 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
2752 block.
2753
2754 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
2755 data string contains the escape sequence \M, pcretest calls pcre_exec() several
2756 times with different match limits, until it finds the minimum value needed for
2757 pcre_exec() to complete. The value is then output. This can be instructive; for
2758 most simple matches the number is quite small, but for pathological cases it
2759 gets very large very quickly.
2760
2761 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
2762 returns the size of the data block pointed to by the study_data field in a
2763 pcre_extra block, that is, the value that was passed as the argument to
2764 pcre_malloc() when PCRE was getting memory in which to place the information
2765 created by pcre_study(). The fourth argument should point to a size_t variable.
2766 pcretest has been extended so that this information is shown after a successful
2767 pcre_study() call when information about the compiled regex is being displayed.
2768
2769 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
2770 because what follows is always an absolute path. (Later: it turns out that this
2771 is more than cosmetic for MinGW, because it doesn't like empty path
2772 components.)
2773
2774 49. Some changes have been made to the callout feature (see 28 above):
2775
2776 (i) A callout function now has three choices for what it returns:
2777
2778 0 => success, carry on matching
2779 > 0 => failure at this point, but backtrack if possible
2780 < 0 => serious error, return this value from pcre_exec()
2781
2782 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
2783 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
2784 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
2785 use by callout functions. It will never be used by PCRE itself.
2786
2787 (ii) The pcre_extra structure (see 45 above) has a void * field called
2788 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
2789 pcre_callout_block structure has a field of the same name. The contents of
2790 the field passed in the pcre_extra structure are passed to the callout
2791 function in the corresponding field in the callout block. This makes it
2792 easier to use the same callout-containing regex from multiple threads. For
2793 testing, the pcretest program has a new data escape
2794
2795 \C*n pass the number n (may be negative) as callout_data
2796
2797 If the callout function in pcretest receives a non-zero value as
2798 callout_data, it returns that value.
2799
2800 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
2801 there were some redundant $(CFLAGS) in commands that are now specified as
2802 $(LINK), which already includes $(CFLAGS).
2803
2804 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
2805 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
2806 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
2807 one-byte characters throughout. Note that case-insensitive matching applies
2808 only to characters whose values are less than 256. PCRE doesn't support the
2809 notion of cases for higher-valued characters.
2810
2811 (i) A character class whose characters are all within 0-255 is handled as
2812 a bit map, and the map is inverted for negative classes. Previously, a
2813 character > 255 always failed to match such a class; however it should
2814 match if the class was a negative one (e.g. [^ab]). This has been fixed.
2815
2816 (ii) A negated character class with a single character < 255 is coded as
2817 "not this character" (OP_NOT). This wasn't working properly when the test
2818 character was multibyte, either singly or repeated.
2819
2820 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
2821 mode, for example: \x{100}{2,3}.
2822
2823 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
2824 singly or repeated) now correctly test multibyte characters. However,
2825 PCRE doesn't recognize any characters with values greater than 255 as
2826 digits, spaces, or word characters. Such characters always match \D, \S,
2827 and \W, and never match \d, \s, or \w.
2828
2829 (v) Classes may now contain characters and character ranges with values
2830 greater than 255. For example: [ab\x{100}-\x{400}].
2831
2832 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
2833 PCRE in UTF-8 mode.
2834
2835 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
2836 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
2837 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
2838 value.)
2839
2840 53. The single man page has become too large. I have therefore split it up into
2841 a number of separate man pages. These also give rise to individual HTML pages;
2842 these are now put in a separate directory, and there is an index.html page that
2843 lists them all. Some hyperlinking between the pages has been installed.
2844
2845 54. Added convenience functions for handling named capturing parentheses.
2846
2847 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
2848 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
2849 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
2850 are faulted.
2851
2852 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
2853 calling configure. These values are used when compiling the dftables.c program
2854 which is run to generate the source of the default character tables. They
2855 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
2856 you will need to set these values.
2857
2858 57. Updated the building process for Windows DLL, as provided by Fred Cox.
2859
2860
2861 Version 3.9 02-Jan-02
2862 ---------------------
2863
2864 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
2865
2866 2. If --disable-static was given, the building process failed when trying to
2867 build pcretest and pcregrep. (For some reason it was using libtool to compile
2868 them, which is not right, as they aren't part of the library.)
2869
2870
2871 Version 3.8 18-Dec-01
2872 ---------------------
2873
2874 1. The experimental UTF-8 code was completely screwed up. It was packing the
2875 bytes in the wrong order. How dumb can you get?
2876
2877
2878 Version 3.7 29-Oct-01
2879 ---------------------
2880
2881 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
2882 This caused pcretest, when used on the test data, to segfault. Unfortunately,
2883 this didn't happen under Solaris 8, where I normally test things.
2884
2885 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
2886 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
2887 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
2888 this fix an hour or so after the initial 3.7 release.)
2889
2890
2891 Version 3.6 23-Oct-01
2892 ---------------------
2893
2894 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
2895 offsets passed as NULL with zero offset count.
2896
2897 2. The config.guess and config.sub files had not been updated when I moved to
2898 the latest autoconf.
2899
2900
2901 Version 3.5 15-Aug-01
2902 ---------------------
2903
2904 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
2905 had been forgotten.
2906
2907 2. By using declared but undefined structures, we can avoid using "void"
2908 definitions in pcre.h while keeping the internal definitions of the structures
2909 private.
2910
2911 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
2912 user point of view, this means that both static and shared libraries are built
2913 by default, but this can be individually controlled. More of the work of
2914 handling this static/shared cases is now inside libtool instead of PCRE's make
2915 file.
2916
2917 4. The pcretest utility is now installed along with pcregrep because it is
2918 useful for users (to test regexs) and by doing this, it automatically gets
2919 relinked by libtool. The documentation has been turned into a man page, so
2920 there are now .1, .txt, and .html versions in /doc.
2921
2922 5. Upgrades to pcregrep:
2923 (i) Added long-form option names like gnu grep.
2924 (ii) Added --help to list all options with an explanatory phrase.
2925 (iii) Added -r, --recursive to recurse into sub-directories.
2926 (iv) Added -f, --file to read patterns from a file.
2927
2928 6. pcre_exec() was referring to its "code" argument before testing that
2929 argument for NULL (and giving an error if it was NULL).
2930
2931 7. Upgraded Makefile.in to allow for compiling in a different directory from
2932 the source directory.
2933
2934 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
2935 options bits, the pointer it was passed was to an int instead of to an unsigned
2936 long int. This mattered only on 64-bit systems.
2937
2938 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
2939 generated) instead of pcre.in, which it its source. Also made the same change
2940 in several of the .c files.
2941
2942 10. A new release of gcc defines printf() as a macro, which broke pcretest
2943 because it had an ifdef in the middle of a string argument for printf(). Fixed
2944 by using separate calls to printf().
2945
2946 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
2947 script, to force use of CR or LF instead of \n in the source. On non-Unix
2948 systems, the value can be set in config.h.
2949
2950 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
2951 absolute limit. Changed the text of the error message to make this clear, and
2952 likewise updated the man page.
2953
2954 13. The limit of 99 on the number of capturing subpatterns has been removed.
2955 The new limit is 65535, which I hope will not be a "real" limit.
2956
2957
2958 Version 3.4 22-Aug-00
2959 ---------------------
2960
2961 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
2962
2963 2. Diagnose condition (?(0) as an error instead of crashing on matching.
2964
2965
2966 Version 3.3 01-Aug-00
2967 ---------------------
2968
2969 1. If an octal character was given, but the value was greater than \377, it
2970 was not getting masked to the least significant bits, as documented. This could
2971 lead to crashes in some systems.
2972
2973 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
2974 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
2975
2976 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
2977 These just pass their arguments on to (pcre_free)(), but they are provided
2978 because some uses of PCRE bind it to non-C systems that can call its functions,
2979 but cannot call free() or pcre_free() directly.
2980
2981 4. Add "make test" as a synonym for "make check". Corrected some comments in
2982 the Makefile.
2983
2984 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
2985 Makefile.
2986
2987 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
2988 command called pgrep for grepping around the active processes.
2989
2990 7. Added the beginnings of support for UTF-8 character strings.
2991
2992 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
2993 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
2994 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
2995 out for the ar command.)
2996
2997
2998 Version 3.2 12-May-00
2999 ---------------------
3000
3001 This is purely a bug fixing release.
3002
3003 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
3004 of ZA. This was just one example of several cases that could provoke this bug,
3005 which was introduced by change 9 of version 2.00. The code for breaking
3006 infinite loops after an iteration that matches an empty string was't working
3007 correctly.
3008
3009 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
3010 when matched against abbab (for example). After matching an empty string, it
3011 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
3012 caused it to match further down the string than it should.
3013
3014 3. The code contained an inclusion of sys/types.h. It isn't clear why this
3015 was there because it doesn't seem to be needed, and it causes trouble on some
3016 systems, as it is not a Standard C header. It has been removed.
3017
3018 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
3019 were reported on the Macintosh. The changes were from
3020
3021 while ((c = *(++ptr)) != 0 && c != '\n');
3022 to
3023 while ((c = *(++ptr)) != 0 && c != '\n') ;
3024
3025 Totally extraordinary, but if that's what it takes...
3026
3027 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
3028 available. Added HAVE_BCOPY and an autoconf test for it; if neither
3029 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
3030 assumes the way PCRE uses memmove() (always moving upwards).
3031
3032 6. PCRE is being used in one environment where strchr() is not available. There
3033 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
3034 faster code anyway.
3035
3036
3037 Version 3.1 09-Feb-00
3038 ---------------------
3039
3040 The only change in this release is the fixing of some bugs in Makefile.in for
3041 the "install" target:
3042
3043 (1) It was failing to install pcreposix.h.
3044
3045 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
3046
3047
3048 Version 3.0 01-Feb-00
3049 ---------------------
3050
3051 1. Add support for the /+ modifier to perltest (to output $` like it does in
3052 pcretest).
3053
3054 2. Add support for the /g modifier to perltest.
3055
3056 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
3057 matches null strings.
3058
3059 4. Fix perltest so that it doesn't do unwanted things when fed an empty
3060 pattern. Perl treats empty patterns specially - it reuses the most recent
3061 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
3062 effect.
3063
3064 5. The POSIX interface was broken in that it was just handing over the POSIX
3065 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
3066 required a bigger vector, with some working space on the end. This means that
3067 the POSIX wrapper now has to get and free some memory, and copy the results.
3068
3069 6. Added some simple autoconf support, placing the test data and the
3070 documentation in separate directories, re-organizing some of the
3071 information files, and making it build pcre-config (a GNU standard). Also added
3072 libtool support for building PCRE as a shared library, which is now the
3073 default.
3074
3075 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
3076 09 are not valid octal constants. Single digits will be used for minor values
3077 less than 10.
3078
3079 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
3080 existing programs that set these in the POSIX interface can use PCRE without
3081 modification.
3082
3083 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
3084 return all that pcre_info() returns, plus additional data. The pcre_info()
3085 function is retained for compatibility, but is considered to be obsolete.
3086
3087 10. Added experimental recursion feature (?R) to handle one common case that
3088 Perl 5.6 will be able to do with (?p{...}).
3089
3090 11. Added support for POSIX character classes like [:alpha:], which Perl is
3091 adopting.
3092
3093
3094 Version 2.08 31-Aug-99
3095 ----------------------
3096
3097 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
3098 trying to match at the startoffset position, but instead was moving forward to
3099 the next newline as if a previous match had failed.
3100
3101 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
3102 and could get into a loop if a null string was matched other than at the start
3103 of the subject.
3104
3105 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
3106 be distinguished at compile time, and for completeness also added PCRE_DATE.
3107
3108 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
3109 in GnuWin32 environments.
3110
3111
3112 Version 2.07 29-Jul-99
3113 ----------------------
3114
3115 1. The documentation is now supplied in plain text form and HTML as well as in
3116 the form of man page sources.
3117
3118 2. C++ compilers don't like assigning (void *) values to other pointer types.
3119 In particular this affects malloc(). Although there is no problem in Standard
3120 C, I've put in casts to keep C++ compilers happy.
3121
3122 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
3123 should be (const char *).
3124
3125 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
3126 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
3127 However, I haven't made this a standard facility. The documentation doesn't
3128 mention it, and the Makefile doesn't support it.
3129
3130 5. The Makefile now contains an "install" target, with editable destinations at
3131 the top of the file. The pcretest program is not installed.
3132
3133 6. pgrep -V now gives the PCRE version number and date.
3134
3135 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
3136 causing the entire string to be ignored, instead of just the last character.
3137
3138 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
3139 non-matching string, it can take a very, very long time, even for strings of
3140 quite modest length, because of the nested recursion. PCRE now does better in
3141 some of these cases. It does this by remembering the last required literal
3142 character in the pattern, and pre-searching the subject to ensure it is present
3143 before running the real match. In other words, it applies a heuristic to detect
3144 some types of certain failure quickly, and in the above example, if presented
3145 with a string that has no trailing " it gives "no match" very quickly.
3146
3147 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
3148 other alternatives are tried instead.
3149
3150
3151 Version 2.06 09-Jun-99
3152 ----------------------
3153
3154 1. Change pcretest's output for amount of store used to show just the code
3155 space, because the remainder (the data block) varies in size between 32-bit and
3156 64-bit systems.
3157
3158 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
3159 start matching at. This allows lookbehinds to work when searching for multiple
3160 occurrences in a string.
3161
3162 3. Added additional options to pcretest for testing multiple occurrences:
3163
3164 /+ outputs the rest of the string that follows a match
3165 /g loops for multiple occurrences, using the new startoffset argument
3166 /G loops for multiple occurrences by passing an incremented pointer
3167
3168 4. PCRE wasn't doing the "first character" optimization for patterns starting
3169 with \b or \B, though it was doing it for other lookbehind assertions. That is,
3170 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
3171 the letter 'x'. On long subject strings, this gives a significant speed-up.
3172
3173
3174 Version 2.05 21-Apr-99
3175 ----------------------
3176
3177 1. Changed the type of magic_number from int to long int so that it works
3178 properly on 16-bit systems.
3179
3180 2. Fixed a bug which caused patterns starting with .* not to work correctly
3181 when the subject string contained newline characters. PCRE was assuming
3182 anchoring for such patterns in all cases, which is not correct because .* will
3183 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
3184 DOTALL is set at top level; otherwise it knows that patterns starting with .*
3185 must be retried after every newline in the subject.
3186
3187
3188 Version 2.04 18-Feb-99
3189 ----------------------
3190
3191 1. For parenthesized subpatterns with repeats whose minimum was zero, the
3192 computation of the store needed to hold the pattern was incorrect (too large).
3193 If such patterns were nested a few deep, this could multiply and become a real
3194 problem.
3195
3196 2. Added /M option to pcretest to show the memory requirement of a specific
3197 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
3198
3199 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
3200 compiled in such a way that the backtracking after subsequent failure was
3201 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
3202 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
3203
3204
3205 Version 2.03 02-Feb-99
3206 ----------------------
3207
3208 1. Fixed typo and small mistake in man page.
3209
3210 2. Added 4th condition (GPL supersedes if conflict) and created separate
3211 LICENCE file containing the conditions.
3212
3213 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
3214 Perl, that is the internal \ allows the delimiter to be included in the
3215 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
3216 the final delimiter, add \ to the end of the pattern (to test the error).
3217
3218 4. Added the convenience functions for extracting substrings after a successful
3219 match. Updated pcretest to make it able to test these functions.
3220
3221
3222 Version 2.02 14-Jan-99
3223 ----------------------
3224
3225 1. Initialized the working variables associated with each extraction so that
3226 their saving and restoring doesn't refer to uninitialized store.
3227
3228 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
3229 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
3230 fix the problem.
3231
3232 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
3233 calls, and wasn't printing the correct value for compiling calls. Increased the
3234 default value of LOOPREPEAT, and the number of significant figures in the
3235 times.
3236
3237 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
3238
3239 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
3240 a building problem on Windows NT with a FAT file system.
3241
3242
3243 Version 2.01 21-Oct-98
3244 ----------------------
3245
3246 1. Changed the API for pcre_compile() to allow for the provision of a pointer
3247 to character tables built by pcre_maketables() in the current locale. If NULL
3248 is passed, the default tables are used.
3249
3250
3251 Version 2.00 24-Sep-98
3252 ----------------------
3253
3254 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
3255 it any more.
3256
3257 2. Allow quantification of (?>) groups, and make it work correctly.
3258
3259 3. The first character computation wasn't working for (?>) groups.
3260
3261 4. Correct the implementation of \Z (it is permitted to match on the \n at the
3262 end of the subject) and add 5.005's \z, which really does match only at the
3263 very end of the subject.
3264
3265 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
3266
3267 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
3268 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
3269 localized options. All options to pcre_study() were also removed.
3270
3271 7. Add other new features from 5.005:
3272
3273 $(?<= positive lookbehind
3274 $(?<! negative lookbehind
3275 (?imsx-imsx) added the unsetting capability
3276 such a setting is global if at outer level; local otherwise
3277 (?imsx-imsx:) non-capturing groups with option setting
3278 (?(cond)re|re) conditional pattern matching
3279
3280 A backreference to itself in a repeated group matches the previous
3281 captured string.
3282
3283 8. General tidying up of studying (both automatic and via "study")
3284 consequential on the addition of new assertions.
3285
3286 9. As in 5.005, unlimited repeated groups that could match an empty substring
3287 are no longer faulted at compile time. Instead, the loop is forcibly broken at
3288 runtime if any iteration does actually match an empty substring.
3289
3290 10. Include the RunTest script in the distribution.
3291
3292 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
3293 discrepancies, some of which were old and were also with respect to 5.004. They
3294 have now been fixed.
3295
3296
3297 Version 1.09 28-Apr-98
3298 ----------------------
3299
3300 1. A negated single character class followed by a quantifier with a minimum
3301 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
3302 program crashes, or just wrong answers. This did not apply to negated classes
3303 containing more than one character, or to minima other than one.
3304
3305
3306 Version 1.08 27-Mar-98
3307 ----------------------
3308
3309 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
3310
3311 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
3312 latter must appear before anything that relies on it in the pattern.
3313
3314
3315 Version 1.07 16-Feb-98
3316 ----------------------
3317
3318 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
3319 repeat of a potentially empty string).
3320
3321
3322 Version 1.06 23-Jan-98
3323 ----------------------
3324
3325 1. Added Markus Oberhumer's little patches for C++.
3326
3327 2. Literal strings longer than 255 characters were broken.
3328
3329
3330 Version 1.05 23-Dec-97
3331 ----------------------
3332
3333 1. Negated character classes containing more than one character were failing if
3334 PCRE_CASELESS was set at run time.
3335
3336
3337 Version 1.04 19-Dec-97
3338 ----------------------
3339
3340 1. Corrected the man page, where some "const" qualifiers had been omitted.
3341
3342 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
3343 input syntax.
3344
3345 3. Fixed memory leak which occurred when a regex with back references was
3346 matched with an offsets vector that wasn't big enough. The temporary memory
3347 that is used in this case wasn't being freed if the match failed.
3348
3349 4. Tidied pcretest to ensure it frees memory that it gets.
3350
3351 5. Temporary memory was being obtained in the case where the passed offsets
3352 vector was exactly big enough.
3353
3354 6. Corrected definition of offsetof() from change 5 below.
3355
3356 7. I had screwed up change 6 below and broken the rules for the use of
3357 setjmp(). Now fixed.
3358
3359
3360 Version 1.03 18-Dec-97
3361 ----------------------
3362
3363 1. A erroneous regex with a missing opening parenthesis was correctly
3364 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
3365 on some systems.
3366
3367 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
3368 it was reported that one broken compiler failed on the former because "code" is
3369 also an independent variable.
3370
3371 3. The erroneous regex a[]b caused an array overrun reference.
3372
3373 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
3374 fail on data ending with that character. (It was going on too far, and checking
3375 the next character, typically a binary zero.) This was specific to the
3376 optimized code for single-character negative classes.
3377
3378 5. Added a contributed patch from the TIN world which does the following:
3379
3380 + Add an undef for memmove, in case the the system defines a macro for it.
3381
3382 + Add a definition of offsetof(), in case there isn't one. (I don't know
3383 the reason behind this - offsetof() is part of the ANSI standard - but
3384 it does no harm).
3385
3386 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
3387 most of the places where whitespace preceded '#'. I have given up and
3388 allowed the remaining 2 cases to be at the margin.
3389
3390 + Rename some variables in pcre to eliminate shadowing. This seems very
3391 pedantic, but does no harm, of course.
3392
3393 6. Moved the call to setjmp() into its own function, to get rid of warnings
3394 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
3395
3396 7. Constructs such as \d{8,} were compiling into the equivalent of
3397 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
3398 outcome, but in this particular case used more store than had been allocated,
3399 which caused the bug to be discovered because it threw up an internal error.
3400
3401 8. The debugging code in both pcre and pcretest for outputting the compiled
3402 form of a regex was going wrong in the case of back references followed by
3403 curly-bracketed repeats.
3404
3405
3406 Version 1.02 12-Dec-97
3407 ----------------------
3408
3409 1. Typos in pcre.3 and comments in the source fixed.
3410
3411 2. Applied a contributed patch to get rid of places where it used to remove
3412 'const' from variables, and fixed some signed/unsigned and uninitialized
3413 variable warnings.
3414
3415 3. Added the "runtest" target to Makefile.
3416
3417 4. Set default compiler flag to -O2 rather than just -O.
3418
3419
3420 Version 1.01 19-Nov-97
3421 ----------------------
3422
3423 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
3424 like /([ab]*)*/, that is, for classes with more than one character in them.
3425
3426 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
3427 as /((?>a*))*/ (a PCRE_EXTRA facility).
3428
3429
3430 Version 1.00 18-Nov-97
3431 ----------------------
3432
3433 1. Added compile-time macros to support systems such as SunOS4 which don't have
3434 memmove() or strerror() but have other things that can be used instead.
3435
3436 2. Arranged that "make clean" removes the executables.
3437
3438
3439 Version 0.99 27-Oct-97
3440 ----------------------
3441
3442 1. Fixed bug in code for optimizing classes with only one character. It was
3443 initializing a 32-byte map regardless, which could cause it to run off the end
3444 of the memory it had got.
3445
3446 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
3447
3448
3449 Version 0.98 22-Oct-97
3450 ----------------------
3451
3452 1. Fixed bug in code for handling temporary memory usage when there are more
3453 back references than supplied space in the ovector. This could cause segfaults.
3454
3455
3456 Version 0.97 21-Oct-97
3457 ----------------------
3458
3459 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
3460
3461 2. Optimized negated single characters not to use a bit map.
3462
3463 3. Brought error texts together as macro definitions; clarified some of them;
3464 fixed one that was wrong - it said "range out of order" when it meant "invalid
3465 escape sequence".
3466
3467 4. Changed some char * arguments to const char *.
3468
3469 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
3470
3471 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
3472 pcretest.
3473
3474
3475 Version 0.96 16-Oct-97
3476 ----------------------
3477
3478 1. Added a simple "pgrep" utility to the distribution.
3479
3480 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
3481 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
3482 where "ddd" means "one or more decimal digits".
3483
3484 3. Fixed serious bug. If a pattern had a back reference, but the call to
3485 pcre_exec() didn't supply a large enough ovector to record the related
3486 identifying subpattern, the match always failed. PCRE now remembers the number
3487 of the largest back reference, and gets some temporary memory in which to save
3488 the offsets during matching if necessary, in order to ensure that
3489 backreferences always work.
3490
3491 4. Increased the compatibility with Perl in a number of ways:
3492
3493 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
3494 to request this handling. The option can be set at compile or exec time.
3495
3496 (b) $ matches before a terminating newline by default; an option
3497 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
3498 mode). The option can be set at compile or exec time.
3499
3500 (c) The handling of \ followed by a digit other than 0 is now supposed to be
3501 the same as Perl's. If the decimal number it represents is less than 10
3502 or there aren't that many previous left capturing parentheses, an octal
3503 escape is read. Inside a character class, it's always an octal escape,
3504 even if it is a single digit.
3505
3506 (d) An escaped but undefined alphabetic character is taken as a literal,
3507 unless PCRE_EXTRA is set. Currently this just reserves the remaining
3508 escapes.
3509
3510 (e) {0} is now permitted. (The previous item is removed from the compiled
3511 pattern).
3512
3513 5. Changed all the names of code files so that the basic parts are no longer
3514 than 10 characters, and abolished the teeny "globals.c" file.
3515
3516 6. Changed the handling of character classes; they are now done with a 32-byte
3517 bit map always.
3518
3519 7. Added the -d and /D options to pcretest to make it possible to look at the
3520 internals of compilation without having to recompile pcre.
3521
3522
3523 Version 0.95 23-Sep-97
3524 ----------------------
3525
3526 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
3527 \x20 at the start of a run of normal characters. These were being treated as
3528 real characters, instead of the source characters being re-checked.
3529
3530
3531 Version 0.94 18-Sep-97
3532 ----------------------
3533
3534 1. The functions are now thread-safe, with the caveat that the global variables
3535 containing pointers to malloc() and free() or alternative functions are the
3536 same for all threads.
3537
3538 2. Get pcre_study() to generate a bitmap of initial characters for non-
3539 anchored patterns when this is possible, and use it if passed to pcre_exec().
3540
3541
3542 Version 0.93 15-Sep-97
3543 ----------------------
3544
3545 1. /(b)|(:+)/ was computing an incorrect first character.
3546
3547 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
3548 but not actually doing anything yet.
3549
3550 3. Treat "-" characters in classes that cannot be part of ranges as literals,
3551 as Perl does (e.g. [-az] or [az-]).
3552
3553 4. Set the anchored flag if a branch starts with .* or .*? because that tests
3554 all possible positions.
3555
3556 5. Split up into different modules to avoid including unneeded functions in a
3557 compiled binary. However, compile and exec are still in one module. The "study"
3558 function is split off.
3559
3560 6. The character tables are now in a separate module whose source is generated
3561 by an auxiliary program - but can then be edited by hand if required. There are
3562 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
3563 toupper() in the code.
3564
3565 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
3566 make them global. Abolish the function for setting them, as the caller can now
3567 set them directly.
3568
3569
3570 Version 0.92 11-Sep-97
3571 ----------------------
3572
3573 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
3574 (e.g. /a{1,3}/) was broken (I mis-optimized it).
3575
3576 2. Caseless matching was not working in character classes if the characters in
3577 the pattern were in upper case.
3578
3579 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
3580
3581 4. Make PCRE_ANCHORED public and accept as a compile option.
3582
3583 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
3584 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
3585 pass them.
3586
3587 6. Give an error if bad option bits passed at compile or run time.
3588
3589 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
3590 pcretest to cause it to pass that flag.
3591
3592 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
3593 options, and the first character, if set.
3594
3595 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
3596
3597
3598 Version 0.91 10-Sep-97
3599 ----------------------
3600
3601 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
3602 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
3603
3604 2. PCRE was looping on encountering an indefinitely repeated back reference to
3605 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
3606 Perl does - treats the match as successful.
3607
3608 ****

Properties

Name Value
svn:eol-style native
svn:keywords "Author Date Id Revision Url"

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12