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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1368 - (show annotations)
Tue Oct 8 10:01:58 2013 UTC (5 years, 11 months ago) by ph10
File size: 253045 byte(s)
Two CMake patches (remove bash-ism, add missing defines).
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.34 xx-xxxx-201x
5 --------------------------
6
7 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
8 executable memory. Patch inspired by Carsten Klein.
9
10 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
11 this macro is never tested and has no effect, because the work to support
12 coverage involves only compiling and linking options and special targets in
13 the Makefile. The comment in config.h implied that defining the macro would
14 enable coverage support, which is totally false. There was also support for
15 setting this macro in the CMake files (my fault, I just copied it from
16 configure). SUPPORT_GCOV has now been removed.
17
18 3. Make a small performance improvement in strlen16() and strlen32() in
19 pcretest.
20
21 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
22 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
23 these cases). There was also one in pcretest.c.
24
25 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
26
27 6. In UTF mode, the code for checking whether a group could match an empty
28 string (which is used for indefinitely repeated groups to allow for
29 breaking an infinite loop) was broken when the group contained a repeated
30 negated single-character class with a character that occupied more than one
31 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
32 UTF-8 mode). The effect was undefined: the group might or might not be
33 deemed as matching an empty string, or the program might have crashed.
34
35 7. The code for checking whether a group could match an empty string was not
36 recognizing that \h, \H, \v, \V, and \R must match a character.
37
38 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
39 an empty string. If it can, pcretest shows this in its information output.
40
41 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
42 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
43 matched by pcre_exec() without using JIT:
44
45 (a) If the rest of the pattern did not match after a maximal run of
46 grapheme clusters, the code for backing up to try with fewer of them
47 did not always back up over a full grapheme when characters that do not
48 have the modifier quality were involved, e.g. Hangul syllables.
49
50 (b) If the match point in a subject started with modifier character, and
51 there was no match, the code could incorrectly back up beyond the match
52 point, and potentially beyond the first character in the subject,
53 leading to a segfault or an incorrect match result.
54
55 10. A conditional group with an assertion condition could lead to PCRE
56 recording an incorrect first data item for a match if no other first data
57 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
58 first data item, and therefore matched "ca" after "c" instead of at the
59 start.
60
61 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
62 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
63
64 12. The source of pcregrep now includes z/OS-specific code so that it can be
65 compiled for z/OS as part of the special z/OS distribution.
66
67 13. Added the -T and -TM options to pcretest.
68
69 14. The code in pcre_compile.c for creating the table of named capturing groups
70 has been refactored. Instead of creating the table dynamically during the
71 actual compiling pass, the information is remembered during the pre-compile
72 pass (on the stack unless there are more than 20 named groups, in which
73 case malloc() is used) and the whole table is created before the actual
74 compile happens. This has simplified the code (it is now nearly 150 lines
75 shorter) and prepared the way for better handling of references to groups
76 with duplicate names.
77
78 15. A back reference to a named subpattern when there is more than one of the
79 same name now checks them in the order in which they appear in the pattern.
80 The first one that is set is used for the reference. Previously only the
81 first one was inspected. This change makes PCRE more compatible with Perl.
82
83 16. Unicode character properties were updated from Unicode 6.3.0.
84
85 17. The compile-time code for auto-possessification has been refactored, based
86 on a patch by Zoltan Herczeg. It now happens after instead of during
87 compilation. The code is cleaner, and more cases are handled. The option
88 PCRE_NO_AUTO_POSSESSIFY is added for testing purposes, and the -O and /O
89 options in pcretest are provided to set it.
90
91 18. The character VT has been added to the set of characters that match \s and
92 are generally treated as white space, following this same change in Perl
93 5.18. There is now no difference between "Perl space" and "POSIX space".
94
95 19. The code for checking named groups as conditions, either for being set or
96 for being recursed, has been refactored (this is related to 14 and 15
97 above). Processing unduplicated named groups should now be as fast at
98 numerical groups, and processing duplicated groups should be faster than
99 before.
100
101 20. Two patches to the CMake build system, by Alexander Barkov:
102
103 (1) Replace the "source" command by "." in CMakeLists.txt because
104 "source" is a bash-ism.
105
106 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
107 without these the CMake build does not work on Solaris.
108
109
110 Version 8.33 28-May-2013
111 ------------------------
112
113 1. Added 'U' to some constants that are compared to unsigned integers, to
114 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
115 variables that are added to signed variables, to ensure the result is
116 signed and can be negated.
117
118 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
119 CMake config files.
120
121 3. Revise the creation of config.h.generic so that all boolean macros are
122 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
123 overriding via -D on the command line possible.
124
125 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
126 to unsigned int is reported to make a quite noticeable speed difference in
127 a specific Windows environment. Testing on Linux did also appear to show
128 some benefit (and it is clearly not harmful). Also fixed the definition of
129 Xop which should be unsigned.
130
131 5. Related to (4), changing the definition of the intermediate variable cc
132 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
133 improvements.
134
135 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
136 unnecessary spaces.
137
138 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
139 and later.
140
141 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
142
143 9. Optimizing fast_forward_start_bits in JIT.
144
145 10. Adding support for callouts in JIT, and fixing some issues revealed
146 during this work. Namely:
147
148 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
149
150 (b) Minimum length was not checked before the matching is started.
151
152 11. The value of capture_last that is passed to callouts was incorrect in some
153 cases when there was a capture on one path that was subsequently abandoned
154 after a backtrack. Also, the capture_last value is now reset after a
155 recursion, since all captures are also reset in this case.
156
157 12. The interpreter no longer returns the "too many substrings" error in the
158 case when an overflowing capture is in a branch that is subsequently
159 abandoned after a backtrack.
160
161 13. In the pathological case when an offset vector of size 2 is used, pcretest
162 now prints out the matched string after a yield of 0 or 1.
163
164 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
165 Only supported by the JIT compiler at the moment.
166
167 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
168
169 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
170 offset of the starting point of the matching process, provided the offsets
171 vector is large enough.
172
173 17. The \A escape now records a lookbehind value of 1, though its execution
174 does not actually inspect the previous character. This is to ensure that,
175 in partial multi-segment matching, at least one character from the old
176 segment is retained when a new segment is processed. Otherwise, if there
177 are no lookbehinds in the pattern, \A might match incorrectly at the start
178 of a new segment.
179
180 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
181
182 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
183 gives some modest performance improvement in 8-bit mode.
184
185 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
186 be expressed in certain programming languages using Universal Character
187 Names.
188
189 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
190 which points out that "non characters" are not "characters that may not
191 appear in Unicode strings" but rather "characters that are reserved for
192 internal use and have only local meaning".
193
194 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
195 there was a conditional group that depended on an assertion, if the
196 assertion was false, the callout that immediately followed the alternation
197 in the condition was skipped when pcre_exec() was used for matching.
198
199 23. Allow an explicit callout to be inserted before an assertion that is the
200 condition for a conditional group, for compatibility with automatic
201 callouts, which always insert a callout at this point.
202
203 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
204 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
205
206 25. (*PRUNE) is now supported by the JIT compiler.
207
208 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
209
210 27. Fix the case where there are two or more SKIPs with arguments that may be
211 ignored.
212
213 28. (*SKIP) is now supported by the JIT compiler.
214
215 29. (*THEN) is now supported by the JIT compiler.
216
217 30. Update RunTest with additional test selector options.
218
219 31. The way PCRE handles backtracking verbs has been changed in two ways.
220
221 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
222 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
223 backtracking. In some cases this makes it more Perl-compatible, but Perl's
224 rather obscure rules do not always do the same thing.
225
226 (2) Previously, backtracking verbs were confined within assertions. This is
227 no longer the case for positive assertions, except for (*ACCEPT). Again,
228 this sometimes improves Perl compatibility, and sometimes does not.
229
230 32. A number of tests that were in test 2 because Perl did things differently
231 have been moved to test 1, because either Perl or PCRE has changed, and
232 these tests are now compatible.
233
234 32. Backtracking control verbs are now handled in the same way in JIT and
235 interpreter.
236
237 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
238 contained a forward subroutine reference caused a compile error.
239
240 34. Auto-detect and optimize limited repetitions in JIT.
241
242 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
243 blocking (*UTF) etc.
244
245 36. In the interpreter, maximizing pattern repetitions for characters and
246 character types now use tail recursion, which reduces stack usage.
247
248 37. The value of the max lookbehind was not correctly preserved if a compiled
249 and saved regex was reloaded on a host of different endianness.
250
251 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
252 of the compiled pattern block, expand the flags field from 16 to 32 bits
253 because it was almost full.
254
255 39. Try madvise first before posix_madvise.
256
257 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
258 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
259 original change locked out all matches of empty strings. This has been
260 changed so that one match of an empty string per line is recognized.
261 Subsequent searches on the same line (for colouring or for --only-matching,
262 for example) do not recognize empty strings.
263
264 41. Applied a user patch to fix a number of spelling mistakes in comments.
265
266 42. Data lines longer than 65536 caused pcretest to crash.
267
268 43. Clarified the data type for length and startoffset arguments for pcre_exec
269 and pcre_dfa_exec in the function-specific man pages, where they were
270 explicitly stated to be in bytes, never having been updated. I also added
271 some clarification to the pcreapi man page.
272
273 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
274 a segmentation fault.
275
276
277 Version 8.32 30-November-2012
278 -----------------------------
279
280 1. Improved JIT compiler optimizations for first character search and single
281 character iterators.
282
283 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
284 Patch by Daniel Richard G.
285
286 3. Single character iterator optimizations in the JIT compiler.
287
288 4. Improved JIT compiler optimizations for character ranges.
289
290 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
291 Reported by Giuseppe D'Angelo.
292
293 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
294 of a line, was being set incorrectly in cases where .* appeared inside
295 atomic brackets at the start of a pattern, or where there was a subsequent
296 *PRUNE or *SKIP.
297
298 7. Improved instruction cache flush for POWER/PowerPC.
299 Patch by Daniel Richard G.
300
301 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
302 grep:
303
304 (a) There is now no limit to the number of patterns to be matched.
305
306 (b) An error is given if a pattern is too long.
307
308 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
309 are now supported.
310
311 (d) --exclude-from and --include-from (multiple use) have been added.
312
313 (e) Exclusions and inclusions now apply to all files and directories, not
314 just to those obtained from scanning a directory recursively.
315
316 (f) Multiple uses of -f and --file-list are now supported.
317
318 (g) In a Windows environment, the default for -d has been changed from
319 "read" (the GNU grep default) to "skip", because otherwise the presence
320 of a directory in the file list provokes an error.
321
322 (h) The documentation has been revised and clarified in places.
323
324 9. Improve the matching speed of capturing brackets.
325
326 10. Changed the meaning of \X so that it now matches a Unicode extended
327 grapheme cluster.
328
329 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
330 out POSIX threads when JIT support is configured.
331
332 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
333
334 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
335 structure could go wrong in environments where size_t is not the same size
336 as int.
337
338 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
339 set.
340
341 15. The EBCDIC support had decayed; later updates to the code had included
342 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
343 general tidy up of EBCDIC-related issues, and the documentation was also
344 not quite right. There is now a test that can be run on ASCII systems to
345 check some of the EBCDIC-related things (but is it not a full test).
346
347 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
348 in a small tidy to the code.
349
350 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
351
352 18. If the --only-matching (-o) option in pcregrep is specified multiple
353 times, each one causes appropriate output. For example, -o1 -o2 outputs the
354 substrings matched by the 1st and 2nd capturing parentheses. A separating
355 string can be specified by --om-separator (default empty).
356
357 19. Improving the first n character searches.
358
359 20. Turn case lists for horizontal and vertical white space into macros so that
360 they are defined only once.
361
362 21. This set of changes together give more compatible Unicode case-folding
363 behaviour for characters that have more than one other case when UCP
364 support is available.
365
366 (a) The Unicode property table now has offsets into a new table of sets of
367 three or more characters that are case-equivalent. The MultiStage2.py
368 script that generates these tables (the pcre_ucd.c file) now scans
369 CaseFolding.txt instead of UnicodeData.txt for character case
370 information.
371
372 (b) The code for adding characters or ranges of characters to a character
373 class has been abstracted into a generalized function that also handles
374 case-independence. In UTF-mode with UCP support, this uses the new data
375 to handle characters with more than one other case.
376
377 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
378 whose other case is greater than 256 are now correctly matched
379 caselessly. Previously, the high codepoint matched the low one, but not
380 vice versa.
381
382 (d) The processing of \h, \H, \v, and \ in character classes now makes use
383 of the new class addition function, using character lists defined as
384 macros alongside the case definitions of 20 above.
385
386 (e) Caseless back references now work with characters that have more than
387 one other case.
388
389 (f) General caseless matching of characters with more than one other case
390 is supported.
391
392 22. Unicode character properties were updated from Unicode 6.2.0
393
394 23. Improved CMake support under Windows. Patch by Daniel Richard G.
395
396 24. Add support for 32-bit character strings, and UTF-32
397
398 25. Major JIT compiler update (code refactoring and bugfixing).
399 Experimental Sparc 32 support is added.
400
401 26. Applied a modified version of Daniel Richard G's patch to create
402 pcre.h.generic and config.h.generic by "make" instead of in the
403 PrepareRelease script.
404
405 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
406 pcre_compile.c when checking for a zero character.
407
408 28. Introducing a native interface for JIT. Through this interface, the compiled
409 machine code can be directly executed. The purpose of this interface is to
410 provide fast pattern matching, so several sanity checks are not performed.
411 However, feature tests are still performed. The new interface provides
412 1.4x speedup compared to the old one.
413
414 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
415 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
416 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
417
418 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
419 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
420 the "old" RFC 2279). Instead, it ended up passing a negative length to
421 pcre_exec().
422
423 31. Add support for GCC's visibility feature to hide internal functions.
424
425 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
426 "unknown -C option" after outputting 0 or 1.
427
428 33. There is now support for generating a code coverage report for the test
429 suite in environments where gcc is the compiler and lcov is installed. This
430 is mainly for the benefit of the developers.
431
432 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
433 unaddressable using valgrind annotations, allowing valgrind to detect
434 invalid memory accesses. This is mainly for the benefit of the developers.
435
436 25. (*UTF) can now be used to start a pattern in any of the three libraries.
437
438 26. Give configure error if --enable-cpp but no C++ compiler found.
439
440
441 Version 8.31 06-July-2012
442 -------------------------
443
444 1. Fixing a wrong JIT test case and some compiler warnings.
445
446 2. Removed a bashism from the RunTest script.
447
448 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
449 to unsigned type, result still unsigned" that was given by an MS compiler
450 on encountering the code "-sizeof(xxx)".
451
452 4. Partial matching support is added to the JIT compiler.
453
454 5. Fixed several bugs concerned with partial matching of items that consist
455 of more than one character:
456
457 (a) /^(..)\1/ did not partially match "aba" because checking references was
458 done on an "all or nothing" basis. This also applied to repeated
459 references.
460
461 (b) \R did not give a hard partial match if \r was found at the end of the
462 subject.
463
464 (c) \X did not give a hard partial match after matching one or more
465 characters at the end of the subject.
466
467 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
468 a partial match for the string "\r".
469
470 (e) When newline was set to CRLF, the metacharacter "." did not recognize
471 a partial match for a CR character at the end of the subject string.
472
473 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
474 running pcretest, the text "(JIT)" added to the output whenever JIT is
475 actually used to run the match.
476
477 7. Individual JIT compile options can be set in pcretest by following -s+[+]
478 or /S+[+] with a digit between 1 and 7.
479
480 8. OP_NOT now supports any UTF character not just single-byte ones.
481
482 9. (*MARK) control verb is now supported by the JIT compiler.
483
484 10. The command "./RunTest list" lists the available tests without actually
485 running any of them. (Because I keep forgetting what they all are.)
486
487 11. Add PCRE_INFO_MAXLOOKBEHIND.
488
489 12. Applied a (slightly modified) user-supplied patch that improves performance
490 when the heap is used for recursion (compiled with --disable-stack-for-
491 recursion). Instead of malloc and free for each heap frame each time a
492 logical recursion happens, frames are retained on a chain and re-used where
493 possible. This sometimes gives as much as 30% improvement.
494
495 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
496 call.
497
498 14. As documented, (*COMMIT) is now confined to within a positive assertion.
499
500 15. It is now possible to link pcretest with libedit as an alternative to
501 libreadline.
502
503 16. (*COMMIT) control verb is now supported by the JIT compiler.
504
505 17. The Unicode data tables have been updated to Unicode 6.1.0.
506
507 18. Added --file-list option to pcregrep.
508
509 19. Added binary file support to pcregrep, including the -a, --binary-files,
510 -I, and --text options.
511
512 20. The madvise function is renamed for posix_madvise for QNX compatibility
513 reasons. Fixed by Giuseppe D'Angelo.
514
515 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
516
517 22. Changed the option for creating symbolic links for 16-bit man pages from
518 -s to -sf so that re-installing does not cause issues.
519
520 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
521
522 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
523 matches in certain environments (the workspace was not being correctly
524 retained). Also added to pcre_dfa_exec() a simple plausibility check on
525 some of the workspace data at the beginning of a restart.
526
527 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
528 was not doing so when it should - probably a typo introduced by SVN 528
529 (change 8.10/14).
530
531 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
532 \w+ when the character tables indicated that \x{c4} was a word character.
533 There were several related cases, all because the tests for doing a table
534 lookup were testing for characters less than 127 instead of 255.
535
536 27. If a pattern contains capturing parentheses that are not used in a match,
537 their slots in the ovector are set to -1. For those that are higher than
538 any matched groups, this happens at the end of processing. In the case when
539 there were back references that the ovector was too small to contain
540 (causing temporary malloc'd memory to be used during matching), and the
541 highest capturing number was not used, memory off the end of the ovector
542 was incorrectly being set to -1. (It was using the size of the temporary
543 memory instead of the true size.)
544
545 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
546 ovector size, it uses memory at the end of the block that it has got.
547
548 29. Check for an overlong MARK name and give an error at compile time. The
549 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
550
551 30. JIT compiler update.
552
553 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
554 Rill for the patch.
555
556 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
557
558 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
559
560 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
561 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
562 was enabled.
563
564 35. Improve JIT code generation for greedy plus quantifier.
565
566 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
567 1 to "aa" instead of to an empty string. The bug affected repeated groups
568 that could potentially match an empty string.
569
570 37. Optimizing single character iterators in JIT.
571
572 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
573 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
574 codepoints that are too big for the mode are faulted, and in a UTF mode,
575 disallowed codepoints are also faulted.
576
577 39. If PCRE was compiled with UTF support, in three places in the DFA
578 matcher there was code that should only have been obeyed in UTF mode, but
579 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
580 processing when bytes with values greater than 127 were present. In 16-bit
581 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
582 both cases the values are those that cannot be the first data item in a UTF
583 character. The three items that might have provoked this were recursions,
584 possessively repeated groups, and atomic groups.
585
586 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
587 and pcregrep, because some OS require shared objects to be explicitly
588 passed to ld, causing the link step to fail if they are not.
589
590 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
591 mode, patterns that started with \h* or \R* might be incorrectly matched.
592
593
594 Version 8.30 04-February-2012
595 -----------------------------
596
597 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
598 name is defined in ctype.h.
599
600 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
601 only in quite long subpatterns.
602
603 3. Removed the function pcre_info(), which has been obsolete and deprecated
604 since it was replaced by pcre_fullinfo() in February 2000.
605
606 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
607 match a (*MARK), and the match failed at the start of the subject, a
608 reference to memory before the start of the subject could occur. This bug
609 was introduced by fix 17 of release 8.21.
610
611 5. A reference to an unset group with zero minimum repetition was giving
612 totally wrong answers (in non-JavaScript-compatibility mode). For example,
613 /(another)?(\1?)test/ matched against "hello world test". This bug was
614 introduced in release 8.13.
615
616 6. Add support for 16-bit character strings (a large amount of work involving
617 many changes and refactorings).
618
619 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
620 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
621 from a file.
622
623 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
624 rounding is not applied in this particular case).
625
626 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
627 if they appear, or are escaped, in patterns.
628
629 10. Get rid of a number of -Wunused-but-set-variable warnings.
630
631 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
632 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
633 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
634 also returns the mark "x". This bug applied to capturing parentheses,
635 non-capturing parentheses, and atomic parentheses. It also applied to some
636 assertions.
637
638 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
639 information out of configure.ac instead of relying on pcre.h.generic, which
640 is not stored in the repository.
641
642 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
643 -lreadline.
644
645 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
646
647 15. Applied Graycode's patch to put the top-level frame on the stack rather
648 than the heap when not using the stack for recursion. This gives a
649 performance improvement in many cases when recursion is not deep.
650
651 16. Experimental code added to "pcretest -C" to output the stack frame size.
652
653
654 Version 8.21 12-Dec-2011
655 ------------------------
656
657 1. Updating the JIT compiler.
658
659 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
660 are added as well.
661
662 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
663 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
664 calling _pcre_jit_exec. Some extra comments are added.
665
666 4. (*MARK) settings inside atomic groups that do not contain any capturing
667 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
668 was introduced by change 18 for 8.20.
669
670 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
671 ECMA-262 standard.
672
673 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
674 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
675 This bug was probably introduced by change 9 of 8.13.
676
677 7. While fixing 6 above, I noticed that a number of other items were being
678 incorrectly rejected as "not fixed length". This arose partly because newer
679 opcodes had not been added to the fixed-length checking code. I have (a)
680 corrected the bug and added tests for these items, and (b) arranged for an
681 error to occur if an unknown opcode is encountered while checking for fixed
682 length instead of just assuming "not fixed length". The items that were
683 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
684 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
685 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
686
687 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
688 being incorrectly compiled and would have given unpredicatble results.
689
690 9. A possessively repeated subpattern with minimum repeat count greater than
691 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
692 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
693 the first (A) could occur when it should not.
694
695 10. Add a cast and remove a redundant test from the code.
696
697 11. JIT should use pcre_malloc/pcre_free for allocation.
698
699 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
700 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
701 is anything other than /usr, -L is still shown).
702
703 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
704
705 14. Perl does not support \N without a following name in a [] class; PCRE now
706 also gives an error.
707
708 15. If a forward reference was repeated with an upper limit of around 2000,
709 it caused the error "internal error: overran compiling workspace". The
710 maximum number of forward references (including repeats) was limited by the
711 internal workspace, and dependent on the LINK_SIZE. The code has been
712 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
713 the default depends on LINK_SIZE. There is a new upper limit (for safety)
714 of around 200,000 forward references. While doing this, I also speeded up
715 the filling in of repeated forward references.
716
717 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
718 incorrectly expecting the subject to contain another "a" after the start.
719
720 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
721 in the match, the SKIP should be ignored. This was not happening; instead
722 the SKIP was being treated as NOMATCH. For patterns such as
723 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
724 tested.
725
726 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
727 now much more compatible with Perl, in particular in cases where the result
728 is a non-match for a non-anchored pattern. For example, if
729 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
730 "m", where previously it did not return a name. A side effect of this
731 change is that for partial matches, the last encountered mark name is
732 returned, as for non matches. A number of tests that were previously not
733 Perl-compatible have been moved into the Perl-compatible test files. The
734 refactoring has had the pleasing side effect of removing one argument from
735 the match() function, thus reducing its stack requirements.
736
737 19. If the /S+ option was used in pcretest to study a pattern using JIT,
738 subsequent uses of /S (without +) incorrectly behaved like /S+.
739
740 21. Retrieve executable code size support for the JIT compiler and fixing
741 some warnings.
742
743 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
744 not work when the shorter character appeared right at the end of the
745 subject string.
746
747 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
748 systems.
749
750 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
751 output it when the /M option is used in pcretest.
752
753 25. The CheckMan script was not being included in the distribution. Also, added
754 an explicit "perl" to run Perl scripts from the PrepareRelease script
755 because this is reportedly needed in Windows.
756
757 26. If study data was being save in a file and studying had not found a set of
758 "starts with" bytes for the pattern, the data written to the file (though
759 never used) was taken from uninitialized memory and so caused valgrind to
760 complain.
761
762 27. Updated RunTest.bat as provided by Sheri Pierce.
763
764 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
765
766 29. Computation of memory usage for the table of capturing group names was
767 giving an unnecessarily large value.
768
769
770 Version 8.20 21-Oct-2011
771 ------------------------
772
773 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
774 a POSIX class. After further experiments with Perl, which convinced me that
775 Perl has bugs and confusions, a closing square bracket is no longer allowed
776 in a POSIX name. This bug also affected patterns with classes that started
777 with full stops.
778
779 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
780 captured substring, but while checking the failing first alternative,
781 substring 1 is temporarily captured. If the output vector supplied to
782 pcre_exec() was not big enough for this capture, the yield of the function
783 was still zero ("insufficient space for captured substrings"). This cannot
784 be totally fixed without adding another stack variable, which seems a lot
785 of expense for a edge case. However, I have improved the situation in cases
786 such as /(a)(b)x|abc/ matched against "abc", where the return code
787 indicates that fewer than the maximum number of slots in the ovector have
788 been set.
789
790 3. Related to (2) above: when there are more back references in a pattern than
791 slots in the output vector, pcre_exec() uses temporary memory during
792 matching, and copies in the captures as far as possible afterwards. It was
793 using the entire output vector, but this conflicts with the specification
794 that only 2/3 is used for passing back captured substrings. Now it uses
795 only the first 2/3, for compatibility. This is, of course, another edge
796 case.
797
798 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
799 main code base, and can be used by building with --enable-jit. When this is
800 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
801 runtime --no-jit option is given.
802
803 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
804 ovector, the return from the function was zero, implying that there were
805 other matches that did not fit. The correct "exactly full" value is now
806 returned.
807
808 6. If a subpattern that was called recursively or as a subroutine contained
809 (*PRUNE) or any other control that caused it to give a non-standard return,
810 invalid errors such as "Error -26 (nested recursion at the same subject
811 position)" or even infinite loops could occur.
812
813 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
814 computing the minimum length on reaching *ACCEPT, and so ended up with the
815 wrong value of 1 rather than 0. Further investigation indicates that
816 computing a minimum subject length in the presence of *ACCEPT is difficult
817 (think back references, subroutine calls), and so I have changed the code
818 so that no minimum is registered for a pattern that contains *ACCEPT.
819
820 8. If (*THEN) was present in the first (true) branch of a conditional group,
821 it was not handled as intended. [But see 16 below.]
822
823 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
824 Sheri Pierce.
825
826 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
827 the first byte in a match must be "a".
828
829 11. Change 17 for 8.13 increased the recursion depth for patterns like
830 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
831 pattern contains any instances of (*THEN). If it does not, the old
832 optimizations are restored. It would be nice to do this on a per-group
833 basis, but at the moment that is not feasible.
834
835 12. In some environments, the output of pcretest -C is CRLF terminated. This
836 broke RunTest's code that checks for the link size. A single white space
837 character after the value is now allowed for.
838
839 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
840 For "fr", it uses the Windows-specific input and output files.
841
842 14. If (*THEN) appeared in a group that was called recursively or as a
843 subroutine, it did not work as intended. [But see next item.]
844
845 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
846 pattern fragments (but not containing any | characters). If A and B are
847 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
848 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
849 D. In other words, Perl considers parentheses that do not contain any |
850 characters to be part of a surrounding alternative, whereas PCRE was
851 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
852 differently. PCRE now behaves in the same way as Perl, except in the case
853 of subroutine/recursion calls such as (?1) which have in any case always
854 been different (but PCRE had them first :-).
855
856 16. Related to 15 above: Perl does not treat the | in a conditional group as
857 creating alternatives. Such a group is treated in the same way as an
858 ordinary group without any | characters when processing (*THEN). PCRE has
859 been changed to match Perl's behaviour.
860
861 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
862 RunGrepTest script failed.
863
864 18. Change 22 for version 13 caused atomic groups to use more stack. This is
865 inevitable for groups that contain captures, but it can lead to a lot of
866 stack use in large patterns. The old behaviour has been restored for atomic
867 groups that do not contain any capturing parentheses.
868
869 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
870 suppress the check for a minimum subject length at run time. (If it was
871 given to pcre_exec() or pcre_dfa_exec() it did work.)
872
873 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
874 fail to work when decoding hex characters in data strings in EBCDIC
875 environments.
876
877 21. It appears that in at least one Mac OS environment, the isxdigit() function
878 is implemented as a macro that evaluates to its argument more than once,
879 contravening the C 90 Standard (I haven't checked a later standard). There
880 was an instance in pcretest which caused it to go wrong when processing
881 \x{...} escapes in subject strings. The has been rewritten to avoid using
882 things like p++ in the argument of isxdigit().
883
884
885 Version 8.13 16-Aug-2011
886 ------------------------
887
888 1. The Unicode data tables have been updated to Unicode 6.0.0.
889
890 2. Two minor typos in pcre_internal.h have been fixed.
891
892 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
893 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
894 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
895
896 4. There were a number of related bugs in the code for matching backrefences
897 caselessly in UTF-8 mode when codes for the characters concerned were
898 different numbers of bytes. For example, U+023A and U+2C65 are an upper
899 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
900 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
901 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
902 2-byte code at the end of the subject (it thought there wasn't enough data
903 left).
904
905 5. Comprehensive information about what went wrong is now returned by
906 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
907 as the output vector has at least 2 elements. The offset of the start of
908 the failing character and a reason code are placed in the vector.
909
910 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
911 now returned is for the first byte of the failing character, instead of the
912 last byte inspected. This is an incompatible change, but I hope it is small
913 enough not to be a problem. It makes the returned offset consistent with
914 pcre_exec() and pcre_dfa_exec().
915
916 7. pcretest now gives a text phrase as well as the error number when
917 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
918 failure, the offset and reason code are output.
919
920 8. When \R was used with a maximizing quantifier it failed to skip backwards
921 over a \r\n pair if the subsequent match failed. Instead, it just skipped
922 back over a single character (\n). This seems wrong (because it treated the
923 two characters as a single entity when going forwards), conflicts with the
924 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
925 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
926 has been changed.
927
928 9. Some internal refactoring has changed the processing so that the handling
929 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
930 time (the PCRE_DOTALL option was changed this way some time ago: version
931 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
932 which was always a bit of a fudge. It also means that there is one less
933 argument for the match() function, which reduces its stack requirements
934 slightly. This change also fixes an incompatibility with Perl: the pattern
935 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
936
937 10. More internal refactoring has drastically reduced the number of recursive
938 calls to match() for possessively repeated groups such as (abc)++ when
939 using pcre_exec().
940
941 11. While implementing 10, a number of bugs in the handling of groups were
942 discovered and fixed:
943
944 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
945 (a|)*(?1) gave a compile-time internal error.
946 ((a|)+)+ did not notice that the outer group could match an empty string.
947 (^a|^)+ was not marked as anchored.
948 (.*a|.*)+ was not marked as matching at start or after a newline.
949
950 12. Yet more internal refactoring has removed another argument from the match()
951 function. Special calls to this function are now indicated by setting a
952 value in a variable in the "match data" data block.
953
954 13. Be more explicit in pcre_study() instead of relying on "default" for
955 opcodes that mean there is no starting character; this means that when new
956 ones are added and accidentally left out of pcre_study(), testing should
957 pick them up.
958
959 14. The -s option of pcretest has been documented for ages as being an old
960 synonym of -m (show memory usage). I have changed it to mean "force study
961 for every regex", that is, assume /S for every regex. This is similar to -i
962 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
963 using it. It makes it easier to run collections of tests with and without
964 study enabled, and thereby test pcre_study() more easily. All the standard
965 tests are now run with and without -s (but some patterns can be marked as
966 "never study" - see 20 below).
967
968 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
969 restoration of the capturing data to the outer values was not happening
970 correctly.
971
972 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
973 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
974 pattern had matched an empty string, and so incorrectly returned a no
975 match.
976
977 17. There was optimizing code for the last branch of non-capturing parentheses,
978 and also for the obeyed branch of a conditional subexpression, which used
979 tail recursion to cut down on stack usage. Unfortunately, now that there is
980 the possibility of (*THEN) occurring in these branches, tail recursion is
981 no longer possible because the return has to be checked for (*THEN). These
982 two optimizations have therefore been removed. [But see 8.20/11 above.]
983
984 18. If a pattern containing \R was studied, it was assumed that \R always
985 matched two bytes, thus causing the minimum subject length to be
986 incorrectly computed because \R can also match just one byte.
987
988 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
989 was incorrectly computed.
990
991 20. If /S is present twice on a test pattern in pcretest input, it now
992 *disables* studying, thereby overriding the use of -s on the command line
993 (see 14 above). This is necessary for one or two tests to keep the output
994 identical in both cases.
995
996 21. When (*ACCEPT) was used in an assertion that matched an empty string and
997 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
998
999 22. When an atomic group that contained a capturing parenthesis was
1000 successfully matched, but the branch in which it appeared failed, the
1001 capturing was not being forgotten if a higher numbered group was later
1002 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1003 group 1 to "a", when in fact it should be unset. This applied to multi-
1004 branched capturing and non-capturing groups, repeated or not, and also to
1005 positive assertions (capturing in negative assertions does not happen
1006 in PCRE) and also to nested atomic groups.
1007
1008 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1009 subject after a captured substring, to make it easier to tell which of a
1010 number of identical substrings has been captured.
1011
1012 24. The way atomic groups are processed by pcre_exec() has been changed so that
1013 if they are repeated, backtracking one repetition now resets captured
1014 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1015 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1016 "aaa". Previously, it would have been "a". As part of this code
1017 refactoring, the way recursive calls are handled has also been changed.
1018
1019 25. If an assertion condition captured any substrings, they were not passed
1020 back unless some other capturing happened later. For example, if
1021 (?(?=(a))a) was matched against "a", no capturing was returned.
1022
1023 26. When studying a pattern that contained subroutine calls or assertions,
1024 the code for finding the minimum length of a possible match was handling
1025 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1026 group 1 called group 2 while simultaneously a separate group 2 called group
1027 1). A stack overflow occurred in this case. I have fixed this by limiting
1028 the recursion depth to 10.
1029
1030 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1031 Fortmann. This supports explicit test numbers on the command line, and has
1032 argument validation and error reporting.
1033
1034 28. An instance of \X with an unlimited repeat could fail if at any point the
1035 first character it looked at was a mark character.
1036
1037 29. Some minor code refactoring concerning Unicode properties and scripts
1038 should reduce the stack requirement of match() slightly.
1039
1040 30. Added the '=' option to pcretest to check the setting of unused capturing
1041 slots at the end of the pattern, which are documented as being -1, but are
1042 not included in the return count.
1043
1044 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1045 compiled something random. Now it gives a compile-time error (as does
1046 Perl).
1047
1048 32. A *MARK encountered during the processing of a positive assertion is now
1049 recorded and passed back (compatible with Perl).
1050
1051 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1052 had alternative anchored branches, the search for a second match in a line
1053 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1054 matched the line "0102" twice. The same bug affected patterns that started
1055 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1056 twice.
1057
1058 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1059 does, and because of capturing effects, quantifying parenthesized
1060 assertions may at times be useful. Quantifiers are now allowed for
1061 parenthesized assertions.
1062
1063 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1064
1065 36. \g was being checked for fancy things in a character class, when it should
1066 just be a literal "g".
1067
1068 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1069 appearance of a nested POSIX class supersedes an apparent external class.
1070 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1071 unescaped square brackets may also appear as part of class names. For
1072 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1073 more like Perl. (But see 8.20/1 above.)
1074
1075 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1076 was because it thought it was \N{name}, which is not supported).
1077
1078 39. Add minix to OS list not supporting the -S option in pcretest.
1079
1080 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1081 cannot analyze patterns in sufficient detail to catch mutual recursions
1082 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1083 subgroup is called recursively as a subpattern for a second time at the
1084 same position in the subject string. In previous releases this might have
1085 been caught by the recursion limit, or it might have run out of stack.
1086
1087 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1088 happen only once. PCRE was, however incorrectly giving a compile time error
1089 "recursive call could loop indefinitely" because it cannot analyze the
1090 pattern in sufficient detail. The compile time test no longer happens when
1091 PCRE is compiling a conditional subpattern, but actual runaway loops are
1092 now caught at runtime (see 40 above).
1093
1094 42. It seems that Perl allows any characters other than a closing parenthesis
1095 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1096 has been changed to be the same.
1097
1098 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1099 as not to get warnings when autogen.sh is called. Also changed
1100 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1101
1102 44. To help people who use pcregrep to scan files containing exceedingly long
1103 lines, the following changes have been made:
1104
1105 (a) The default value of the buffer size parameter has been increased from
1106 8K to 20K. (The actual buffer used is three times this size.)
1107
1108 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1109 PCRE is built.
1110
1111 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1112 to be set at run time.
1113
1114 (d) Numerical values in pcregrep options can be followed by K or M, for
1115 example --buffer-size=50K.
1116
1117 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1118 given and the return code is set to 2.
1119
1120 45. Add a pointer to the latest mark to the callout data block.
1121
1122 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1123 partial match of an empty string instead of no match. This was specific to
1124 the use of ".".
1125
1126 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1127 complete match instead of a partial match. This bug was dependent on both
1128 the PCRE_UTF8 and PCRE_DOTALL options being set.
1129
1130 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1131 starting byte set, because \b was not being ignored.
1132
1133
1134 Version 8.12 15-Jan-2011
1135 ------------------------
1136
1137 1. Fixed some typos in the markup of the man pages, and wrote a script that
1138 checks for such things as part of the documentation building process.
1139
1140 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1141 --match-limit and --recursion-limit options (added for 8.11). In
1142 particular, this made one of the standard tests fail. (The integer value
1143 went into the wrong half of a long int.)
1144
1145 3. If the --colour option was given to pcregrep with -v (invert match), it
1146 did strange things, either producing crazy output, or crashing. It should,
1147 of course, ignore a request for colour when reporting lines that do not
1148 match.
1149
1150 4. Another pcregrep bug caused similar problems if --colour was specified with
1151 -M (multiline) and the pattern match finished with a line ending.
1152
1153 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1154 matched in multiline mode, the following line was shown as part of the
1155 match. This seems wrong, so I have changed it.
1156
1157 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1158 the check for further matches in the same line (so they could be coloured)
1159 to overrun the end of the current line. If another match was found, it was
1160 incorrectly shown (and then shown again when found in the next line).
1161
1162 7. If pcregrep was compiled under Windows, there was a reference to the
1163 function pcregrep_exit() before it was defined. I am assuming this was
1164 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1165 reported by a user. I've moved the definition above the reference.
1166
1167
1168 Version 8.11 10-Dec-2010
1169 ------------------------
1170
1171 1. (*THEN) was not working properly if there were untried alternatives prior
1172 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1173 backtracked to try for "b" instead of moving to the next alternative branch
1174 at the same level (in this case, to look for "c"). The Perl documentation
1175 is clear that when (*THEN) is backtracked onto, it goes to the "next
1176 alternative in the innermost enclosing group".
1177
1178 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1179 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1180 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1181 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1182 (*THEN).
1183
1184 3. If \s appeared in a character class, it removed the VT character from
1185 the class, even if it had been included by some previous item, for example
1186 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1187 of \s, but is part of the POSIX "space" class.)
1188
1189 4. A partial match never returns an empty string (because you can always
1190 match an empty string at the end of the subject); however the checking for
1191 an empty string was starting at the "start of match" point. This has been
1192 changed to the "earliest inspected character" point, because the returned
1193 data for a partial match starts at this character. This means that, for
1194 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1195 (previously it gave "no match").
1196
1197 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1198 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1199 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1200 has an implication that the given string is incomplete (because a partial
1201 match is preferred over a full match). For this reason, these items now
1202 give a partial match in this situation. [Aside: previously, the one case
1203 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1204 match rather than a full match, which was wrong by the old rules, but is
1205 now correct.]
1206
1207 6. There was a bug in the handling of #-introduced comments, recognized when
1208 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1209 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1210 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1211 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1212 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1213 places in pcre_compile().
1214
1215 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1216 comments when looking ahead for named forward references to subpatterns,
1217 the only newline sequence it recognized was NL. It now handles newlines
1218 according to the set newline convention.
1219
1220 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1221 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1222 cater for a lack of strerror(). These oversights have been fixed.
1223
1224 9. Added --match-limit and --recursion-limit to pcregrep.
1225
1226 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1227
1228 11. When the -o option was used, pcregrep was setting a return code of 1, even
1229 when matches were found, and --line-buffered was not being honoured.
1230
1231 12. Added an optional parentheses number to the -o and --only-matching options
1232 of pcregrep.
1233
1234 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1235 can match an empty string. The code to do it in pcretest and pcredemo
1236 needed fixing:
1237
1238 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1239 only one byte after an empty string match just before CRLF (this case
1240 just got forgotten; "any" and "anycrlf" were OK).
1241
1242 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1243 mode when an empty string match preceded an ASCII character followed by
1244 a non-ASCII character. (The code for advancing by one character rather
1245 than one byte was nonsense.)
1246
1247 (c) The pcredemo.c sample program did not have any code at all to handle
1248 the cases when CRLF is a valid newline sequence.
1249
1250 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1251 as a starting offset was within the subject string. There is now a new
1252 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1253 negative or greater than the length of the string. In order to test this,
1254 pcretest is extended to allow the setting of negative starting offsets.
1255
1256 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1257 starting offset points to the beginning of a UTF-8 character was
1258 unnecessarily clumsy. I tidied it up.
1259
1260 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1261 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1262
1263 17. Nobody had reported that the --include_dir option, which was added in
1264 release 7.7 should have been called --include-dir (hyphen, not underscore)
1265 for compatibility with GNU grep. I have changed it to --include-dir, but
1266 left --include_dir as an undocumented synonym, and the same for
1267 --exclude-dir, though that is not available in GNU grep, at least as of
1268 release 2.5.4.
1269
1270 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1271 characters from a string of bytes) have been redefined so as not to use
1272 loops, in order to improve performance in some environments. At the same
1273 time, I abstracted some of the common code into auxiliary macros to save
1274 repetition (this should not affect the compiled code).
1275
1276 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1277 compile-time error is now given if \c is not followed by an ASCII
1278 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1279 different, and any byte value is allowed.)
1280
1281 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1282 START_OPTIMIZE option, which is now allowed at compile time - but just
1283 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1284 to pcregrep and other applications that have no direct access to PCRE
1285 options. The new /Y option in pcretest sets this option when calling
1286 pcre_compile().
1287
1288 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1289 back references. Groups containing recursive back references were forced to
1290 be atomic by that change, but in the case of named groups, the amount of
1291 memory required was incorrectly computed, leading to "Failed: internal
1292 error: code overflow". This has been fixed.
1293
1294 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1295 pcretest.c, to avoid build problems in some Borland environments.
1296
1297
1298 Version 8.10 25-Jun-2010
1299 ------------------------
1300
1301 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1302 THEN.
1303
1304 2. (*ACCEPT) was not working when inside an atomic group.
1305
1306 3. Inside a character class, \B is treated as a literal by default, but
1307 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1308 causes the error). The code is unchanged, but I tidied the documentation.
1309
1310 4. Inside a character class, PCRE always treated \R and \X as literals,
1311 whereas Perl faults them if its -w option is set. I have changed PCRE so
1312 that it faults them when PCRE_EXTRA is set.
1313
1314 5. Added support for \N, which always matches any character other than
1315 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1316
1317 6. When compiling pcregrep with newer versions of gcc which may have
1318 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1319 declared with attribute warn_unused_result" were given. Just casting the
1320 result to (void) does not stop the warnings; a more elaborate fudge is
1321 needed. I've used a macro to implement this.
1322
1323 7. Minor change to pcretest.c to avoid a compiler warning.
1324
1325 8. Added four artifical Unicode properties to help with an option to make
1326 \s etc use properties (see next item). The new properties are: Xan
1327 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1328
1329 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1330 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1331 this option. Modified pcretest to add /W to test this facility. Added
1332 REG_UCP to make it available via the POSIX interface.
1333
1334 10. Added --line-buffered to pcregrep.
1335
1336 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1337 studied, and the match started with a letter with a code point greater than
1338 127 whose first byte was different to the first byte of the other case of
1339 the letter, the other case of this starting letter was not recognized
1340 (#976).
1341
1342 12. If a pattern that was studied started with a repeated Unicode property
1343 test, for example, \p{Nd}+, there was the theoretical possibility of
1344 setting up an incorrect bitmap of starting bytes, but fortunately it could
1345 not have actually happened in practice until change 8 above was made (it
1346 added property types that matched character-matching opcodes).
1347
1348 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1349 possible starting bytes for non-anchored patterns.
1350
1351 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1352 \R, and also a number of cases that involve Unicode properties, both
1353 explicit and implicit when PCRE_UCP is set.
1354
1355 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1356 input, it could crash or give wrong results if characters with values
1357 greater than 0xc0 were present in the subject string. (Detail: it assumed
1358 UTF-8 input when processing these items.)
1359
1360 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1361 size_t is 64-bit (#991).
1362
1363 17. Added a check for running out of memory when PCRE is compiled with
1364 --disable-stack-for-recursion (#990).
1365
1366 18. If the last data line in a file for pcretest does not have a newline on
1367 the end, a newline was missing in the output.
1368
1369 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1370 less than 128) in its various bitmaps. However, there is a facility for
1371 generating tables according to the current locale when PCRE is compiled. It
1372 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1373 characters, are recognized by isspace() and therefore were getting set in
1374 these tables, and indeed these tables seem to approximate to ISO 8859. This
1375 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1376 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1377 which of course cannot start UTF-8 characters. I have changed the code so
1378 that only real ASCII characters (less than 128) and the correct starting
1379 bytes for UTF-8 encodings are set for characters greater than 127 when in
1380 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1381 altogether.)
1382
1383 20. Added the /T option to pcretest so as to be able to run tests with non-
1384 standard character tables, thus making it possible to include the tests
1385 used for 19 above in the standard set of tests.
1386
1387 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1388 reference to a subpattern the other side of a comment that contains an
1389 opening parenthesis caused either an internal compiling error, or a
1390 reference to the wrong subpattern.
1391
1392
1393 Version 8.02 19-Mar-2010
1394 ------------------------
1395
1396 1. The Unicode data tables have been updated to Unicode 5.2.0.
1397
1398 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1399 configured.
1400
1401 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1402 original author of that file, following a query about its status.
1403
1404 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1405 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1406
1407 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1408 quantifier applied to a forward-referencing subroutine call, could compile
1409 incorrect code or give the error "internal error: previously-checked
1410 referenced subpattern not found".
1411
1412 6. Both MS Visual Studio and Symbian OS have problems with initializing
1413 variables to point to external functions. For these systems, therefore,
1414 pcre_malloc etc. are now initialized to local functions that call the
1415 relevant global functions.
1416
1417 7. There were two entries missing in the vectors called coptable and poptable
1418 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1419 I've fixed the data, and added a kludgy way of testing at compile time that
1420 the lengths are correct (equal to the number of opcodes).
1421
1422 8. Following on from 7, I added a similar kludge to check the length of the
1423 eint vector in pcreposix.c.
1424
1425 9. Error texts for pcre_compile() are held as one long string to avoid too
1426 much relocation at load time. To find a text, the string is searched,
1427 counting zeros. There was no check for running off the end of the string,
1428 which could happen if a new error number was added without updating the
1429 string.
1430
1431 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1432
1433 11. \K was not working if it appeared in an atomic group or in a group that
1434 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1435 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1436 the assertion is positive, but not if it is negative.
1437
1438 12. Change 11 fortuitously reduced the size of the stack frame used in the
1439 "match()" function of pcre_exec.c by one pointer. Forthcoming
1440 implementation of support for (*MARK) will need an extra pointer on the
1441 stack; I have reserved it now, so that the stack frame size does not
1442 decrease.
1443
1444 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1445 item in branch that calls a recursion is a subroutine call - as in the
1446 second branch in the above example - was incorrectly given the compile-
1447 time error "recursive call could loop indefinitely" because pcre_compile()
1448 was not correctly checking the subroutine for matching a non-empty string.
1449
1450 14. The checks for overrunning compiling workspace could trigger after an
1451 overrun had occurred. This is a "should never occur" error, but it can be
1452 triggered by pathological patterns such as hundreds of nested parentheses.
1453 The checks now trigger 100 bytes before the end of the workspace.
1454
1455 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1456
1457
1458 Version 8.01 19-Jan-2010
1459 ------------------------
1460
1461 1. If a pattern contained a conditional subpattern with only one branch (in
1462 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1463 computed the wrong minimum data length (which is of course zero for such
1464 subpatterns). This could cause incorrect "no match" results.
1465
1466 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1467 the pattern is reset in the first branch, pcre_compile() failed with
1468 "internal error: code overflow at offset...". This happened only when
1469 the reset was to the original external option setting. (An optimization
1470 abstracts leading options settings into an external setting, which was the
1471 cause of this.)
1472
1473 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1474 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1475 assertion pattern did not match (meaning that the assertion was true), it
1476 was incorrectly treated as false if the SKIP had been reached during the
1477 matching. This also applied to assertions used as conditions.
1478
1479 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1480 assertion subpattern, including such a pattern used as a condition,
1481 unpredictable results occurred, instead of the error return
1482 PCRE_ERROR_DFA_UITEM.
1483
1484 5. The C++ GlobalReplace function was not working like Perl for the special
1485 situation when an empty string is matched. It now does the fancy magic
1486 stuff that is necessary.
1487
1488 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1489 removed. (These were left over from very, very early versions of PCRE.)
1490
1491 7. Some cosmetic changes to the code to make life easier when compiling it
1492 as part of something else:
1493
1494 (a) Change DEBUG to PCRE_DEBUG.
1495
1496 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1497 called "current" as "current_branch", to prevent a collision with the
1498 Linux macro when compiled as a kernel module.
1499
1500 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1501 prevent a collision with the Linux macro when compiled as a kernel
1502 module.
1503
1504 8. In pcre_compile() there are some checks for integer overflows that used to
1505 cast potentially large values to (double). This has been changed to that
1506 when building, a check for int64_t is made, and if it is found, it is used
1507 instead, thus avoiding the use of floating point arithmetic. (There is no
1508 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1509 double.
1510
1511 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1512 2005 (difference between two addresses compared to an unsigned value).
1513
1514 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1515 custom one, because of the following reported problem in Windows:
1516
1517 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1518 under Win32.
1519 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1520 therefore missing the function definition.
1521 - The compiler thus generates a "C" signature for the test function.
1522 - The linker fails to find the "C" function.
1523 - PCRE fails to configure if asked to do so against libbz2.
1524
1525 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1526 messages were output:
1527
1528 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1529 rerunning libtoolize, to keep the correct libtool macros in-tree.
1530 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1531
1532 I have done both of these things.
1533
1534 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1535 most of the time, it *can* run out if it is given a pattern that contains a
1536 runaway infinite recursion. I updated the discussion in the pcrestack man
1537 page.
1538
1539 13. Now that we have gone to the x.xx style of version numbers, the minor
1540 version may start with zero. Using 08 or 09 is a bad idea because users
1541 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1542 interpreted as invalid octal numbers. I've updated the previous comment in
1543 configure.ac, and also added a check that gives an error if 08 or 09 are
1544 used.
1545
1546 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1547 causing partial matching to fail when the end of the subject matched \W
1548 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1549
1550 15. There were some discrepancies between the declarations in pcre_internal.h
1551 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1552 their definitions. The declarations used "const uschar *" and the
1553 definitions used USPTR. Even though USPTR is normally defined as "const
1554 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1555 reported that: "This difference in casting confuses some C++ compilers, for
1556 example, SunCC recognizes above declarations as different functions and
1557 generates broken code for hbpcre." I have changed the declarations to use
1558 USPTR.
1559
1560 16. GNU libtool is named differently on some systems. The autogen.sh script now
1561 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1562 (FreeBSD).
1563
1564 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1565 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1566 comment: "Figure out how to create a longlong from a string: strtoll and
1567 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1568 instance, but it only takes 2 args instead of 3!"
1569
1570 18. A subtle bug concerned with back references has been fixed by a change of
1571 specification, with a corresponding code fix. A pattern such as
1572 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1573 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1574 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1575 same bug. Such groups have to be quantified to be useful, or contained
1576 inside another quantified group. (If there's no repetition, the reference
1577 can never match.) The problem arises because, having left the group and
1578 moved on to the rest of the pattern, a later failure that backtracks into
1579 the group uses the captured value from the final iteration of the group
1580 rather than the correct earlier one. I have fixed this in PCRE by forcing
1581 any group that contains a reference to itself to be an atomic group; that
1582 is, there cannot be any backtracking into it once it has completed. This is
1583 similar to recursive and subroutine calls.
1584
1585
1586 Version 8.00 19-Oct-09
1587 ----------------------
1588
1589 1. The table for translating pcre_compile() error codes into POSIX error codes
1590 was out-of-date, and there was no check on the pcre_compile() error code
1591 being within the table. This could lead to an OK return being given in
1592 error.
1593
1594 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1595 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1596 in a Windows environment.
1597
1598 3. The pcregrep --count option prints the count for each file even when it is
1599 zero, as does GNU grep. However, pcregrep was also printing all files when
1600 --files-with-matches was added. Now, when both options are given, it prints
1601 counts only for those files that have at least one match. (GNU grep just
1602 prints the file name in this circumstance, but including the count seems
1603 more useful - otherwise, why use --count?) Also ensured that the
1604 combination -clh just lists non-zero counts, with no names.
1605
1606 4. The long form of the pcregrep -F option was incorrectly implemented as
1607 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1608 but it seems right to fix it, and I didn't think it was worth preserving
1609 the old behaviour.
1610
1611 5. The command line items --regex=pattern and --regexp=pattern were not
1612 recognized by pcregrep, which required --regex pattern or --regexp pattern
1613 (with a space rather than an '='). The man page documented the '=' forms,
1614 which are compatible with GNU grep; these now work.
1615
1616 6. No libpcreposix.pc file was created for pkg-config; there was just
1617 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1618
1619 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1620 when UCP support is not needed, by modifying the Python script that
1621 generates it from Unicode data files. This should not matter if the module
1622 is correctly used as a library, but I received one complaint about 50K of
1623 unwanted data. My guess is that the person linked everything into his
1624 program rather than using a library. Anyway, it does no harm.
1625
1626 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1627 was a minimum greater than 1 for a wide character in a possessive
1628 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1629 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1630 character. Chaos in the form of incorrect output or a compiling loop could
1631 result.
1632
1633 9. The restrictions on what a pattern can contain when partial matching is
1634 requested for pcre_exec() have been removed. All patterns can now be
1635 partially matched by this function. In addition, if there are at least two
1636 slots in the offset vector, the offset of the earliest inspected character
1637 for the match and the offset of the end of the subject are set in them when
1638 PCRE_ERROR_PARTIAL is returned.
1639
1640 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1641 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1642 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1643 and may be more useful for multi-segment matching.
1644
1645 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1646 used to be given if ever the end of the subject was reached; now it is
1647 given only if matching could not proceed because another character was
1648 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1649 string "Z", which now yields "no match" instead of "partial match". In the
1650 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1651 final character ended with (*FAIL).
1652
1653 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1654 if the pattern had a "must contain" character that was already found in the
1655 earlier partial match, unless partial matching was again requested. For
1656 example, with the pattern /dog.(body)?/, the "must contain" character is
1657 "g". If the first part-match was for the string "dog", restarting with
1658 "sbody" failed. This bug has been fixed.
1659
1660 13. The string returned by pcre_dfa_exec() after a partial match has been
1661 changed so that it starts at the first inspected character rather than the
1662 first character of the match. This makes a difference only if the pattern
1663 starts with a lookbehind assertion or \b or \B (\K is not supported by
1664 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1665 matching functions compatible, and I think it's the right thing to do.
1666
1667 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1668 so that the demonstration program is easily available in environments where
1669 PCRE has not been installed from source.
1670
1671 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1672 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1673 library.
1674
1675 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1676 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1677 is not the first non-POSIX option to be added. Clearly some people find
1678 these options useful.
1679
1680 17. If a caller to the POSIX matching function regexec() passes a non-zero
1681 value for nmatch with a NULL value for pmatch, the value of
1682 nmatch is forced to zero.
1683
1684 18. RunGrepTest did not have a test for the availability of the -u option of
1685 the diff command, as RunTest does. It now checks in the same way as
1686 RunTest, and also checks for the -b option.
1687
1688 19. If an odd number of negated classes containing just a single character
1689 interposed, within parentheses, between a forward reference to a named
1690 subpattern and the definition of the subpattern, compilation crashed with
1691 an internal error, complaining that it could not find the referenced
1692 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1693 [The bug was that it was starting one character too far in when skipping
1694 over the character class, thus treating the ] as data rather than
1695 terminating the class. This meant it could skip too much.]
1696
1697 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1698 /g option in pcretest when the pattern contains \K, which makes it possible
1699 to have an empty string match not at the start, even when the pattern is
1700 anchored. Updated pcretest and pcredemo to use this option.
1701
1702 21. If the maximum number of capturing subpatterns in a recursion was greater
1703 than the maximum at the outer level, the higher number was returned, but
1704 with unset values at the outer level. The correct (outer level) value is
1705 now given.
1706
1707 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1708 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1709 make it do so. The string so far is captured, making this feature
1710 compatible with Perl.
1711
1712 23. The tests have been re-organized, adding tests 11 and 12, to make it
1713 possible to check the Perl 5.10 features against Perl 5.10.
1714
1715 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1716 pattern matches a fixed length string. PCRE did not allow this; now it
1717 does. Neither allows recursion.
1718
1719 25. I finally figured out how to implement a request to provide the minimum
1720 length of subject string that was needed in order to match a given pattern.
1721 (It was back references and recursion that I had previously got hung up
1722 on.) This code has now been added to pcre_study(); it finds a lower bound
1723 to the length of subject needed. It is not necessarily the greatest lower
1724 bound, but using it to avoid searching strings that are too short does give
1725 some useful speed-ups. The value is available to calling programs via
1726 pcre_fullinfo().
1727
1728 26. While implementing 25, I discovered to my embarrassment that pcretest had
1729 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1730 study optimizations had never been tested with that matching function.
1731 Oops. What is worse, even when it was passed study data, there was a bug in
1732 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1733 were also very few tests of studied patterns with pcre_dfa_exec().
1734
1735 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1736 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1737 on the other side of the coin, they are no longer allowed to have different
1738 names, because these cannot be distinguished in PCRE, and this has caused
1739 confusion. (This is a difference from Perl.)
1740
1741 28. When duplicate subpattern names are present (necessarily with different
1742 numbers, as required by 27 above), and a test is made by name in a
1743 conditional pattern, either for a subpattern having been matched, or for
1744 recursion in such a pattern, all the associated numbered subpatterns are
1745 tested, and the overall condition is true if the condition is true for any
1746 one of them. This is the way Perl works, and is also more like the way
1747 testing by number works.
1748
1749
1750 Version 7.9 11-Apr-09
1751 ---------------------
1752
1753 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1754 (pcretest), all targets were linked against these libraries. This included
1755 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1756 libraries. This caused unwanted dependencies to be created. This problem
1757 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1758 pcretest is linked with readline.
1759
1760 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1761 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1762 moved outside it again, because FALSE and TRUE are already defined in AIX,
1763 but BOOL is not.
1764
1765 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1766 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1767
1768 4. The pcregrep documentation said spaces were inserted as well as colons (or
1769 hyphens) following file names and line numbers when outputting matching
1770 lines. This is not true; no spaces are inserted. I have also clarified the
1771 wording for the --colour (or --color) option.
1772
1773 5. In pcregrep, when --colour was used with -o, the list of matching strings
1774 was not coloured; this is different to GNU grep, so I have changed it to be
1775 the same.
1776
1777 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1778 each matching line was coloured. Now it goes on to look for further matches
1779 of any of the test patterns, which is the same behaviour as GNU grep.
1780
1781 7. A pattern that could match an empty string could cause pcregrep to loop; it
1782 doesn't make sense to accept an empty string match in pcregrep, so I have
1783 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1784 seems to be how GNU grep behaves. [But see later change 40 for release
1785 8.33.]
1786
1787 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1788 start or after a newline", because the conditional assertion was not being
1789 correctly handled. The rule now is that both the assertion and what follows
1790 in the first alternative must satisfy the test.
1791
1792 9. If auto-callout was enabled in a pattern with a conditional group whose
1793 condition was an assertion, PCRE could crash during matching, both with
1794 pcre_exec() and pcre_dfa_exec().
1795
1796 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1797 used for matching.
1798
1799 11. Unicode property support in character classes was not working for
1800 characters (bytes) greater than 127 when not in UTF-8 mode.
1801
1802 12. Added the -M command line option to pcretest.
1803
1804 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1805
1806 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1807
1808 16. Added comments and documentation about mis-use of no_arg in the C++
1809 wrapper.
1810
1811 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1812 from Martin Jerabek that uses macro names for all relevant character and
1813 string constants.
1814
1815 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1816 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1817 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1818 these, but not everybody uses configure.
1819
1820 19. A conditional group that had only one branch was not being correctly
1821 recognized as an item that could match an empty string. This meant that an
1822 enclosing group might also not be so recognized, causing infinite looping
1823 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1824 with the subject "ab", where knowledge that the repeated group can match
1825 nothing is needed in order to break the loop.
1826
1827 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1828 exec(), but without supplying a callout function, matching went wrong.
1829
1830 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1831 leak if the size of the offset vector was greater than 30. When the vector
1832 is smaller, the saved offsets during recursion go onto a local stack
1833 vector, but for larger vectors malloc() is used. It was failing to free
1834 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1835 error, in fact).
1836
1837 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1838 heapframe that is used only when UTF-8 support is enabled. This caused no
1839 problem, but was untidy.
1840
1841 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1842 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1843 included within another project.
1844
1845 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1846 slightly modified by me:
1847
1848 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1849 not building pcregrep.
1850
1851 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1852 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1853
1854 25. Forward references, both numeric and by name, in patterns that made use of
1855 duplicate group numbers, could behave incorrectly or give incorrect errors,
1856 because when scanning forward to find the reference group, PCRE was not
1857 taking into account the duplicate group numbers. A pattern such as
1858 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1859
1860 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1861 the feature of a custom pointer more persuasive (as requested by a user).
1862
1863 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1864 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1865 pre-defined.
1866
1867 28. Added support for (*UTF8) at the start of a pattern.
1868
1869 29. Arrange for flags added by the "release type" setting in CMake to be shown
1870 in the configuration summary.
1871
1872
1873 Version 7.8 05-Sep-08
1874 ---------------------
1875
1876 1. Replaced UCP searching code with optimized version as implemented for Ad
1877 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1878 stage table and inline lookup instead of a function, giving speed ups of 2
1879 to 5 times on some simple patterns that I tested. Permission was given to
1880 distribute the MultiStage2.py script that generates the tables (it's not in
1881 the tarball, but is in the Subversion repository).
1882
1883 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1884 scripts.
1885
1886 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1887 a group with a zero qualifier. The result of the study could be incorrect,
1888 or the function might crash, depending on the pattern.
1889
1890 4. Caseless matching was not working for non-ASCII characters in back
1891 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1892 It now works when Unicode Property Support is available.
1893
1894 5. In pcretest, an escape such as \x{de} in the data was always generating
1895 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1896 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1897 truncation.
1898
1899 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1900
1901 7. Added two (int) casts to pcregrep when printing the difference of two
1902 pointers, in case they are 64-bit values.
1903
1904 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1905 test 2 if it fails.
1906
1907 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1908 and a #define of that name to empty if it is not externally set. This is to
1909 allow users of MSVC to set it if necessary.
1910
1911 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1912 the convenience functions in the pcre_get.c source file.
1913
1914 11. An option change at the start of a pattern that had top-level alternatives
1915 could cause overwriting and/or a crash. This command provoked a crash in
1916 some environments:
1917
1918 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1919
1920 This potential security problem was recorded as CVE-2008-2371.
1921
1922 12. For a pattern where the match had to start at the beginning or immediately
1923 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1924 pcre_dfa_exec() could read past the end of the passed subject if there was
1925 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1926 pcretest so that it places the subject at the end of its malloc-ed buffer.
1927
1928 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1929 exec() might read past the end of the data buffer in UTF-8 mode.
1930
1931 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1932 the data contained the byte 0x85 as part of a UTF-8 character within its
1933 first line. This applied both to normal and DFA matching.
1934
1935 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1936 /^[^d]*?$/8 failed to match "abc".
1937
1938 16. Added a missing copyright notice to pcrecpp_internal.h.
1939
1940 17. Make it more clear in the documentation that values returned from
1941 pcre_exec() in ovector are byte offsets, not character counts.
1942
1943 18. Tidied a few places to stop certain compilers from issuing warnings.
1944
1945 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1946 supplied by Stefan Weber. I made a further small update for 7.8 because
1947 there is a change of source arrangements: the pcre_searchfuncs.c module is
1948 replaced by pcre_ucd.c.
1949
1950
1951 Version 7.7 07-May-08
1952 ---------------------
1953
1954 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1955 a string to a long long, pretend we don't even have a long long." This is
1956 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1957
1958 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1959 pre-7.6 versions, which defined a global no_arg variable instead of putting
1960 it in the RE class. (See also #8 below.)
1961
1962 3. Remove a line of dead code, identified by coverity and reported by Nuno
1963 Lopes.
1964
1965 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1966
1967 (1) The include/exclude patterns were being applied to the whole pathnames
1968 of files, instead of just to the final components.
1969
1970 (2) If there was more than one level of directory, the subdirectories were
1971 skipped unless they satisfied the include/exclude conditions. This is
1972 inconsistent with GNU grep (and could even be seen as contrary to the
1973 pcregrep specification - which I improved to make it absolutely clear).
1974 The action now is always to scan all levels of directory, and just
1975 apply the include/exclude patterns to regular files.
1976
1977 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1978 --exclude_dir in the tests to avoid scanning .svn directories.
1979
1980 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1981 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1982 doesn't support NULs in patterns.
1983
1984 7. Added some missing "const"s to declarations of static tables in
1985 pcre_compile.c and pcre_dfa_exec.c.
1986
1987 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1988 caused by fix #2 above. (Subsequently also a second patch to fix the
1989 first patch. And a third patch - this was a messy problem.)
1990
1991 9. Applied Craig's patch to remove the use of push_back().
1992
1993 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1994 matching function regexec().
1995
1996 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1997 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1998 references. PCRE supports relative numbers with this syntax (I don't think
1999 Oniguruma does).
2000
2001 12. Previously, a group with a zero repeat such as (...){0} was completely
2002 omitted from the compiled regex. However, this means that if the group
2003 was called as a subroutine from elsewhere in the pattern, things went wrong
2004 (an internal error was given). Such groups are now left in the compiled
2005 pattern, with a new opcode that causes them to be skipped at execution
2006 time.
2007
2008 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2009 to the way PCRE behaves:
2010
2011 (a) A lone ] character is dis-allowed (Perl treats it as data).
2012
2013 (b) A back reference to an unmatched subpattern matches an empty string
2014 (Perl fails the current match path).
2015
2016 (c) A data ] in a character class must be notated as \] because if the
2017 first data character in a class is ], it defines an empty class. (In
2018 Perl it is not possible to have an empty class.) The empty class []
2019 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2020 The negative empty class [^] matches any one character, independently
2021 of the DOTALL setting.
2022
2023 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2024 non-existent subpattern following a character class starting with ']' and
2025 containing () gave an internal compiling error instead of "reference to
2026 non-existent subpattern". Fortunately, when the pattern did exist, the
2027 compiled code was correct. (When scanning forwards to check for the
2028 existence of the subpattern, it was treating the data ']' as terminating
2029 the class, so got the count wrong. When actually compiling, the reference
2030 was subsequently set up correctly.)
2031
2032 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2033 it was being rejected as not supported by pcre_dfa_exec(), even though
2034 other assertions are supported. I have made pcre_dfa_exec() support
2035 (*FAIL).
2036
2037 16. The implementation of 13c above involved the invention of a new opcode,
2038 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2039 cannot be changed at match time, I realized I could make a small
2040 improvement to matching performance by compiling OP_ALLANY instead of
2041 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2042 on the OP_ANY path.
2043
2044 17. Compiling pcretest on Windows with readline support failed without the
2045 following two fixes: (1) Make the unistd.h include conditional on
2046 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2047
2048 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2049 ncurses library to be included for pcretest when ReadLine support is
2050 requested, but also to allow for it to be overridden. This patch came from
2051 Daniel Bergström.
2052
2053 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2054 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2055 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2056 spotting this.
2057
2058
2059 Version 7.6 28-Jan-08
2060 ---------------------
2061
2062 1. A character class containing a very large number of characters with
2063 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2064 overflow.
2065
2066 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2067 HAVE_LONG_LONG is not defined.
2068
2069 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2070 bring them up to date and include new features. This patch includes:
2071
2072 - Fixed PH's badly added libz and libbz2 support.
2073 - Fixed a problem with static linking.
2074 - Added pcredemo. [But later removed - see 7 below.]
2075 - Fixed dftables problem and added an option.
2076 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2077 HAVE_LONG_LONG.
2078 - Added readline support for pcretest.
2079 - Added an listing of the option settings after cmake has run.
2080
2081 4. A user submitted a patch to Makefile that makes it easy to create
2082 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2083 Makefile.am that cause it to include this special target, without
2084 affecting anything else. Note that the same mingw target plus all
2085 the other distribution libraries and programs are now supported
2086 when configuring with CMake (see 6 below) instead of with
2087 Configure/Make.
2088
2089 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2090 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2091 exported in the Windows port". It has not yet been confirmed that the patch
2092 solves the problem, but it does no harm.
2093
2094 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2095 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2096 with CMake, and also correct the comment about stack recursion.
2097
2098 7. Remove the automatic building of pcredemo from the ./configure system and
2099 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2100 of a program that users should build themselves after PCRE is installed, so
2101 building it automatically is not really right. What is more, it gave
2102 trouble in some build environments.
2103
2104 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2105
2106
2107 Version 7.5 10-Jan-08
2108 ---------------------
2109
2110 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2111 values in parens when parsing an RE using the C++ wrapper."
2112
2113 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2114 Characters greater than 255 were excluded from the class instead of being
2115 included.
2116
2117 3. The same bug as (2) above applied to negated POSIX classes such as
2118 [:^space:].
2119
2120 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2121 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2122 I have changed it.
2123
2124 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2125 first named subpattern) and a construct such as (?&a) would reference the
2126 first named subpattern whose name started with "a" (in other words, the
2127 length check was missing). Both these problems are fixed. "Subpattern name
2128 expected" is now given for (?&) (a zero-length name), and this patch also
2129 makes it give the same error for \k'' (previously it complained that that
2130 was a reference to a non-existent subpattern).
2131
2132 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2133 this is right because (?- can be followed by option settings as well as by
2134 digits. I have, however, made the messages clearer.
2135
2136 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2137 than the number used in the conditional) now cause a compile-time error.
2138 This is actually not compatible with Perl, which accepts such patterns, but
2139 treats the conditional as always being FALSE (as PCRE used to), but it
2140 seems to me that giving a diagnostic is better.
2141
2142 8. Change "alphameric" to the more common word "alphanumeric" in comments
2143 and messages.
2144
2145 9. Fix two occurrences of "backslash" in comments that should have been
2146 "backspace".
2147
2148 10. Remove two redundant lines of code that can never be obeyed (their function
2149 was moved elsewhere).
2150
2151 11. The program that makes PCRE's Unicode character property table had a bug
2152 which caused it to generate incorrect table entries for sequences of
2153 characters that have the same character type, but are in different scripts.
2154 It amalgamated them into a single range, with the script of the first of
2155 them. In other words, some characters were in the wrong script. There were
2156 thirteen such cases, affecting characters in the following ranges:
2157
2158 U+002b0 - U+002c1
2159 U+0060c - U+0060d
2160 U+0061e - U+00612
2161 U+0064b - U+0065e
2162 U+0074d - U+0076d
2163 U+01800 - U+01805
2164 U+01d00 - U+01d77
2165 U+01d9b - U+01dbf
2166 U+0200b - U+0200f
2167 U+030fc - U+030fe
2168 U+03260 - U+0327f
2169 U+0fb46 - U+0fbb1
2170 U+10450 - U+1049d
2171
2172 12. The -o option (show only the matching part of a line) for pcregrep was not
2173 compatible with GNU grep in that, if there was more than one match in a
2174 line, it showed only the first of them. It now behaves in the same way as
2175 GNU grep.
2176
2177 13. If the -o and -v options were combined for pcregrep, it printed a blank
2178 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2179 does the same. The return code can be used to tell if there were any
2180 non-matching lines.
2181
2182 14. Added --file-offsets and --line-offsets to pcregrep.
2183
2184 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2185 infinitely looping recursion. The bug was that positive lookaheads were not
2186 being skipped when checking for a possible empty match (negative lookaheads
2187 and both kinds of lookbehind were skipped).
2188
2189 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2190 inclusion of <windows.h> to before rather than after the definition of
2191 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2192
2193 17. Specifying a possessive quantifier with a specific limit for a Unicode
2194 character property caused pcre_compile() to compile bad code, which led at
2195 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2196 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2197 caused the error; without that there was no problem.
2198
2199 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2200
2201 19. Added --enable-pcretest-libreadline.
2202
2203 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2204 RE::GlobalReplace(). As a result, the number of replacements returned was
2205 double what it should be. I removed one of the increments, but Craig sent a
2206 later patch that removed the other one (the right fix) and added unit tests
2207 that check the return values (which was not done before).
2208
2209 21. Several CMake things:
2210
2211 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2212 the names libpcre and libpcreposix, not just pcre and pcreposix.
2213
2214 (2) The above change means that pcretest and pcregrep are now correctly
2215 linked with the newly-built libraries, not previously installed ones.
2216
2217 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2218
2219 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2220 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2221 UTF-8 newline character). The key issue is that the pattern starts .*;
2222 this means that the match must be either at the beginning, or after a
2223 newline. The bug was in the code for advancing after a failed match and
2224 checking that the new position followed a newline. It was not taking
2225 account of UTF-8 characters correctly.
2226
2227 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2228 character classes. PCRE was not treating the sequence [:...:] as a
2229 character class unless the ... were all letters. Perl, however, seems to
2230 allow any characters between [: and :], though of course it rejects as
2231 unknown any "names" that contain non-letters, because all the known class
2232 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2233 for example, whereas PCRE did not - it did not recognize a POSIX character
2234 class. This seemed a bit dangerous, so the code has been changed to be
2235 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2236 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2237 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2238 Perl does, and where it didn't before.
2239
2240 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2241 Windows environments %n is disabled by default.
2242
2243
2244 Version 7.4 21-Sep-07
2245 ---------------------
2246
2247 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2248 means that a class such as [\s] counted as "explicit reference to CR or
2249 LF". That isn't really right - the whole point of the change was to try to
2250 help when there was an actual mention of one of the two characters. So now
2251 the change happens only if \r or \n (or a literal CR or LF) character is
2252 encountered.
2253
2254 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2255 of both had grown to the point where there were only 3 bits left.
2256 Fortunately, there was spare space in the data structure, and so I have
2257 moved the internal flags into a new 16-bit field to free up more option
2258 bits.
2259
2260 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2261 but did not set the internal JCHANGED flag - either of these is enough to
2262 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2263 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2264 start sets both bits.
2265
2266 4. Added options (at build time, compile time, exec time) to change \R from
2267 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2268
2269 5. doc/pcresyntax.html was missing from the distribution.
2270
2271 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2272 compatibility, even though it is no longer used.
2273
2274 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2275 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2276 windows.h file is present (where different names are used). [This was
2277 reversed later after testing - see 16 below.]
2278
2279 8. Changed all #include <config.h> to #include "config.h". There were also
2280 some further <pcre.h> cases that I changed to "pcre.h".
2281
2282 9. When pcregrep was used with the --colour option, it missed the line ending
2283 sequence off the lines that it output.
2284
2285 10. It was pointed out to me that arrays of string pointers cause lots of
2286 relocations when a shared library is dynamically loaded. A technique of
2287 using a single long string with a table of offsets can drastically reduce
2288 these. I have refactored PCRE in four places to do this. The result is
2289 dramatic:
2290
2291 Originally: 290
2292 After changing UCP table: 187
2293 After changing error message table: 43
2294 After changing table of "verbs" 36
2295 After changing table of Posix names 22
2296
2297 Thanks to the folks working on Gregex for glib for this insight.
2298
2299 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2300 unicode-properties was also set.
2301
2302 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2303
2304 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2305 checked only for CRLF.
2306
2307 14. Added casts to pcretest.c to avoid compiler warnings.
2308
2309 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2310
2311 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2312 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2313 entirely. This removes changes made in 7 above.
2314
2315 17. The CMake files have been updated, and there is now more information about
2316 building with CMake in the NON-UNIX-USE document.
2317
2318
2319 Version 7.3 28-Aug-07
2320 ---------------------
2321
2322 1. In the rejigging of the build system that eventually resulted in 7.1, the
2323 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2324 brackets there is not right, since it causes compilers to look for an
2325 installed pcre.h, not the version that is in the source that is being
2326 compiled (which of course may be different). I have changed it back to:
2327
2328 #include "pcre.h"
2329
2330 I have a vague recollection that the change was concerned with compiling in
2331 different directories, but in the new build system, that is taken care of
2332 by the VPATH setting the Makefile.
2333
2334 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2335 when the subject happened to end in the byte 0x85 (e.g. if the last
2336 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2337 characters but of course it shouldn't be taken as a newline when it is part
2338 of another character. The bug was that, for an unlimited repeat of . in
2339 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2340 characters when looking for a newline.
2341
2342 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2343
2344 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2345 in debug output.
2346
2347 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2348 long printing in the pcrecpp unittest when running under MinGW.
2349
2350 6. ESC_K was left out of the EBCDIC table.
2351
2352 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2353 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2354 limit also applies to "virtual nesting" when a pattern is recursive, and in
2355 this case 1000 isn't so big. I have been able to remove this limit at the
2356 expense of backing off one optimization in certain circumstances. Normally,
2357 when pcre_exec() would call its internal match() function recursively and
2358 immediately return the result unconditionally, it uses a "tail recursion"
2359 feature to save stack. However, when a subpattern that can match an empty
2360 string has an unlimited repetition quantifier, it no longer makes this
2361 optimization. That gives it a stack frame in which to save the data for
2362 checking that an empty string has been matched. Previously this was taken
2363 from the 1000-entry workspace that had been reserved. So now there is no
2364 explicit limit, but more stack is used.
2365
2366 8. Applied Daniel's patches to solve problems with the import/export magic
2367 syntax that is required for Windows, and which was going wrong for the
2368 pcreposix and pcrecpp parts of the library. These were overlooked when this
2369 problem was solved for the main library.
2370
2371 9. There were some crude static tests to avoid integer overflow when computing
2372 the size of patterns that contain repeated groups with explicit upper
2373 limits. As the maximum quantifier is 65535, the maximum group length was
2374 set at 30,000 so that the product of these two numbers did not overflow a
2375 32-bit integer. However, it turns out that people want to use groups that
2376 are longer than 30,000 bytes (though not repeat them that many times).
2377 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2378 made it possible to implement the integer overflow checks in a much more
2379 dynamic way, which I have now done. The artificial limitation on group
2380 length has been removed - we now have only the limit on the total length of
2381 the compiled pattern, which depends on the LINK_SIZE setting.
2382
2383 10. Fixed a bug in the documentation for get/copy named substring when
2384 duplicate names are permitted. If none of the named substrings are set, the
2385 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2386 empty string.
2387
2388 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2389 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2390 because the ] is interpreted as the first data character and the
2391 terminating ] is not found. PCRE has been made compatible with Perl in this
2392 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2393 cause memory overwriting.
2394
2395 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2396 string has been matched (to stop an infinite loop). It was not recognizing
2397 a conditional subpattern that could match an empty string if that
2398 subpattern was within another subpattern. For example, it looped when
2399 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2400 condition was not nested. This bug has been fixed.
2401
2402 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2403 past the start of the subject in the presence of bytes with the top bit
2404 set, for example "\x8aBCD".
2405
2406 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2407 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2408
2409 14. Optimized (?!) to (*FAIL).
2410
2411 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2412 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2413 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2414 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2415 does: it's just the validity check that is more restrictive.
2416
2417 16. Inserted checks for integer overflows during escape sequence (backslash)
2418 processing, and also fixed erroneous offset values for syntax errors during
2419 backslash processing.
2420
2421 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2422 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2423
2424 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2425 caused an overrun.
2426
2427 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2428 something other than just ASCII characters) inside a group that had an
2429 unlimited repeat caused a loop at compile time (while checking to see
2430 whether the group could match an empty string).
2431
2432 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2433 [\P{Any}] did so. (Error in the code for printing property names.)
2434
2435 21. An orphan \E inside a character class could cause a crash.
2436
2437 22. A repeated capturing bracket such as (A)? could cause a wild memory
2438 reference during compilation.
2439
2440 23. There are several functions in pcre_compile() that scan along a compiled
2441 expression for various reasons (e.g. to see if it's fixed length for look
2442 behind). There were bugs in these functions when a repeated \p or \P was
2443 present in the pattern. These operators have additional parameters compared
2444 with \d, etc, and these were not being taken into account when moving along
2445 the compiled data. Specifically:
2446
2447 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2448 length.
2449
2450 (b) An item such as \pL+ within a repeated group could cause crashes or
2451 loops.
2452
2453 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2454 "reference to non-existent subpattern" error.
2455
2456 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2457
2458 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2459 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2460
2461 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2462
2463 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2464 character were causing crashes (broken optimization).
2465
2466 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2467 \p or \P) caused a compile-time loop.
2468
2469 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2470 break. For example, the unstudied pattern [\r\n]A does not match the string
2471 "\r\nA" because change 7.0/46 below moves the current point on by two
2472 characters after failing to match at the start. However, the pattern \nA
2473 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2474 the same is true. There doesn't seem any very clean way out of this, but
2475 what I have chosen to do makes the common cases work: PCRE now takes note
2476 of whether there can be an explicit match for \r or \n anywhere in the
2477 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2478 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2479 pattern has explicit CR or LF references.
2480
2481 29. Added (*CR) etc for changing newline setting at start of pattern.
2482
2483
2484 Version 7.2 19-Jun-07
2485 ---------------------
2486
2487 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2488 which is apparently normally available under Windows.
2489
2490 2. Re-jig the pcregrep tests with different newline settings in an attempt
2491 to make them independent of the local environment's newline setting.
2492
2493 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2494
2495 4. Some of the "internals" tests were previously cut out when the link size
2496 was not 2, because the output contained actual offsets. The recent new
2497 "Z" feature of pcretest means that these can be cut out, making the tests
2498 usable with all link sizes.
2499
2500 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2501 stack recursion. This gives a massive performance boost under BSD, but just
2502 a small improvement under Linux. However, it saves one field in the frame
2503 in all cases.
2504
2505 6. Added more features from the forthcoming Perl 5.10:
2506
2507 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2508 recursion call. It refers to the nth most recently opened parentheses.
2509
2510 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2511 to be opened parentheses.
2512
2513 (c) Conditions that refer to capturing parentheses can be specified
2514 relatively, for example, (?(-2)... or (?(+3)...
2515
2516 (d) \K resets the start of the current match so that everything before
2517 is not part of it.
2518
2519 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2520
2521 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2522 reference syntax.
2523
2524 (g) (?| introduces a group in which the numbering of parentheses in each
2525 alternative starts with the same number.
2526
2527 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2528
2529 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2530 PCRE_INFO_JCHANGED.
2531
2532 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2533 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2534 for detecting groups that can match an empty string.
2535
2536 9. A pattern with a very large number of alternatives (more than several
2537 hundred) was running out of internal workspace during the pre-compile
2538 phase, where pcre_compile() figures out how much memory will be needed. A
2539 bit of new cunning has reduced the workspace needed for groups with
2540 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2541 workspace instead of running out of the 4096 that are available.
2542
2543 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2544
2545 11. Applied patch from Google to remove an optimization that didn't quite work.
2546 The report of the bug said:
2547
2548 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2549 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2550 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2551
2552 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2553 it matched the wrong number of bytes.
2554
2555
2556 Version 7.1 24-Apr-07
2557 ---------------------
2558
2559 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2560 that is more "standard", making use of automake and other Autotools. There
2561 is some re-arrangement of the files and adjustment of comments consequent
2562 on this.
2563
2564 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2565 for recursive directory scanning broke on some systems because the files
2566 are not scanned in any specific order and on different systems the order
2567 was different. A call to "sort" has been inserted into RunGrepTest for the
2568 approprate test as a short-term fix. In the longer term there may be an
2569 alternative.
2570
2571 3. I had an email from Eric Raymond about problems translating some of PCRE's
2572 man pages to HTML (despite the fact that I distribute HTML pages, some
2573 people do their own conversions for various reasons). The problems
2574 concerned the use of low-level troff macros .br and .in. I have therefore
2575 removed all such uses from the man pages (some were redundant, some could
2576 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2577 HTML has been updated to handle .nf/.fi and to complain if it encounters
2578 .br or .in.
2579
2580 4. Updated comments in configure.ac that get placed in config.h.in and also
2581 arranged for config.h to be included in the distribution, with the name
2582 config.h.generic, for the benefit of those who have to compile without
2583 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2584
2585 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2586 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2587 makevp.bat for the current PCRE, using the additional files
2588 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2589
2590 6. A Windows user reported a minor discrepancy with test 2, which turned out
2591 to be caused by a trailing space on an input line that had got lost in his
2592 copy. The trailing space was an accident, so I've just removed it.
2593
2594 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2595 that is needed.
2596
2597 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2598 as "const" (a) because they are and (b) because it helps the PHP
2599 maintainers who have recently made a script to detect big data structures
2600 in the php code that should be moved to the .rodata section. I remembered
2601 to update Builducptable as well, so it won't revert if ucptable.h is ever
2602 re-created.
2603
2604 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2605 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2606 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2607 support is not required. This saves 1.5-2K of code, which is important in
2608 some applications.
2609
2610 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2611 so as not to refer to the tables, even though these functions will never be
2612 called when UTF-8 support is disabled. Otherwise there are problems with a
2613 shared library.
2614
2615 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2616
2617 (a) It was defining its arguments as char * instead of void *.
2618
2619 (b) It was assuming that all moves were upwards in memory; this was true
2620 a long time ago when I wrote it, but is no longer the case.
2621
2622 The emulated memove() is provided for those environments that have neither
2623 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2624 is clearly not the case, as these two bugs were recently reported.
2625
2626 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2627 and Detrail to create the HTML documentation, the .txt form of the man
2628 pages, and it removes trailing spaces from listed files. It also creates
2629 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2630 case, it wraps all the #defines with #ifndefs. This script should be run
2631 before "make dist".
2632
2633 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2634 with Unicode property support.
2635
2636 (a) For a maximizing quantifier, if the two different cases of the
2637 character were of different lengths in their UTF-8 codings (there are
2638 some cases like this - I found 11), and the matching function had to
2639 back up over a mixture of the two cases, it incorrectly assumed they
2640 were both the same length.
2641
2642 (b) When PCRE was configured to use the heap rather than the stack for
2643 recursion during matching, it was not correctly preserving the data for
2644 the other case of a UTF-8 character when checking ahead for a match
2645 while processing a minimizing repeat. If the check also involved
2646 matching a wide character, but failed, corruption could cause an
2647 erroneous result when trying to check for a repeat of the original
2648 character.
2649
2650 13. Some tidying changes to the testing mechanism:
2651
2652 (a) The RunTest script now detects the internal link size and whether there
2653 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2654 values substituted by "configure". (The RunGrepTest script already did
2655 this for UTF-8.) The configure.ac script no longer substitutes the
2656 relevant variables.
2657
2658 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2659 with length and offset values. This means that the output is different
2660 for different internal link sizes. Test 2 is skipped for link sizes
2661 other than 2 because of this, bypassing the problem. Unfortunately,
2662 there was also a test in test 3 (the locale tests) that used /B and
2663 failed for link sizes other than 2. Rather than cut the whole test out,
2664 I have added a new /Z option to pcretest that replaces the length and
2665 offset values with spaces. This is now used to make test 3 independent
2666 of link size. (Test 2 will be tidied up later.)
2667
2668 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2669 segmentation fault instead of returning the appropriate error message.
2670
2671 15. In multiline mode when the newline sequence was set to "any", the pattern
2672 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2673 This doesn't seem right; it now treats the CRLF combination as the line
2674 ending, and so does not match in that case. It's only a pattern such as ^$
2675 that would hit this one: something like ^ABC$ would have failed after \r
2676 and then tried again after \r\n.
2677
2678 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2679 in an attempt to make files that differ only in their line terminators
2680 compare equal. This works on Linux.
2681
2682 17. Under certain error circumstances pcregrep might try to free random memory
2683 as it exited. This is now fixed, thanks to valgrind.
2684
2685 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2686 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2687 was because its rules for how to advance for /g after matching an empty
2688 string at the end of a line did not allow for this case. They now check for
2689 it specially.
2690
2691 20. pcretest is supposed to handle patterns and data of any length, by
2692 extending its buffers when necessary. It was getting this wrong when the
2693 buffer for a data line had to be extended.
2694
2695 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2696 CRLF as a newline sequence.
2697
2698 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2699 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2700 I have nevertheless tidied it up.
2701
2702 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2703
2704 24. Added a man page for pcre-config.
2705
2706
2707 Version 7.0 19-Dec-06
2708 ---------------------
2709
2710 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2711 moving to gcc 4.1.1.
2712
2713 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2714 sys/time.h, which is documented as needed for this function. It doesn't
2715 seem to matter on Linux, but it showed up on some releases of OS X.
2716
2717 3. It seems that there are systems where bytes whose values are greater than
2718 127 match isprint() in the "C" locale. The "C" locale should be the
2719 default when a C program starts up. In most systems, only ASCII printing
2720 characters match isprint(). This difference caused the output from pcretest
2721 to vary, making some of the tests fail. I have changed pcretest so that:
2722
2723 (a) When it is outputting text in the compiled version of a pattern, bytes
2724 other than 32-126 are always shown as hex escapes.
2725
2726 (b) When it is outputting text that is a matched part of a subject string,
2727 it does the same, unless a different locale has been set for the match
2728 (using the /L modifier). In this case, it uses isprint() to decide.
2729
2730 4. Fixed a major bug that caused incorrect computation of the amount of memory
2731 required for a compiled pattern when options that changed within the
2732 pattern affected the logic of the preliminary scan that determines the
2733 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2734 that the computed length was too small. The symptoms of this bug were
2735 either the PCRE error "internal error: code overflow" from pcre_compile(),
2736 or a glibc crash with a message such as "pcretest: free(): invalid next
2737 size (fast)". Examples of patterns that provoked this bug (shown in
2738 pcretest format) are:
2739
2740 /(?-x: )/x
2741 /(?x)(?-x: \s*#\s*)/
2742 /((?i)[\x{c0}])/8
2743 /(?i:[\x{c0}])/8
2744
2745 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2746 is now done differently.
2747
2748 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2749 wrapper classes; (b) implement a new function in the C++ scanner that is
2750 more efficient than the old way of doing things because it avoids levels of
2751 recursion in the regex matching; (c) add a paragraph to the documentation
2752 for the FullMatch() function.
2753
2754 6. The escape sequence \n was being treated as whatever was defined as
2755 "newline". Not only was this contrary to the documentation, which states
2756 that \n is character 10 (hex 0A), but it also went horribly wrong when
2757 "newline" was defined as CRLF. This has been fixed.
2758
2759 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2760 was being set to -1 for the "end of line" case (supposedly a value that no
2761 character can have). Though this value is never used (the check for end of
2762 line is "zero bytes in current character"), it caused compiler complaints.
2763 I've changed it to 0xffffffff.
2764
2765 8. In pcre_version.c, the version string was being built by a sequence of
2766 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2767 string (as it is for production releases) called a macro with an empty
2768 argument. The C standard says the result of this is undefined. The gcc
2769 compiler treats it as an empty string (which was what was wanted) but it is
2770 reported that Visual C gives an error. The source has been hacked around to
2771 avoid this problem.
2772
2773 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2774 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2775 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2776 of them did).
2777
2778 10. Originally, pcretest opened its input and output without "b"; then I was
2779 told that "b" was needed in some environments, so it was added for release
2780 5.0 to both the input and output. (It makes no difference on Unix-like
2781 systems.) Later I was told that it is wrong for the input on Windows. I've
2782 now abstracted the modes into two macros, to make it easier to fiddle with
2783 them, and removed "b" from the input mode under Windows.
2784
2785 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2786
2787 12. Added -help and --help to pcretest as an official way of being reminded
2788 of the options.
2789
2790 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2791 and pcrecpp.cc because they annoy compilers at high warning levels.
2792
2793 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2794
2795 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2796 scripts are not C programs :-) and which was not noticed because it works
2797 on Linux.
2798
2799 16. pcretest is supposed to handle any length of pattern and data line (as one
2800 line or as a continued sequence of lines) by extending its input buffer if
2801 necessary. This feature was broken for very long pattern lines, leading to
2802 a string of junk being passed to pcre_compile() if the pattern was longer
2803 than about 50K.
2804
2805 17. I have done a major re-factoring of the way pcre_compile() computes the
2806 amount of memory needed for a compiled pattern. Previously, there was code
2807 that made a preliminary scan of the pattern in order to do this. That was
2808 OK when PCRE was new, but as the facilities have expanded, it has become
2809 harder and harder to keep it in step with the real compile phase, and there
2810 have been a number of bugs (see for example, 4 above). I have now found a
2811 cunning way of running the real compile function in a "fake" mode that
2812 enables it to compute how much memory it would need, while actually only
2813 ever using a few hundred bytes of working memory and without too many
2814 tests of the mode. This should make future maintenance and development
2815 easier. A side effect of this work is that the limit of 200 on the nesting
2816 depth of parentheses has been removed (though this was never a serious
2817 limitation, I suspect). However, there is a downside: pcre_compile() now
2818 runs more slowly than before (30% or more, depending on the pattern). I
2819 hope this isn't a big issue. There is no effect on runtime performance.
2820
2821 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2822 newline (only possible for the last line of a file) and it was a
2823 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2824
2825 19. Added additional timing features to pcretest. (1) The -tm option now times
2826 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2827 separate command line item, by a number that specifies the number of
2828 repeats to use when timing. The default is 50000; this gives better
2829 precision, but takes uncomfortably long for very large patterns.
2830
2831 20. Extended pcre_study() to be more clever in cases where a branch of a
2832 subpattern has no definite first character. For example, (a*|b*)[cd] would
2833 previously give no result from pcre_study(). Now it recognizes that the
2834 first character must be a, b, c, or d.
2835
2836 21. There was an incorrect error "recursive call could loop indefinitely" if
2837 a subpattern (or the entire pattern) that was being tested for matching an
2838 empty string contained only one non-empty item after a nested subpattern.
2839 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2840 incorrectly, because the \d was being skipped in the check.
2841
2842 22. The pcretest program now has a new pattern option /B and a command line
2843 option -b, which is equivalent to adding /B to every pattern. This causes
2844 it to show the compiled bytecode, without the additional information that
2845 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2846 is the same as /B/I).
2847
2848 23. A new optimization is now able automatically to treat some sequences such
2849 as a*b as a*+b. More specifically, if something simple (such as a character
2850 or a simple class like \d) has an unlimited quantifier, and is followed by
2851 something that cannot possibly match the quantified thing, the quantifier
2852 is automatically "possessified".
2853
2854 24. A recursive reference to a subpattern whose number was greater than 39
2855 went wrong under certain circumstances in UTF-8 mode. This bug could also
2856 have affected the operation of pcre_study().
2857
2858 25. Realized that a little bit of performance could be had by replacing
2859 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2860
2861 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2862
2863 27. Possessive quantifiers such as a++ were previously implemented by turning
2864 them into atomic groups such as ($>a+). Now they have their own opcodes,
2865 which improves performance. This includes the automatically created ones
2866 from 23 above.
2867
2868 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2869 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2870 the first matched character to be a colon. This applied both to named and
2871 numbered groups.
2872
2873 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2874
2875 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2876 building PCRE on Windows easier, so I have included it in the distribution.
2877
2878 31. There is now a check in pcretest against a ridiculously large number being
2879 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2880 loop, the loop is abandoned.
2881
2882 32. Forward references to subpatterns in conditions such as (?(2)...) where
2883 subpattern 2 is defined later cause pcre_compile() to search forwards in
2884 the pattern for the relevant set of parentheses. This search went wrong
2885 when there were unescaped parentheses in a character class, parentheses
2886 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2887
2888 33. "Subroutine" calls and backreferences were previously restricted to
2889 referencing subpatterns earlier in the regex. This restriction has now
2890 been removed.
2891
2892 34. Added a number of extra features that are going to be in Perl 5.10. On the
2893 whole, these are just syntactic alternatives for features that PCRE had
2894 previously implemented using the Python syntax or my own invention. The
2895 other formats are all retained for compatibility.
2896
2897 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2898 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2899 also .NET compatible.
2900
2901 (b) A recursion or subroutine call to a named group can now be defined as
2902 (?&name) as well as (?P>name).
2903
2904 (c) A backreference to a named group can now be defined as \k<name> or
2905 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2906 5.10, are also .NET compatible.
2907
2908 (d) A conditional reference to a named group can now use the syntax
2909 (?(<name>) or (?('name') as well as (?(name).
2910
2911 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2912 groups (named and numbered) that are never evaluated inline, but can be
2913 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2914 is always false. There may be only one alternative in such a group.
2915
2916 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2917 as the simple (?(R). The condition is true only if the most recent
2918 recursion is that of the given number or name. It does not search out
2919 through the entire recursion stack.
2920
2921 (g) The escape \gN or \g{N} has been added, where N is a positive or
2922 negative number, specifying an absolute or relative reference.
2923
2924 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2925 some "unreachable code" warnings.
2926
2927 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2928 things, this adds five new scripts.
2929
2930 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2931 There were also incompatibilities regarding the handling of \Q..\E inside
2932 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2933 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2934
2935 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2936 matches an empty string, and forcibly breaks the loop. There were bugs in
2937 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2938 against aaaa the result was just "a" rather than "aaaa", for example. Two
2939 separate and independent bugs (that affected different cases) have been
2940 fixed.
2941
2942 39. Refactored the code to abolish the use of different opcodes for small
2943 capturing bracket numbers. This is a tidy that I avoided doing when I
2944 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2945 The new approach is not only tidier, it makes it possible to reduce the
2946 memory needed to fix the previous bug (38).
2947
2948 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2949 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2950 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2951 mode.
2952
2953 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2954 report.
2955
2956 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2957 copy construction and assignment in the C++ wrapper.
2958
2959 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2960 couple of bugs that could have given wrong results in the "--newline=crlf"
2961 case.
2962
2963 44. Added a number of casts and did some reorganization of signed/unsigned int
2964 variables following suggestions from Dair Grant. Also renamed the variable
2965 "this" as "item" because it is a C++ keyword.
2966
2967 45. Arranged for dftables to add
2968
2969 #include "pcre_internal.h"
2970
2971 to pcre_chartables.c because without it, gcc 4.x may remove the array
2972 definition from the final binary if PCRE is built into a static library and
2973 dead code stripping is activated.
2974
2975 46. For an unanchored pattern, if a match attempt fails at the start of a
2976 newline sequence, and the newline setting is CRLF or ANY, and the next two
2977 characters are CRLF, advance by two characters instead of one.
2978
2979
2980 Version 6.7 04-Jul-06
2981 ---------------------
2982
2983 1. In order to handle tests when input lines are enormously long, pcretest has
2984 been re-factored so that it automatically extends its buffers when
2985 necessary. The code is crude, but this _is_ just a test program. The
2986 default size has been increased from 32K to 50K.
2987
2988 2. The code in pcre_study() was using the value of the re argument before
2989 testing it for NULL. (Of course, in any sensible call of the function, it
2990 won't be NULL.)
2991
2992 3. The memmove() emulation function in pcre_internal.h, which is used on
2993 systems that lack both memmove() and bcopy() - that is, hardly ever -
2994 was missing a "static" storage class specifier.
2995
2996 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2997 containing an extended class (one that cannot be represented by a bitmap
2998 because it contains high-valued characters or Unicode property items, e.g.
2999 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3000 pattern, but PCRE should not loop if you do not (it no longer does).
3001 [Detail: two cases were found: (a) a repeated subpattern containing an
3002 extended class; (b) a recursive reference to a subpattern that followed a
3003 previous extended class. It wasn't skipping over the extended class
3004 correctly when UTF-8 mode was not set.]
3005
3006 5. A negated single-character class was not being recognized as fixed-length
3007 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3008 compile error "lookbehind assertion is not fixed length".
3009
3010 6. The RunPerlTest auxiliary script was showing an unexpected difference
3011 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3012 write a Perl script that can interpret lines of an input file either as
3013 byte characters or as UTF-8, which is what "perltest" was being required to
3014 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3015 can't do is switch easily at run time between having the "use utf8;" pragma
3016 or not. In the end, I fudged it by using the RunPerlTest script to insert
3017 "use utf8;" explicitly for the UTF-8 tests.
3018
3019 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3020 the end of the subject string, contrary to the documentation and to what
3021 Perl does. This was true of both matching functions. Now it matches only at
3022 the start of the subject and immediately after *internal* newlines.
3023
3024 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3025 a pointer to an int instead of a pointer to an unsigned long int. This
3026 caused problems on 64-bit systems.
3027
3028 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3029 instance of the 'standard' template library not being so standard".
3030
3031 10. There was no check on the number of named subpatterns nor the maximum
3032 length of a subpattern name. The product of these values is used to compute
3033 the size of the memory block for a compiled pattern. By supplying a very
3034 long subpattern name and a large number of named subpatterns, the size
3035 computation could be caused to overflow. This is now prevented by limiting
3036 the length of names to 32 characters, and the number of named subpatterns
3037 to 10,000.
3038
3039 11. Subpatterns that are repeated with specific counts have to be replicated in
3040 the compiled pattern. The size of memory for this was computed from the
3041 length of the subpattern and the repeat count. The latter is limited to
3042 65535, but there was no limit on the former, meaning that integer overflow
3043 could in principle occur. The compiled length of a repeated subpattern is
3044 now limited to 30,000 bytes in order to prevent this.
3045
3046 12. Added the optional facility to have named substrings with the same name.
3047
3048 13. Added the ability to use a named substring as a condition, using the
3049 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3050 are numbers (not recommended). Forward references are permitted.
3051
3052 14. Added forward references in named backreferences (if you see what I mean).
3053
3054 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3055 pattern could run off the end of the subject. For example, the pattern
3056 "(?s)(.{1,5})"8 did this with the subject "ab".
3057
3058 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3059 PCRE_CASELESS was set when matching characters that were quantified with ?
3060 or *.
3061
3062 17. A character class other than a single negated character that had a minimum
3063 but no maximum quantifier - for example [ab]{6,} - was not handled
3064 correctly by pce_dfa_exec(). It would match only one character.
3065
3066 18. A valid (though odd) pattern that looked like a POSIX character
3067 class but used an invalid character after [ (for example [[,abc,]]) caused
3068 pcre_compile() to give the error "Failed: internal error: code overflow" or
3069 in some cases to crash with a glibc free() error. This could even happen if
3070 the pattern terminated after [[ but there just happened to be a sequence of
3071 letters, a binary zero, and a closing ] in the memory that followed.
3072
3073 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3074 over the years. Originally (before any Unicode support), just the bottom 8
3075 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3076 output from "man perlunicode" includes this:
3077
3078 The regular expression compiler produces polymorphic opcodes. That
3079 is, the pattern adapts to the data and automatically switches to
3080 the Unicode character scheme when presented with Unicode data--or
3081 instead uses a traditional byte scheme when presented with byte
3082 data.
3083
3084 Sadly, a wide octal escape does not cause a switch, and in a string with
3085 no other multibyte characters, these octal escapes are treated as before.
3086 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3087 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3088 Unicode string.
3089
3090 I have not perpetrated such confusion in PCRE. Up till now, it took just
3091 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3092 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3093 translate to the appropriate multibyte character.
3094
3095 29. Applied some refactoring to reduce the number of warnings from Microsoft
3096 and Borland compilers. This has included removing the fudge introduced
3097 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3098 a warning about an unused variable.
3099
3100 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3101 characters since release 4.0, because Perl (from release 5.004) does not.
3102 [Or at least, is documented not to: some releases seem to be in conflict
3103 with the documentation.] However, when a pattern was studied with
3104 pcre_study() and all its branches started with \s, PCRE still included VT
3105 as a possible starting character. Of course, this did no harm; it just
3106 caused an unnecessary match attempt.
3107
3108 22. Removed a now-redundant internal flag bit that recorded the fact that case
3109 dependency changed within the pattern. This was once needed for "required
3110 byte" processing, but is no longer used. This recovers a now-scarce options
3111 bit. Also moved the least significant internal flag bit to the most-
3112 significant bit of the word, which was not previously used (hangover from
3113 the days when it was an int rather than a uint) to free up another bit for
3114 the future.
3115
3116 23. Added support for CRLF line endings as well as CR and LF. As well as the
3117 default being selectable at build time, it can now be changed at runtime
3118 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3119 specify that it is scanning data with non-default line endings.
3120
3121 24. Changed the definition of CXXLINK to make it agree with the definition of
3122 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3123
3124 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3125 recursions. This makes a big different to stack usage for some patterns.
3126
3127 26. If a subpattern containing a named recursion or subroutine reference such
3128 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3129 the space required for the compiled pattern went wrong and gave too small a
3130 value. Depending on the environment, this could lead to "Failed: internal
3131 error: code overflow at offset 49" or "glibc detected double free or
3132 corruption" errors.
3133
3134 27. Applied patches from Google (a) to support the new newline modes and (b) to
3135 advance over multibyte UTF-8 characters in GlobalReplace.
3136
3137 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3138 difference for some implementation of PCRE in some Windows version.
3139
3140 29. Added some extra testing facilities to pcretest:
3141
3142 \q<number> in a data line sets the "match limit" value
3143 \Q<number> in a data line sets the "match recursion limt" value
3144 -S <number> sets the stack size, where <number> is in megabytes
3145
3146 The -S option isn't available for Windows.
3147
3148
3149 Version 6.6 06-Feb-06
3150 ---------------------
3151
3152 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3153 in pcreposix.h. I have copied the definition from pcre.h.
3154
3155 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3156 because pcre.h is no longer a built file.
3157
3158 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3159 not normally included in the compiled code.
3160
3161
3162 Version 6.5 01-Feb-06
3163 ---------------------
3164
3165 1. When using the partial match feature with pcre_dfa_exec(), it was not
3166 anchoring the second and subsequent partial matches at the new starting
3167 point. This could lead to incorrect results. For example, with the pattern
3168 /1234/, partially matching against "123" and then "a4" gave a match.
3169
3170 2. Changes to pcregrep:
3171
3172 (a) All non-match returns from pcre_exec() were being treated as failures
3173 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3174 error message is output. Some extra information is given for the
3175 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3176 probably the only errors that are likely to be caused by users (by
3177 specifying a regex that has nested indefinite repeats, for instance).
3178 If there are more than 20 of these errors, pcregrep is abandoned.
3179
3180 (b) A binary zero was treated as data while matching, but terminated the
3181 output line if it was written out. This has been fixed: binary zeroes
3182 are now no different to any other data bytes.
3183
3184 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3185 used to set a locale for matching. The --locale=xxxx long option has
3186 been added (no short equivalent) to specify a locale explicitly on the
3187 pcregrep command, overriding the environment variables.
3188
3189 (d) When -B was used with -n, some line numbers in the output were one less
3190 than they should have been.
3191
3192 (e) Added the -o (--only-matching) option.
3193
3194 (f) If -A or -C was used with -c (count only), some lines of context were
3195 accidentally printed for the final match.
3196
3197 (g) Added the -H (--with-filename) option.
3198
3199 (h) The combination of options -rh failed to suppress file names for files
3200 that were found from directory arguments.
3201
3202 (i) Added the -D (--devices) and -d (--directories) options.
3203
3204 (j) Added the -F (--fixed-strings) option.
3205
3206 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3207
3208 (l) Added the --colo(u)r option.
3209
3210 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3211 is not present by default.
3212
3213 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3214 items such as (?R) or (?1), when the recursion could match a number of
3215 alternatives. If it matched one of the alternatives, but subsequently,
3216 outside the recursion, there was a failure, the code tried to back up into
3217 the recursion. However, because of the way PCRE is implemented, this is not
3218 possible, and the result was an incorrect result from the match.
3219
3220 In order to prevent this happening, the specification of recursion has
3221 been changed so that all such subpatterns are automatically treated as
3222 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3223
3224 4. I had overlooked the fact that, in some locales, there are characters for
3225 which isalpha() is true but neither isupper() nor islower() are true. In
3226 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3227 and ordfeminine) are like this. This affected the treatment of \w and \W
3228 when they appeared in character classes, but not when they appeared outside
3229 a character class. The bit map for "word" characters is now created
3230 separately from the results of isalnum() instead of just taking it from the
3231 upper, lower, and digit maps. (Plus the underscore character, of course.)
3232
3233 5. The above bug also affected the handling of POSIX character classes such as
3234 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3235 permanent tables. Instead, the bit maps for such a class were previously
3236 created as the appropriate unions of the upper, lower, and digit bitmaps.
3237 Now they are created by subtraction from the [[:word:]] class, which has
3238 its own bitmap.
3239
3240 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3241 It is created by subtracting the vertical space characters (\x09, \x0a,
3242 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3243 subtraction was done in the overall bitmap for a character class, meaning
3244 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3245 be recognized. This bug has been fixed.
3246
3247 7. Patches from the folks at Google:
3248
3249 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3250 real life, but is still worth protecting against".
3251
3252 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3253 regular expressions".
3254
3255 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3256 have it.
3257
3258 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3259 "configure" and the latter not, in order to fix a problem somebody had
3260 with compiling the Arg class on HP-UX.
3261
3262 (e) Improve the error-handling of the C++ wrapper a little bit.
3263
3264 (f) New tests for checking recursion limiting.
3265
3266 8. The pcre_memmove() function, which is used only if the environment does not
3267 have a standard memmove() function (and is therefore rarely compiled),
3268 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3269 returning a result (though PCRE never actually uses the result).
3270
3271 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3272 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3273 returned instead of calling malloc() with an overflowing number that would
3274 most likely cause subsequent chaos.
3275
3276 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3277
3278 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3279 with this option is matched, the nmatch and pmatch options of regexec() are
3280 ignored.
3281
3282 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3283 provided in case anyone wants to the the POSIX interface with UTF-8
3284 strings.
3285
3286 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3287 C++ linking (needed for some HP-UX environments).
3288
3289 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3290 (unused parameter) and in the pcre_printint() function (omitted "default"
3291 switch label when the default is to do nothing).
3292
3293 15. Added some code to make it possible, when PCRE is compiled as a C++
3294 library, to replace subject pointers for pcre_exec() with a smart pointer
3295 class, thus making it possible to process discontinuous strings.
3296
3297 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3298 much the same function. They were added by different people who were trying
3299 to make PCRE easy to compile on non-Unix systems. It has been suggested
3300 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3301 for compiling on Windows systems. I have therefore replaced it with
3302 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3303 defaults to "extern" for C or "extern C" for C++, which works fine on
3304 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3305 SCOPE with something explicit in config.h. In addition:
3306
3307 (a) pcreposix.h still had just "extern" instead of either of these macros;
3308 I have replaced it with PCRE_DATA_SCOPE.
3309
3310 (b) Functions such as _pcre_xclass(), which are internal to the library,
3311 but external in the C sense, all had PCRE_EXPORT in their definitions.
3312 This is apparently wrong for the Windows case, so I have removed it.
3313 (It makes no difference on Unix-like systems.)
3314
3315 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3316 of recursive calls to match(). This is different to MATCH_LIMIT because
3317 that limits the total number of calls to match(), not all of which increase
3318 the depth of recursion. Limiting the recursion depth limits the amount of
3319 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3320 when PCRE is compiled, and changed at run time. A patch from Google adds
3321 this functionality to the C++ interface.
3322
3323 18. Changes to the handling of Unicode character properties:
3324
3325 (a) Updated the table to Unicode 4.1.0.
3326
3327 (b) Recognize characters that are not in the table as "Cn" (undefined).
3328
3329 (c) I revised the way the table is implemented to a much improved format
3330 which includes recognition of ranges. It now supports the ranges that
3331 are defined in UnicodeData.txt, and it also amalgamates other
3332 characters into ranges. This has reduced the number of entries in the
3333 table from around 16,000 to around 3,000, thus reducing its size
3334 considerably. I realized I did not need to use a tree structure after
3335 all - a binary chop search is just as efficient. Having reduced the
3336 number of entries, I extended their size from 6 bytes to 8 bytes to
3337 allow for more data.
3338
3339 (d) Added support for Unicode script names via properties such as \p{Han}.
3340
3341 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3342 matching that character.
3343
3344 20. When matching a repeated Unicode property with a minimum greater than zero,
3345 (for example \pL{2,}), PCRE could look past the end of the subject if it
3346 reached it while seeking the minimum number of characters. This could
3347 happen only if some of the characters were more than one byte long, because
3348 there is a check for at least the minimum number of bytes.
3349
3350 21. Refactored the implementation of \p and \P so as to be more general, to
3351 allow for more different types of property in future. This has changed the
3352 compiled form incompatibly. Anybody with saved compiled patterns that use
3353 \p or \P will have to recompile them.
3354
3355 22. Added "Any" and "L&" to the supported property types.
3356
3357 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3358 but give a compile time error if the value is greater than 0xff.
3359
3360 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3361 accidentally not being installed or uninstalled.
3362
3363 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3364 made were to insert the current release number. This seemed silly, because
3365 it made things harder for people building PCRE on systems that don't run
3366 "configure". I have turned pcre.h into a distributed file, no longer built
3367 by "configure", with the version identification directly included. There is
3368 no longer a pcre.h.in file.
3369
3370 However, this change necessitated a change to the pcre-config script as
3371 well. It is built from pcre-config.in, and one of the substitutions was the
3372 release number. I have updated configure.ac so that ./configure now finds
3373 the release number by grepping pcre.h.
3374
3375 26. Added the ability to run the tests under valgrind.
3376
3377
3378 Version 6.4 05-Sep-05
3379 ---------------------
3380
3381 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3382 "--" to be printed when multiple files were scanned, even when none of the
3383 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3384 consider it to be a bug, and have restored the previous behaviour.
3385
3386 2. A couple of code tidies to get rid of compiler warnings.
3387
3388 3. The pcretest program used to cheat by referring to symbols in the library
3389 whose names begin with _pcre_. These are internal symbols that are not
3390 really supposed to be visible externally, and in some environments it is
3391 possible to suppress them. The cheating is now confined to including
3392 certain files from the library's source, which is a bit cleaner.
3393
3394 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3395 file's purpose clearer.
3396
3397 5. Reorganized pcre_ucp_findchar().
3398
3399
3400 Version 6.3 15-Aug-05
3401 ---------------------
3402
3403 1. The file libpcre.pc.in did not have general read permission in the tarball.
3404
3405 2. There were some problems when building without C++ support:
3406
3407 (a) If C++ support was not built, "make install" and "make test" still
3408 tried to test it.
3409
3410 (b) There were problems when the value of CXX was explicitly set. Some
3411 changes have been made to try to fix these, and ...
3412
3413 (c) --disable-cpp can now be used to explicitly disable C++ support.
3414
3415 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3416 backslash in a target when C++ was disabled. This confuses some
3417 versions of "make", apparently. Using an intermediate variable solves
3418 this. (Same for CPP_LOBJ.)
3419
3420 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3421 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3422 necessary on certain architectures.
3423
3424 4. Added a setting of -export-symbols-regex to the link command to remove
3425 those symbols that are exported in the C sense, but actually are local
3426 within the library, and not documented. Their names all begin with
3427 "_pcre_". This is not a perfect job, because (a) we have to except some
3428 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3429 available (and never for static libraries). I have made a note to try to
3430 find a way round (a) in the future.
3431
3432
3433 Version 6.2 01-Aug-05
3434 ---------------------
3435
3436 1. There was no test for integer overflow of quantifier values. A construction
3437 such as {1111111111111111} would give undefined results. What is worse, if
3438 a minimum quantifier for a parenthesized subpattern overflowed and became
3439 negative, the calculation of the memory size went wrong. This could have
3440 led to memory overwriting.
3441
3442 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3443
3444 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3445 operating environments where this matters.
3446
3447 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3448 PCRE options from within the C++ wrapper.
3449
3450 5. Named capturing subpatterns were not being correctly counted when a pattern
3451 was compiled. This caused two problems: (a) If there were more than 100
3452 such subpatterns, the calculation of the memory needed for the whole
3453 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3454 back references of the form \12, where the number was greater than 9, were
3455 not recognized as back references, even though there were sufficient
3456 previous subpatterns.
3457
3458 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3459 versions of gcc, e.g. 2.95.4.
3460
3461
3462 Version 6.1 21-Jun-05
3463 ---------------------
3464
3465 1. There was one reference to the variable "posix" in pcretest.c that was not
3466 surrounded by "#if !defined NOPOSIX".
3467
3468 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3469 the cross-check on the old pcre_info() function, for the benefit of the
3470 cut-down version of PCRE that is currently imported into Exim.
3471
3472 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3473 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3474 patterns that start (?i) but I don't think that matters. The use of (?i) is
3475 just an example; this all applies to the other options as well.
3476
3477 4. Since libtool seems to echo the compile commands it is issuing, the output
3478 from "make" can be reduced a bit by putting "@" in front of each libtool
3479 compile command.
3480
3481 5. Patch from the folks at Google for configure.in to be a bit more thorough
3482 in checking for a suitable C++ installation before trying to compile the
3483 C++ stuff. This should fix a reported problem when a compiler was present,
3484 but no suitable headers.
3485
3486 6. The man pages all had just "PCRE" as their title. I have changed them to
3487 be the relevant file name. I have also arranged that these names are
3488 retained in the file doc/pcre.txt, which is a concatenation in text format
3489 of all the man pages except the little individual ones for each function.
3490
3491 7. The NON-UNIX-USE file had not been updated for the different set of source
3492 files that come with release 6. I also added a few comments about the C++
3493 wrapper.
3494
3495
3496 Version 6.0 07-Jun-05
3497 ---------------------
3498
3499 1. Some minor internal re-organization to help with my DFA experiments.
3500
3501 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3502 didn't matter for the library itself when fully configured, but did matter
3503 when compiling without UCP support, or within Exim, where the ucp files are
3504 not imported.
3505
3506 3. Refactoring of the library code to split up the various functions into
3507 different source modules. The addition of the new DFA matching code (see
3508 below) to a single monolithic source would have made it really too
3509 unwieldy, quite apart from causing all the code to be include in a
3510 statically linked application, when only some functions are used. This is
3511 relevant even without the DFA addition now that patterns can be compiled in
3512 one application and matched in another.
3513
3514 The downside of splitting up is that there have to be some external
3515 functions and data tables that are used internally in different modules of
3516 the library but which are not part of the API. These have all had their
3517 names changed to start with "_pcre_" so that they are unlikely to clash
3518 with other external names.
3519
3520 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3521 a different (DFA) algorithm. Although it is slower than the original
3522 function, it does have some advantages for certain types of matching
3523 problem.
3524
3525 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3526 including restarting after a partial match.
3527
3528 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3529 defined when compiling for Windows was sent to me. I have put it into the
3530 code, though I have no means of testing or verifying it.
3531
3532 7. Added the pcre_refcount() auxiliary function.
3533
3534 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3535 match before or at the first newline in the subject string. In pcretest,
3536 the /f option on a pattern can be used to set this.
3537
3538 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3539 would behave wrongly. This has been present in PCRE since release 4.0.
3540
3541 10. A number of changes to the pcregrep command:
3542
3543 (a) Refactored how -x works; insert ^(...)$ instead of setting
3544 PCRE_ANCHORED and checking the length, in preparation for adding
3545 something similar for -w.
3546
3547 (b) Added the -w (match as a word) option.
3548
3549 (c) Refactored the way lines are read and buffered so as to have more
3550 than one at a time available.
3551
3552 (d) Implemented a pcregrep test script.
3553
3554 (e) Added the -M (multiline match) option. This allows patterns to match
3555 over several lines of the subject. The buffering ensures that at least
3556 8K, or the rest of the document (whichever is the shorter) is available
3557 for matching (and similarly the previous 8K for lookbehind assertions).
3558
3559 (f) Changed the --help output so that it now says
3560
3561 -w, --word-regex(p)
3562
3563 instead of two lines, one with "regex" and the other with "regexp"
3564 because that confused at least one person since the short forms are the
3565 same. (This required a bit of code, as the output is generated
3566 automatically from a table. It wasn't just a text change.)
3567
3568 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3569 option but starts with a hyphen. Could be a pattern or a path name
3570 starting with a hyphen, for instance.
3571
3572 (h) "-" can be given as a file name to represent stdin.
3573
3574 (i) When file names are being printed, "(standard input)" is used for
3575 the standard input, for compatibility with GNU grep. Previously
3576 "<stdin>" was used.
3577
3578 (j) The option --label=xxx can be used to supply a name to be used for
3579 stdin when file names are being printed. There is no short form.
3580
3581 (k) Re-factored the options decoding logic because we are going to add
3582 two more options that take data. Such options can now be given in four
3583 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3584
3585 (l) Added the -A, -B, and -C options for requesting that lines of context
3586 around matches be printed.
3587
3588 (m) Added the -L option to print the names of files that do not contain
3589 any matching lines, that is, the complement of -l.
3590
3591 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3592 continue to scan other files.
3593
3594 (o) The -s option was incorrectly implemented. For compatibility with other
3595 greps, it now suppresses the error message for a non-existent or non-
3596 accessible file (but not the return code). There is a new option called
3597 -q that suppresses the output of matching lines, which was what -s was
3598 previously doing.
3599
3600 (p) Added --include and --exclude options to specify files for inclusion
3601 and exclusion when recursing.
3602
3603 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3604 Hopefully, it now does.
3605
3606 12. Missing cast in pcre_study().
3607
3608 13. Added an "uninstall" target to the makefile.
3609
3610 14. Replaced "extern" in the function prototypes in Makefile.in with
3611 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3612 world, but is set differently for Windows.
3613
3614 15. Added a second compiling function called pcre_compile2(). The only
3615 difference is that it has an extra argument, which is a pointer to an
3616 integer error code. When there is a compile-time failure, this is set
3617 non-zero, in addition to the error test pointer being set to point to an
3618 error message. The new argument may be NULL if no error number is required
3619 (but then you may as well call pcre_compile(), which is now just a
3620 wrapper). This facility is provided because some applications need a
3621 numeric error indication, but it has also enabled me to tidy up the way
3622 compile-time errors are handled in the POSIX wrapper.
3623
3624 16. Added VPATH=.libs to the makefile; this should help when building with one
3625 prefix path and installing with another. (Or so I'm told by someone who
3626 knows more about this stuff than I do.)
3627
3628 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3629 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3630 match everything, including newlines. This is not POSIX-compatible, but
3631 somebody wanted the feature. From pcretest it can be activated by using
3632 both the P and the s flags.
3633
3634 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3635
3636 19. libpcre.pc was being incorrectly installed as executable.
3637
3638 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3639 it now also looks for '\r' so that it will work unmodified on Windows.
3640
3641 21. Added Google's contributed C++ wrapper to the distribution.
3642
3643 22. Added some untidy missing memory free() calls in pcretest, to keep
3644 Electric Fence happy when testing.
3645
3646
3647
3648 Version 5.0 13-Sep-04
3649 ---------------------
3650
3651 1. Internal change: literal characters are no longer packed up into items
3652 containing multiple characters in a single byte-string. Each character
3653 is now matched using a separate opcode. However, there may be more than one
3654 byte in the character in UTF-8 mode.
3655
3656 2. The pcre_callout_block structure has two new fields: pattern_position and
3657 next_item_length. These contain the offset in the pattern to the next match
3658 item, and its length, respectively.
3659
3660 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3661 insertion of callouts before each pattern item. Added the /C option to
3662 pcretest to make use of this.
3663
3664 4. On the advice of a Windows user, the lines
3665
3666 #if defined(_WIN32) || defined(WIN32)
3667 _setmode( _fileno( stdout ), 0x8000 );
3668 #endif /* defined(_WIN32) || defined(WIN32) */
3669
3670 have been added to the source of pcretest. This apparently does useful
3671 magic in relation to line terminators.
3672
3673 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3674 for the benefit of those environments where the "b" makes a difference.
3675
3676 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3677 to know about it. I have put a hack into configure.in that adds in code
3678 to set GCC=yes if CC=icc. This seems to end up at a point in the
3679 generated configure script that is early enough to affect the setting of
3680 compiler options, which is what is needed, but I have no means of testing
3681 whether it really works. (The user who reported this had patched the
3682 generated configure script, which of course I cannot do.)
3683
3684 LATER: After change 22 below (new libtool files), the configure script
3685 seems to know about icc (and also ecc). Therefore, I have commented out
3686 this hack in configure.in.
3687
3688 7. Added support for pkg-config (2 patches were sent in).
3689
3690 8. Negated POSIX character classes that used a combination of internal tables
3691 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3692 [[:^ascii]]. Typically, they would match almost any characters. The other
3693 POSIX classes were not broken in this way.
3694
3695 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3696 to find the match, as PCRE was deluded into thinking that the match had to
3697 start at the start point or following a newline. The same bug applied to
3698 patterns with negative forward assertions or any backward assertions
3699 preceding ".*" at the start, unless the pattern required a fixed first
3700 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3701
3702 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3703 starting at the last subject character, bytes beyond the end of the subject
3704 string were read.
3705
3706 11. Renamed the variable "class" as "classbits" to make life easier for C++
3707 users. (Previously there was a macro definition, but it apparently wasn't
3708 enough.)
3709
3710 12. Added the new field "tables" to the extra data so that tables can be passed
3711 in at exec time, or the internal tables can be re-selected. This allows
3712 a compiled regex to be saved and re-used at a later time by a different
3713 program that might have everything at different addresses.
3714
3715 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3716 -R library as well as a -L library.
3717
3718 14. The debugging options of pcretest (-d on the command line or D on a
3719 pattern) showed incorrect output for anything following an extended class
3720 that contained multibyte characters and which was followed by a quantifier.
3721
3722 15. Added optional support for general category Unicode character properties
3723 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3724 support. It adds about 90K to the size of the library. The meanings of the
3725 inbuilt class escapes such as \d and \s have NOT been changed.
3726
3727 16. Updated pcredemo.c to include calls to free() to release the memory for the
3728 compiled pattern.
3729
3730 17. The generated file chartables.c was being created in the source directory
3731 instead of in the building directory. This caused the build to fail if the
3732 source directory was different from the building directory, and was
3733 read-only.
3734
3735 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3736 file. No doubt somebody will tell me if they don't make sense... Also added
3737 Dan Mooney's comments about building on OpenVMS.
3738
3739 19. Added support for partial matching via the PCRE_PARTIAL option for
3740 pcre_exec() and the \P data escape in pcretest.
3741
3742 20. Extended pcretest with 3 new pattern features:
3743
3744 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3745 write the compiled pattern to the file whose name is "rest-of-line".
3746 This is a straight binary dump of the data, with the saved pointer to
3747 the character tables forced to be NULL. The study data, if any, is
3748 written too. After writing, pcretest reads a new pattern.
3749
3750 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3751 compiled pattern from the given file. There must not be any
3752 occurrences of "<" in the file name (pretty unlikely); if there are,
3753 pcretest will instead treat the initial "<" as a pattern delimiter.
3754 After reading in the pattern, pcretest goes on to read data lines as
3755 usual.
3756
3757 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3758 and 16-bit fields in a compiled pattern, to simulate a pattern that
3759 was compiled on a host of opposite endianness.
3760
3761 21. The pcre-exec() function can now cope with patterns that were compiled on
3762 hosts of opposite endianness, with this restriction:
3763
3764 As for any compiled expression that is saved and used later, the tables
3765 pointer field cannot be preserved; the extra_data field in the arguments
3766 to pcre_exec() should be used to pass in a tables address if a value
3767 other than the default internal tables were used at compile time.
3768
3769 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3770 now diagnosed as an error. Previously, most of the time, a negative number
3771 would have been treated as zero, but if in addition "ovector" was passed as
3772 NULL, a crash could occur.
3773
3774 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3775 new versions from the libtool 1.5 distribution (the last one is a copy of
3776 a file called libtool.m4). This seems to have fixed the need to patch
3777 "configure" to support Darwin 1.3 (which I used to do). However, I still
3778 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3779 workstation).
3780
3781 24. Changed the PCRE licence to be the more standard "BSD" licence.
3782
3783
3784 Version 4.5 01-Dec-03
3785 ---------------------
3786
3787 1. There has been some re-arrangement of the code for the match() function so
3788 that it can be compiled in a version that does not call itself recursively.
3789 Instead, it keeps those local variables that need separate instances for
3790 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3791 needs to "recurse". Keeping track of where control must go is done by means
3792 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3793 hide most of the details from the main code, and operates only if
3794 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3795 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3796 operating.
3797
3798 To make it easier for callers to provide specially tailored get/free
3799 functions for this usage, two new functions, pcre_stack_malloc, and
3800 pcre_stack_free, are used. They are always called in strict stacking order,
3801 and the size of block requested is always the same.
3802
3803 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3804 PCRE has been compiled to use the stack or the heap for recursion. The
3805 -C option of pcretest uses this to show which version is compiled.
3806
3807 A new data escape \S, is added to pcretest; it causes the amounts of store
3808 obtained and freed by both kinds of malloc/free at match time to be added
3809 to the output.
3810
3811 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3812 what's available on my current Linux desktop machine.
3813
3814 3. When matching a UTF-8 string, the test for a valid string at the start has
3815 been extended. If start_offset is not zero, PCRE now checks that it points
3816 to a byte that is the start of a UTF-8 character. If not, it returns
3817 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3818 this is necessary because there may be backward assertions in the pattern.
3819 When matching the same subject several times, it may save resources to use
3820 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3821
3822 4. The code for checking the validity of UTF-8 strings has been tightened so
3823 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3824 containing "overlong sequences".
3825
3826 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3827 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3828 should have been "&", but it just so happened that all the cases this let
3829 through by mistake were picked up later in the function.
3830
3831 6. I had used a variable called "isblank" - this is a C99 function, causing
3832 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3833
3834 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3835 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3836 so the version is shown; (c) stop "make test" echoing "./RunTest".
3837
3838 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3839
3840 9. The prototype for memmove() for systems that don't have it was using
3841 size_t, but the inclusion of the header that defines size_t was later. I've
3842 moved the #includes for the C headers earlier to avoid this.
3843
3844 10. Added some adjustments to the code to make it easier to compiler on certain
3845 special systems:
3846
3847 (a) Some "const" qualifiers were missing.
3848 (b) Added the macro EXPORT before all exported functions; by default this
3849 is defined to be empty.
3850 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3851 that it reads its output file name as an argument instead of writing
3852 to the standard output and assuming this can be redirected.
3853
3854 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3855 class containing characters with values greater than 255, PCRE compilation
3856 went into a loop.
3857
3858 12. A recursive reference to a subpattern that was within another subpattern
3859 that had a minimum quantifier of zero caused PCRE to crash. For example,
3860 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3861 recursion. If the recursively-called subpattern itself had a zero repeat,
3862 that was OK.
3863
3864 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3865 buffer into which it was copied (for escape processing) was still set at
3866 1024, so long lines caused crashes.
3867
3868 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3869 "internal error: code overflow...". This applied to any character class
3870 that was followed by a possessive quantifier.
3871
3872 15. Modified the Makefile to add libpcre.la as a prerequisite for
3873 libpcreposix.la because I was told this is needed for a parallel build to
3874 work.
3875
3876 16. If a pattern that contained .* following optional items at the start was
3877 studied, the wrong optimizing data was generated, leading to matching
3878 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3879 matching string must start with a or b or c. The correct conclusion for
3880 this pattern is that a match can start with any character.
3881
3882
3883 Version 4.4 13-Aug-03
3884 ---------------------
3885
3886 1. In UTF-8 mode, a character class containing characters with values between
3887 127 and 255 was not handled correctly if the compiled pattern was studied.
3888 In fixing this, I have also improved the studying algorithm for such
3889 classes (slightly).
3890
3891 2. Three internal functions had redundant arguments passed to them. Removal
3892 might give a very teeny performance improvement.
3893
3894 3. Documentation bug: the value of the capture_top field in a callout is *one
3895 more than* the number of the hightest numbered captured substring.
3896
3897 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3898 in incorrectly linking with a previously installed version. They now link
3899 explicitly with libpcre.la.
3900
3901 5. configure.in no longer needs to recognize Cygwin specially.
3902
3903 6. A problem in pcre.in for Windows platforms is fixed.
3904
3905 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3906 pcretest, it used to include the size of the study block as part of its
3907 output. Unfortunately, the structure contains a field that has a different
3908 size on different hardware architectures. This meant that the tests that
3909 showed this size failed. As the block is currently always of a fixed size,
3910 this information isn't actually particularly useful in pcretest output, so
3911 I have just removed it.
3912
3913 8. Three pre-processor statements accidentally did not start in column 1.
3914 Sadly, there are *still* compilers around that complain, even though
3915 standard C has not required this for well over a decade. Sigh.
3916
3917 9. In pcretest, the code for checking callouts passed small integers in the
3918 callout_data field, which is a void * field. However, some picky compilers
3919 complained about the casts involved for this on 64-bit systems. Now
3920 pcretest passes the address of the small integer instead, which should get
3921 rid of the warnings.
3922
3923 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3924 both compile and run time, and gives an error if an invalid UTF-8 sequence
3925 is found. There is a option for disabling this check in cases where the
3926 string is known to be correct and/or the maximum performance is wanted.
3927
3928 11. In response to a bug report, I changed one line in Makefile.in from
3929
3930 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3931 to
3932 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3933
3934 to look similar to other lines, but I have no way of telling whether this
3935 is the right thing to do, as I do not use Windows. No doubt I'll get told
3936 if it's wrong...
3937
3938
3939 Version 4.3 21-May-03
3940 ---------------------
3941
3942 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3943 Makefile.
3944
3945 2. Some refactoring to improve the quality of the code:
3946
3947 (i) The utf8_table... variables are now declared "const".
3948
3949 (ii) The code for \cx, which used the "case flipping" table to upper case
3950 lower case letters, now just substracts 32. This is ASCII-specific,
3951 but the whole concept of \cx is ASCII-specific, so it seems
3952 reasonable.
3953
3954 (iii) PCRE was using its character types table to recognize decimal and
3955 hexadecimal digits in the pattern. This is silly, because it handles
3956 only 0-9, a-f, and A-F, but the character types table is locale-
3957 specific, which means strange things might happen. A private
3958 table is now used for this - though it costs 256 bytes, a table is
3959 much faster than multiple explicit tests. Of course, the standard
3960 character types table is still used for matching digits in subject
3961 strings against \d.
3962
3963 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3964 ending in _t are). So I've renamed it as ESC_tee.
3965
3966 3. The first argument for regexec() in the POSIX wrapper should have been
3967 defined as "const".
3968
3969 4. Changed pcretest to use malloc() for its buffers so that they can be
3970 Electric Fenced for debugging.
3971
3972 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3973 to read one or more bytes before the start of the subject string. Often this
3974 had no effect on PCRE's behaviour, but in some circumstances it could
3975 provoke a segmentation fault.
3976
3977 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3978 to try to read one or more bytes before the start of the subject string.
3979
3980 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3981 UTF-8 support could misbehave in various ways if the subject string
3982 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3983 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3984 back over UTF-8 characters.)
3985
3986
3987 Version 4.2 14-Apr-03
3988 ---------------------
3989
3990 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3991
3992 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3993 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3994 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3995 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3996 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3997 and BUILD_EXEEXT
3998 Note: automatic setting of the BUILD variables is not yet working
3999 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4000 compile-time but not at link-time
4001 [LINK]: use for linking executables only
4002 make different versions for Windows and non-Windows
4003 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4004 libraries
4005 [LINK_FOR_BUILD]: new variable
4006 [OBJEXT]: use throughout
4007 [EXEEXT]: use throughout
4008 <winshared>: new target
4009 <wininstall>: new target
4010 <dftables.o>: use native compiler
4011 <dftables>: use native linker
4012 <install>: handle Windows platform correctly
4013 <clean>: ditto
4014 <check>: ditto
4015 copy DLL to top builddir before testing
4016
4017 As part of these changes, -no-undefined was removed again. This was reported
4018 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4019 in any case.
4020
4021 3. Some tidies to get rid of compiler warnings:
4022
4023 . In the match_data structure, match_limit was an unsigned long int, whereas
4024 match_call_count was an int. I've made them both unsigned long ints.
4025
4026 . In pcretest the fact that a const uschar * doesn't automatically cast to
4027 a void * provoked a warning.
4028
4029 . Turning on some more compiler warnings threw up some "shadow" variables
4030 and a few more missing casts.
4031
4032 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4033 option, a class that contained a single character with a value between 128
4034 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4035
4036 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4037 option, a class that contained several characters, but with at least one
4038 whose value was between 128 and 255 caused PCRE to crash.
4039
4040
4041 Version 4.1 12-Mar-03
4042 ---------------------
4043
4044 1. Compiling with gcc -pedantic found a couple of places where casts were
4045 needed, and a string in dftables.c that was longer than standard compilers are
4046 required to support.
4047
4048 2. Compiling with Sun's compiler found a few more places where the code could
4049 be tidied up in order to avoid warnings.
4050
4051 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4052 first of these names is deprecated in the latest Autoconf in favour of the name
4053 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4054 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4055 analogy I have changed it to CFLAGS_FOR_BUILD.
4056
4057 4. Added -no-undefined to the linking command in the Makefile, because this is
4058 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4059 linking step for the pcreposix library.
4060
4061 5. PCRE was failing to diagnose the case of two named groups with the same
4062 name.
4063
4064 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4065 literal character that is needed in the subject for a match, and scans along to
4066 ensure that it is present before embarking on the full matching process. This
4067 saves time in cases of nested unlimited repeats that are never going to match.
4068 Problem: the scan can take a lot of time if the subject is very long (e.g.
4069 megabytes), thus penalizing straightforward matches. It is now done only if the
4070 amount of subject to be scanned is less than 1000 bytes.
4071
4072 7. A lesser problem with the same optimization is that it was recording the
4073 first character of an anchored pattern as "needed", thus provoking a search
4074 right along the subject, even when the first match of the pattern was going to
4075 fail. The "needed" character is now not set for anchored patterns, unless it
4076 follows something in the pattern that is of non-fixed length. Thus, it still
4077 fulfils its original purpose of finding quick non-matches in cases of nested
4078 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4079
4080
4081 Version 4.0 17-Feb-03
4082 ---------------------
4083
4084 1. If a comment in an extended regex that started immediately after a meta-item
4085 extended to the end of string, PCRE compiled incorrect data. This could lead to
4086 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4087
4088 2. Moved to autoconf 2.53 and libtool 1.4.2.
4089
4090 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4091 the special perltest8 script is no longer needed - all the tests can be run
4092 from a single perltest script.
4093
4094 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4095 by \s. It has now been removed in PCRE. This means it isn't recognized as
4096 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4097 class [:space:] *does* include VT, thereby creating a mess.
4098
4099 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4100 space and tab.
4101
4102 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4103 its new features into the main test script, reducing the number of scripts.
4104
4105 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4106 were backward compatible, and made the (?i) apply to the whole pattern, as if
4107 /i were given. Now it behaves more logically, and applies the option setting
4108 only to what follows. PCRE has been changed to follow suit. However, if it
4109 finds options settings right at the start of the pattern, it extracts them into
4110 the global options, as before. Thus, they show up in the info data.
4111
4112 8. Added support for the \Q...\E escape sequence. Characters in between are
4113 treated as literals. This is slightly different from Perl in that $ and @ are
4114 also handled as literals inside the quotes. In Perl, they will cause variable
4115 interpolation. Note the following examples:
4116
4117 Pattern PCRE matches Perl matches
4118
4119 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4120 \Qabc\$xyz\E abc\$xyz abc\$xyz
4121 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4122
4123 For compatibility with Perl, \Q...\E sequences are recognized inside character
4124 classes as well as outside them.
4125
4126 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4127 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4128 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4129 signed/unsigned warnings.
4130
4131 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4132 option for pcretest, so I've replaced it by a simple function that does just
4133 that job.
4134
4135 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4136 "pcregrep -".
4137
4138 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4139 Java package. This provides some syntactic sugar for simple cases of what my
4140 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4141 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4142 item, you can use this simplified notation. Note that only makes sense with
4143 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4144 greediness, whatever the setting of the PCRE_UNGREEDY option.
4145
4146 13. A change of greediness default within a pattern was not taking effect at
4147 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4148 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4149 was abstracted outside.
4150
4151 14. PCRE now supports the \G assertion. It is true when the current matching
4152 position is at the start point of the match. This differs from \A when the
4153 starting offset is non-zero. Used with the /g option of pcretest (or similar
4154 code), it works in the same way as it does for Perl's /g option. If all
4155 alternatives of a regex begin with \G, the expression is anchored to the start
4156 match position, and the "anchored" flag is set in the compiled expression.
4157
4158 15. Some bugs concerning the handling of certain option changes within patterns
4159 have been fixed. These applied to options other than (?ims). For example,
4160 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4161 been the other way round. Some of this was related to change 7 above.
4162
4163 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4164 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4165 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4166 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4167
4168 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4169 mode. Unlike ".", it always matches newline, whatever the setting of
4170 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4171 assertions. Perl allows it, but it doesn't (in general) work because it can't
4172 calculate the length of the lookbehind. At least, that's the case for Perl
4173 5.8.0 - I've been told they are going to document that it doesn't work in
4174 future.
4175
4176 18. Added an error diagnosis for escapes that PCRE does not support: these are
4177 \L, \l, \N, \P, \p, \U, \u, and \X.
4178
4179 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4180 reading past the end of the pattern in cases such as /[abcd/.
4181
4182 20. PCRE was getting more memory than necessary for patterns with classes that
4183 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4184
4185 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4186 compiling PCRE for use with Virtual Pascal.
4187
4188 22. Small fix to the Makefile to make it work properly if the build is done
4189 outside the source tree.
4190
4191 23. Added a new extension: a condition to go with recursion. If a conditional
4192 subpattern starts with (?(R) the "true" branch is used if recursion has
4193 happened, whereas the "false" branch is used only at the top level.
4194
4195 24. When there was a very long string of literal characters (over 255 bytes
4196 without UTF support, over 250 bytes with UTF support), the computation of how
4197 much memory was required could be incorrect, leading to segfaults or other
4198 strange effects.
4199
4200 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4201 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4202 there was a subsequent back reference to those brackets. This meant that, for
4203 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4204 possible to check for precisely this case. All we can do is abandon the
4205 optimization if .* occurs inside capturing brackets when there are any back
4206 references whatsoever. (See below for a better fix that came later.)
4207
4208 26. The handling of the optimization for finding the first character of a
4209 non-anchored pattern, and for finding a character that is required later in the
4210 match were failing in some cases. This didn't break the matching; it just
4211 failed to optimize when it could. The way this is done has been re-implemented.
4212
4213 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4214
4215 28. Added a new feature that provides some of the functionality that Perl
4216 provides with (?{...}). The facility is termed a "callout". The way it is done
4217 in PCRE is for the caller to provide an optional function, by setting
4218 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4219 global variable. By default it is unset, which disables all calling out. To get
4220 the function called, the regex must include (?C) at appropriate points. This
4221 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4222 This provides a means of identifying different callout points. When PCRE
4223 reaches such a point in the regex, if pcre_callout has been set, the external
4224 function is called. It is provided with data in a structure called
4225 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4226 matching continues; if it returns a non-zero value, the match at the current
4227 point fails. However, backtracking will occur if possible. [This was changed
4228 later and other features added - see item 49 below.]
4229
4230 29. pcretest is upgraded to test the callout functionality. It provides a
4231 callout function that displays information. By default, it shows the start of
4232 the match and the current position in the text. There are some new data escapes
4233 to vary what happens:
4234
4235 \C+ in addition, show current contents of captured substrings
4236 \C- do not supply a callout function
4237 \C!n return 1 when callout number n is reached
4238 \C!n!m return 1 when callout number n is reached for the mth time
4239
4240 30. If pcregrep was called with the -l option and just a single file name, it
4241 output "<stdin>" if a match was found, instead of the file name.
4242
4243 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4244 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4245 pcre_exec(). This saves a malloc/free per call. The default value of
4246 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4247 when configuring.
4248
4249 32. The default maximum size of a compiled pattern is 64K. There have been a
4250 few cases of people hitting this limit. The code now uses macros to handle the
4251 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4252 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4253 configuring. Tests 2 and 5 work only with 2-byte links because they output
4254 debugging information about compiled patterns.
4255
4256 33. Internal code re-arrangements:
4257
4258 (a) Moved the debugging function for printing out a compiled regex into
4259 its own source file (printint.c) and used #include to pull it into
4260 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4261 separate copies.
4262
4263 (b) Defined the list of op-code names for debugging as a macro in
4264 internal.h so that it is next to the definition of the opcodes.
4265
4266 (c) Defined a table of op-code lengths for simpler skipping along compiled
4267 code. This is again a macro in internal.h so that it is next to the
4268 definition of the opcodes.
4269
4270 34. Added support for recursive calls to individual subpatterns, along the
4271 lines of Robin Houston's patch (but implemented somewhat differently).
4272
4273 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4274 allow it to read and process whole directories in Win32. This code was
4275 contributed by Lionel Fourquaux; it has not been tested by me.
4276
4277 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4278 used to name a group. Names consist of alphanumerics and underscores, and must
4279 be unique. Back references use the syntax (?P=name) and recursive calls use
4280 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4281 numbers. The function pcre_fullinfo() can be used after compilation to extract
4282 a name/number map. There are three relevant calls:
4283
4284 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4285 PCRE_INFO_NAMECOUNT yields the number of entries
4286 PCRE_INFO_NAMETABLE yields a pointer to the map.
4287
4288 The map is a vector of fixed-size entries. The size of each entry depends on
4289 the length of the longest name used. The first two bytes of each entry are the
4290 group number, most significant byte first. There follows the corresponding
4291 name, zero terminated. The names are in alphabetical order.
4292
4293 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4294 case instead of 255. Making it the same both with and without UTF-8 support
4295 means that the same test output works with both.
4296
4297 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4298 calling malloc() with a zero argument.
4299
4300 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4301 optimization. I've improved things by keeping a bitmap of backreferences with
4302 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4303 fact referenced, the optimization can be applied. It is unlikely that a
4304 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4305 the match to follow \n) will appear inside brackets with a number greater than
4306 31, but if it does, any back reference > 31 suppresses the optimization.
4307
4308 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4309 of disabling numbered capturing parentheses. Any opening parenthesis that is
4310 not followed by ? behaves as if it were followed by ?: but named parentheses
4311 can still be used for capturing (and they will acquire numbers in the usual
4312 way).
4313
4314 41. Redesigned the return codes from the match() function into yes/no/error so
4315 that errors can be passed back from deep inside the nested calls. A malloc
4316 failure while inside a recursive subpattern call now causes the
4317 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4318
4319 42. It is now possible to set a limit on the number of times the match()
4320 function is called in a call to pcre_exec(). This facility makes it possible to
4321 limit the amount of recursion and backtracking, though not in a directly
4322 obvious way, because the match() function is used in a number of different
4323 circumstances. The count starts from zero for each position in the subject
4324 string (for non-anchored patterns). The default limit is, for compatibility, a
4325 large number, namely 10 000 000. You can change this in two ways:
4326
4327 (a) When configuring PCRE before making, you can use --with-match-limit=n
4328 to set a default value for the compiled library.
4329
4330 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4331 a different value is set. See 45 below.
4332
4333 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4334
4335 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4336 of things that can be changed at compile time. The first argument specifies
4337 what is wanted and the second points to where the information is to be placed.
4338 The current list of available information is:
4339
4340 PCRE_CONFIG_UTF8
4341
4342 The output is an integer that is set to one if UTF-8 support is available;
4343 otherwise it is set to zero.
4344
4345 PCRE_CONFIG_NEWLINE
4346
4347 The output is an integer that it set to the value of the code that is used for
4348 newline. It is either LF (10) or CR (13).
4349
4350 PCRE_CONFIG_LINK_SIZE
4351
4352 The output is an integer that contains the number of bytes used for internal
4353 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4354
4355 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4356
4357 The output is an integer that contains the threshold above which the POSIX
4358 interface uses malloc() for output vectors. See item 31 above.
4359
4360 PCRE_CONFIG_MATCH_LIMIT
4361
4362 The output is an unsigned integer that contains the default limit of the number
4363 of match() calls in a pcre_exec() execution. See 42 above.
4364
4365 44. pcretest has been upgraded by the addition of the -C option. This causes it
4366 to extract all the available output from the new pcre_config() function, and to
4367 output it. The program then exits immediately.
4368
4369 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4370 order to support additional features. One way would have been to define
4371 pcre_exec2() (for example) with extra arguments, but this would not have been
4372 extensible, and would also have required all calls to the original function to
4373 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4374 is used for passing in "extra" data from pcre_study().
4375
4376 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4377 contains the following fields:
4378
4379 flags a bitmap indicating which of the following fields are set
4380 study_data opaque data from pcre_study()
4381 match_limit a way of specifying a limit on match() calls for a specific
4382 call to pcre_exec()
4383 callout_data data for callouts (see 49 below)
4384
4385 The flag bits are also defined in pcre.h, and are
4386
4387 PCRE_EXTRA_STUDY_DATA
4388 PCRE_EXTRA_MATCH_LIMIT
4389 PCRE_EXTRA_CALLOUT_DATA
4390
4391 The pcre_study() function now returns one of these new pcre_extra blocks, with
4392 the actual study data pointed to by the study_data field, and the
4393 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4394 before. That is, this change is entirely upwards-compatible and requires no
4395 change to existing code.
4396
4397 If you want to pass in additional data to pcre_exec(), you can either place it
4398 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4399 block.
4400
4401 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4402 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4403 times with different match limits, until it finds the minimum value needed for
4404 pcre_exec() to complete. The value is then output. This can be instructive; for
4405 most simple matches the number is quite small, but for pathological cases it
4406 gets very large very quickly.
4407
4408 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4409 returns the size of the data block pointed to by the study_data field in a
4410 pcre_extra block, that is, the value that was passed as the argument to
4411 pcre_malloc() when PCRE was getting memory in which to place the information
4412 created by pcre_study(). The fourth argument should point to a size_t variable.
4413 pcretest has been extended so that this information is shown after a successful
4414 pcre_study() call when information about the compiled regex is being displayed.
4415
4416 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4417 because what follows is always an absolute path. (Later: it turns out that this
4418 is more than cosmetic for MinGW, because it doesn't like empty path
4419 components.)
4420
4421 49. Some changes have been made to the callout feature (see 28 above):
4422
4423 (i) A callout function now has three choices for what it returns:
4424
4425 0 => success, carry on matching
4426 > 0 => failure at this point, but backtrack if possible
4427 < 0 => serious error, return this value from pcre_exec()
4428
4429 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4430 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4431 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4432 use by callout functions. It will never be used by PCRE itself.
4433
4434 (ii) The pcre_extra structure (see 45 above) has a void * field called
4435 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4436 pcre_callout_block structure has a field of the same name. The contents of
4437 the field passed in the pcre_extra structure are passed to the callout
4438 function in the corresponding field in the callout block. This makes it
4439 easier to use the same callout-containing regex from multiple threads. For
4440 testing, the pcretest program has a new data escape
4441
4442 \C*n pass the number n (may be negative) as callout_data
4443
4444 If the callout function in pcretest receives a non-zero value as
4445 callout_data, it returns that value.
4446
4447 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4448 there were some redundant $(CFLAGS) in commands that are now specified as
4449 $(LINK), which already includes $(CFLAGS).
4450
4451 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4452 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4453 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4454 one-byte characters throughout. Note that case-insensitive matching applies
4455 only to characters whose values are less than 256. PCRE doesn't support the
4456 notion of cases for higher-valued characters.
4457
4458 (i) A character class whose characters are all within 0-255 is handled as
4459 a bit map, and the map is inverted for negative classes. Previously, a
4460 character > 255 always failed to match such a class; however it should
4461 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4462
4463 (ii) A negated character class with a single character < 255 is coded as
4464 "not this character" (OP_NOT). This wasn't working properly when the test
4465 character was multibyte, either singly or repeated.
4466
4467 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4468 mode, for example: \x{100}{2,3}.
4469
4470 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4471 singly or repeated) now correctly test multibyte characters. However,
4472 PCRE doesn't recognize any characters with values greater than 255 as
4473 digits, spaces, or word characters. Such characters always match \D, \S,
4474 and \W, and never match \d, \s, or \w.
4475
4476 (v) Classes may now contain characters and character ranges with values
4477 greater than 255. For example: [ab\x{100}-\x{400}].
4478
4479 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4480 PCRE in UTF-8 mode.
4481
4482 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4483 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4484 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4485 value.)
4486
4487 53. The single man page has become too large. I have therefore split it up into
4488 a number of separate man pages. These also give rise to individual HTML pages;
4489 these are now put in a separate directory, and there is an index.html page that
4490 lists them all. Some hyperlinking between the pages has been installed.
4491
4492 54. Added convenience functions for handling named capturing parentheses.
4493
4494 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4495 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4496 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4497 are faulted.
4498
4499 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4500 calling configure. These values are used when compiling the dftables.c program
4501 which is run to generate the source of the default character tables. They
4502 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4503 you will need to set these values.
4504
4505 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4506
4507
4508 Version 3.9 02-Jan-02
4509 ---------------------
4510
4511 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4512
4513 2. If --disable-static was given, the building process failed when trying to
4514 build pcretest and pcregrep. (For some reason it was using libtool to compile
4515 them, which is not right, as they aren't part of the library.)
4516
4517
4518 Version 3.8 18-Dec-01
4519 ---------------------
4520
4521 1. The experimental UTF-8 code was completely screwed up. It was packing the
4522 bytes in the wrong order. How dumb can you get?
4523
4524
4525 Version 3.7 29-Oct-01
4526 ---------------------
4527
4528 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4529 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4530 this didn't happen under Solaris 8, where I normally test things.
4531
4532 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4533 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4534 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4535 this fix an hour or so after the initial 3.7 release.)
4536
4537
4538 Version 3.6 23-Oct-01
4539 ---------------------
4540
4541 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4542 offsets passed as NULL with zero offset count.
4543
4544 2. The config.guess and config.sub files had not been updated when I moved to
4545 the latest autoconf.
4546
4547
4548 Version 3.5 15-Aug-01
4549 ---------------------
4550
4551 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4552 had been forgotten.
4553
4554 2. By using declared but undefined structures, we can avoid using "void"
4555 definitions in pcre.h while keeping the internal definitions of the structures
4556 private.
4557
4558 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4559 user point of view, this means that both static and shared libraries are built
4560 by default, but this can be individually controlled. More of the work of
4561 handling this static/shared cases is now inside libtool instead of PCRE's make
4562 file.
4563
4564 4. The pcretest utility is now installed along with pcregrep because it is
4565 useful for users (to test regexs) and by doing this, it automatically gets
4566 relinked by libtool. The documentation has been turned into a man page, so
4567 there are now .1, .txt, and .html versions in /doc.
4568
4569 5. Upgrades to pcregrep:
4570 (i) Added long-form option names like gnu grep.
4571 (ii) Added --help to list all options with an explanatory phrase.
4572 (iii) Added -r, --recursive to recurse into sub-directories.
4573 (iv) Added -f, --file to read patterns from a file.
4574
4575 6. pcre_exec() was referring to its "code" argument before testing that
4576 argument for NULL (and giving an error if it was NULL).
4577
4578 7. Upgraded Makefile.in to allow for compiling in a different directory from
4579 the source directory.
4580
4581 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4582 options bits, the pointer it was passed was to an int instead of to an unsigned
4583 long int. This mattered only on 64-bit systems.
4584
4585 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4586 generated) instead of pcre.in, which it its source. Also made the same change
4587 in several of the .c files.
4588
4589 10. A new release of gcc defines printf() as a macro, which broke pcretest
4590 because it had an ifdef in the middle of a string argument for printf(). Fixed
4591 by using separate calls to printf().
4592
4593 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4594 script, to force use of CR or LF instead of \n in the source. On non-Unix
4595 systems, the value can be set in config.h.
4596
4597 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4598 absolute limit. Changed the text of the error message to make this clear, and
4599 likewise updated the man page.
4600
4601 13. The limit of 99 on the number of capturing subpatterns has been removed.
4602 The new limit is 65535, which I hope will not be a "real" limit.
4603
4604
4605 Version 3.4 22-Aug-00
4606 ---------------------
4607
4608 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4609
4610 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4611
4612
4613 Version 3.3 01-Aug-00
4614 ---------------------
4615
4616 1. If an octal character was given, but the value was greater than \377, it
4617 was not getting masked to the least significant bits, as documented. This could
4618 lead to crashes in some systems.
4619
4620 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4621 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4622
4623 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4624 These just pass their arguments on to (pcre_free)(), but they are provided
4625 because some uses of PCRE bind it to non-C systems that can call its functions,
4626 but cannot call free() or pcre_free() directly.
4627
4628 4. Add "make test" as a synonym for "make check". Corrected some comments in
4629 the Makefile.
4630
4631 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4632 Makefile.
4633
4634 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4635 command called pgrep for grepping around the active processes.
4636
4637 7. Added the beginnings of support for UTF-8 character strings.
4638
4639 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4640 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4641 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4642 out for the ar command.)
4643
4644
4645 Version 3.2 12-May-00
4646 ---------------------
4647
4648 This is purely a bug fixing release.
4649
4650 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4651 of ZA. This was just one example of several cases that could provoke this bug,
4652 which was introduced by change 9 of version 2.00. The code for breaking
4653 infinite loops after an iteration that matches an empty string was't working
4654 correctly.
4655
4656 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4657 when matched against abbab (for example). After matching an empty string, it
4658 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4659 caused it to match further down the string than it should.
4660
4661 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4662 was there because it doesn't seem to be needed, and it causes trouble on some
4663 systems, as it is not a Standard C header. It has been removed.
4664
4665 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4666 were reported on the Macintosh. The changes were from
4667
4668 while ((c = *(++ptr)) != 0 && c != '\n');
4669 to
4670 while ((c = *(++ptr)) != 0 && c != '\n') ;
4671
4672 Totally extraordinary, but if that's what it takes...
4673
4674 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4675 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4676 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4677 assumes the way PCRE uses memmove() (always moving upwards).
4678
4679 6. PCRE is being used in one environment where strchr() is not available. There
4680 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4681 faster code anyway.
4682
4683
4684 Version 3.1 09-Feb-00
4685 ---------------------
4686
4687 The only change in this release is the fixing of some bugs in Makefile.in for
4688 the "install" target:
4689
4690 (1) It was failing to install pcreposix.h.
4691
4692 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4693
4694
4695 Version 3.0 01-Feb-00
4696 ---------------------
4697
4698 1. Add support for the /+ modifier to perltest (to output $` like it does in
4699 pcretest).
4700
4701 2. Add support for the /g modifier to perltest.
4702
4703 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4704 matches null strings.
4705
4706 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4707 pattern. Perl treats empty patterns specially - it reuses the most recent
4708 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4709 effect.
4710
4711 5. The POSIX interface was broken in that it was just handing over the POSIX
4712 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4713 required a bigger vector, with some working space on the end. This means that
4714 the POSIX wrapper now has to get and free some memory, and copy the results.
4715
4716 6. Added some simple autoconf support, placing the test data and the
4717 documentation in separate directories, re-organizing some of the
4718 information files, and making it build pcre-config (a GNU standard). Also added
4719 libtool support for building PCRE as a shared library, which is now the
4720 default.
4721
4722 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4723 09 are not valid octal constants. Single digits will be used for minor values
4724 less than 10.
4725
4726 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4727 existing programs that set these in the POSIX interface can use PCRE without
4728 modification.
4729
4730 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4731 return all that pcre_info() returns, plus additional data. The pcre_info()
4732 function is retained for compatibility, but is considered to be obsolete.
4733
4734 10. Added experimental recursion feature (?R) to handle one common case that
4735 Perl 5.6 will be able to do with (?p{...}).
4736
4737 11. Added support for POSIX character classes like [:alpha:], which Perl is
4738 adopting.
4739
4740
4741 Version 2.08 31-Aug-99
4742 ----------------------
4743
4744 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4745 trying to match at the startoffset position, but instead was moving forward to
4746 the next newline as if a previous match had failed.
4747
4748 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4749 and could get into a loop if a null string was matched other than at the start
4750 of the subject.
4751
4752 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4753 be distinguished at compile time, and for completeness also added PCRE_DATE.
4754
4755 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4756 in GnuWin32 environments.
4757
4758
4759 Version 2.07 29-Jul-99
4760 ----------------------
4761
4762 1. The documentation is now supplied in plain text form and HTML as well as in
4763 the form of man page sources.
4764
4765 2. C++ compilers don't like assigning (void *) values to other pointer types.
4766 In particular this affects malloc(). Although there is no problem in Standard
4767 C, I've put in casts to keep C++ compilers happy.
4768
4769 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4770 should be (const char *).
4771
4772 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4773 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4774 However, I haven't made this a standard facility. The documentation doesn't
4775 mention it, and the Makefile doesn't support it.
4776
4777 5. The Makefile now contains an "install" target, with editable destinations at
4778 the top of the file. The pcretest program is not installed.
4779
4780 6. pgrep -V now gives the PCRE version number and date.
4781
4782 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4783 causing the entire string to be ignored, instead of just the last character.
4784
4785 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4786 non-matching string, it can take a very, very long time, even for strings of
4787 quite modest length, because of the nested recursion. PCRE now does better in
4788 some of these cases. It does this by remembering the last required literal
4789 character in the pattern, and pre-searching the subject to ensure it is present
4790 before running the real match. In other words, it applies a heuristic to detect
4791 some types of certain failure quickly, and in the above example, if presented
4792 with a string that has no trailing " it gives "no match" very quickly.
4793
4794 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4795 other alternatives are tried instead.
4796
4797
4798 Version 2.06 09-Jun-99
4799 ----------------------
4800
4801 1. Change pcretest's output for amount of store used to show just the code
4802 space, because the remainder (the data block) varies in size between 32-bit and
4803 64-bit systems.
4804
4805 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4806 start matching at. This allows lookbehinds to work when searching for multiple
4807 occurrences in a string.
4808
4809 3. Added additional options to pcretest for testing multiple occurrences:
4810
4811 /+ outputs the rest of the string that follows a match
4812 /g loops for multiple occurrences, using the new startoffset argument
4813 /G loops for multiple occurrences by passing an incremented pointer
4814
4815 4. PCRE wasn't doing the "first character" optimization for patterns starting
4816 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4817 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4818 the letter 'x'. On long subject strings, this gives a significant speed-up.
4819
4820
4821 Version 2.05 21-Apr-99
4822 ----------------------
4823
4824 1. Changed the type of magic_number from int to long int so that it works
4825 properly on 16-bit systems.
4826
4827 2. Fixed a bug which caused patterns starting with .* not to work correctly
4828 when the subject string contained newline characters. PCRE was assuming
4829 anchoring for such patterns in all cases, which is not correct because .* will
4830 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4831 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4832 must be retried after every newline in the subject.
4833
4834
4835 Version 2.04 18-Feb-99
4836 ----------------------
4837
4838 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4839 computation of the store needed to hold the pattern was incorrect (too large).
4840 If such patterns were nested a few deep, this could multiply and become a real
4841 problem.
4842
4843 2. Added /M option to pcretest to show the memory requirement of a specific
4844 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4845
4846 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4847 compiled in such a way that the backtracking after subsequent failure was
4848 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4849 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4850
4851
4852 Version 2.03 02-Feb-99
4853 ----------------------
4854
4855 1. Fixed typo and small mistake in man page.
4856
4857 2. Added 4th condition (GPL supersedes if conflict) and created separate
4858 LICENCE file containing the conditions.
4859
4860 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4861 Perl, that is the internal \ allows the delimiter to be included in the
4862 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4863 the final delimiter, add \ to the end of the pattern (to test the error).
4864
4865 4. Added the convenience functions for extracting substrings after a successful
4866 match. Updated pcretest to make it able to test these functions.
4867
4868
4869 Version 2.02 14-Jan-99
4870 ----------------------
4871
4872 1. Initialized the working variables associated with each extraction so that
4873 their saving and restoring doesn't refer to uninitialized store.
4874
4875 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4876 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4877 fix the problem.
4878
4879 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4880 calls, and wasn't printing the correct value for compiling calls. Increased the
4881 default value of LOOPREPEAT, and the number of significant figures in the
4882 times.
4883
4884 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4885
4886 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4887 a building problem on Windows NT with a FAT file system.
4888
4889
4890 Version 2.01 21-Oct-98
4891 ----------------------
4892
4893 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4894 to character tables built by pcre_maketables() in the current locale. If NULL
4895 is passed, the default tables are used.
4896
4897
4898 Version 2.00 24-Sep-98
4899 ----------------------
4900
4901 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4902 it any more.
4903
4904 2. Allow quantification of (?>) groups, and make it work correctly.
4905
4906 3. The first character computation wasn't working for (?>) groups.
4907
4908 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4909 end of the subject) and add 5.005's \z, which really does match only at the
4910 very end of the subject.
4911
4912 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4913
4914 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4915 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4916 localized options. All options to pcre_study() were also removed.
4917
4918 7. Add other new features from 5.005:
4919
4920 $(?<= positive lookbehind
4921 $(?<! negative lookbehind
4922 (?imsx-imsx) added the unsetting capability
4923 such a setting is global if at outer level; local otherwise
4924 (?imsx-imsx:) non-capturing groups with option setting
4925 (?(cond)re|re) conditional pattern matching
4926
4927 A backreference to itself in a repeated group matches the previous
4928 captured string.
4929
4930 8. General tidying up of studying (both automatic and via "study")
4931 consequential on the addition of new assertions.
4932
4933 9. As in 5.005, unlimited repeated groups that could match an empty substring
4934 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4935 runtime if any iteration does actually match an empty substring.
4936
4937 10. Include the RunTest script in the distribution.
4938
4939 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4940 discrepancies, some of which were old and were also with respect to 5.004. They
4941 have now been fixed.
4942
4943
4944 Version 1.09 28-Apr-98
4945 ----------------------
4946
4947 1. A negated single character class followed by a quantifier with a minimum
4948 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4949 program crashes, or just wrong answers. This did not apply to negated classes
4950 containing more than one character, or to minima other than one.
4951
4952
4953 Version 1.08 27-Mar-98
4954 ----------------------
4955
4956 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4957
4958 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4959 latter must appear before anything that relies on it in the pattern.
4960
4961
4962 Version 1.07 16-Feb-98
4963 ----------------------
4964
4965 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4966 repeat of a potentially empty string).
4967
4968
4969 Version 1.06 23-Jan-98
4970 ----------------------
4971
4972 1. Added Markus Oberhumer's little patches for C++.
4973
4974 2. Literal strings longer than 255 characters were broken.
4975
4976
4977 Version 1.05 23-Dec-97
4978 ----------------------
4979
4980 1. Negated character classes containing more than one character were failing if
4981 PCRE_CASELESS was set at run time.
4982
4983
4984 Version 1.04 19-Dec-97
4985 ----------------------
4986
4987 1. Corrected the man page, where some "const" qualifiers had been omitted.
4988
4989 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4990 input syntax.
4991
4992 3. Fixed memory leak which occurred when a regex with back references was
4993 matched with an offsets vector that wasn't big enough. The temporary memory
4994 that is used in this case wasn't being freed if the match failed.
4995
4996 4. Tidied pcretest to ensure it frees memory that it gets.
4997
4998 5. Temporary memory was being obtained in the case where the passed offsets
4999 vector was exactly big enough.
5000
5001 6. Corrected definition of offsetof() from change 5 below.
5002
5003 7. I had screwed up change 6 below and broken the rules for the use of
5004 setjmp(). Now fixed.
5005
5006
5007 Version 1.03 18-Dec-97
5008 ----------------------
5009
5010 1. A erroneous regex with a missing opening parenthesis was correctly
5011 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5012 on some systems.
5013
5014 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5015 it was reported that one broken compiler failed on the former because "code" is
5016 also an independent variable.
5017
5018 3. The erroneous regex a[]b caused an array overrun reference.
5019
5020 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5021 fail on data ending with that character. (It was going on too far, and checking
5022 the next character, typically a binary zero.) This was specific to the
5023 optimized code for single-character negative classes.
5024
5025 5. Added a contributed patch from the TIN world which does the following:
5026
5027 + Add an undef for memmove, in case the the system defines a macro for it.
5028
5029 + Add a definition of offsetof(), in case there isn't one. (I don't know
5030 the reason behind this - offsetof() is part of the ANSI standard - but
5031 it does no harm).
5032
5033 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5034 most of the places where whitespace preceded '#'. I have given up and
5035 allowed the remaining 2 cases to be at the margin.
5036
5037 + Rename some variables in pcre to eliminate shadowing. This seems very
5038 pedantic, but does no harm, of course.
5039
5040 6. Moved the call to setjmp() into its own function, to get rid of warnings
5041 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5042
5043 7. Constructs such as \d{8,} were compiling into the equivalent of
5044 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5045 outcome, but in this particular case used more store than had been allocated,
5046 which caused the bug to be discovered because it threw up an internal error.
5047
5048 8. The debugging code in both pcre and pcretest for outputting the compiled
5049 form of a regex was going wrong in the case of back references followed by
5050 curly-bracketed repeats.
5051
5052
5053 Version 1.02 12-Dec-97
5054 ----------------------
5055
5056 1. Typos in pcre.3 and comments in the source fixed.
5057
5058 2. Applied a contributed patch to get rid of places where it used to remove
5059 'const' from variables, and fixed some signed/unsigned and uninitialized
5060 variable warnings.
5061
5062 3. Added the "runtest" target to Makefile.
5063
5064 4. Set default compiler flag to -O2 rather than just -O.
5065
5066
5067 Version 1.01 19-Nov-97
5068 ----------------------
5069
5070 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5071 like /([ab]*)*/, that is, for classes with more than one character in them.
5072
5073 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5074 as /((?>a*))*/ (a PCRE_EXTRA facility).
5075
5076
5077 Version 1.00 18-Nov-97
5078 ----------------------
5079
5080 1. Added compile-time macros to support systems such as SunOS4 which don't have
5081 memmove() or strerror() but have other things that can be used instead.
5082
5083 2. Arranged that "make clean" removes the executables.
5084
5085
5086 Version 0.99 27-Oct-97
5087 ----------------------
5088
5089 1. Fixed bug in code for optimizing classes with only one character. It was
5090 initializing a 32-byte map regardless, which could cause it to run off the end
5091 of the memory it had got.
5092
5093 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5094
5095
5096 Version 0.98 22-Oct-97
5097 ----------------------
5098
5099 1. Fixed bug in code for handling temporary memory usage when there are more
5100 back references than supplied space in the ovector. This could cause segfaults.
5101
5102
5103 Version 0.97 21-Oct-97
5104 ----------------------
5105
5106 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5107
5108 2. Optimized negated single characters not to use a bit map.
5109
5110 3. Brought error texts together as macro definitions; clarified some of them;
5111 fixed one that was wrong - it said "range out of order" when it meant "invalid
5112 escape sequence".
5113
5114 4. Changed some char * arguments to const char *.
5115
5116 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5117
5118 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5119 pcretest.
5120
5121
5122 Version 0.96 16-Oct-97
5123 ----------------------
5124
5125 1. Added a simple "pgrep" utility to the distribution.
5126
5127 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5128 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5129 where "ddd" means "one or more decimal digits".
5130
5131 3. Fixed serious bug. If a pattern had a back reference, but the call to
5132 pcre_exec() didn't supply a large enough ovector to record the related
5133 identifying subpattern, the match always failed. PCRE now remembers the number
5134 of the largest back reference, and gets some temporary memory in which to save
5135 the offsets during matching if necessary, in order to ensure that
5136 backreferences always work.
5137
5138 4. Increased the compatibility with Perl in a number of ways:
5139
5140 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5141 to request this handling. The option can be set at compile or exec time.
5142
5143 (b) $ matches before a terminating newline by default; an option
5144 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5145 mode). The option can be set at compile or exec time.
5146
5147 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5148 the same as Perl's. If the decimal number it represents is less than 10
5149 or there aren't that many previous left capturing parentheses, an octal
5150 escape is read. Inside a character class, it's always an octal escape,
5151 even if it is a single digit.
5152
5153 (d) An escaped but undefined alphabetic character is taken as a literal,
5154 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5155 escapes.
5156
5157 (e) {0} is now permitted. (The previous item is removed from the compiled
5158 pattern).
5159
5160 5. Changed all the names of code files so that the basic parts are no longer
5161 than 10 characters, and abolished the teeny "globals.c" file.
5162
5163 6. Changed the handling of character classes; they are now done with a 32-byte
5164 bit map always.
5165
5166 7. Added the -d and /D options to pcretest to make it possible to look at the
5167 internals of compilation without having to recompile pcre.
5168
5169
5170 Version 0.95 23-Sep-97
5171 ----------------------
5172
5173 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5174 \x20 at the start of a run of normal characters. These were being treated as
5175 real characters, instead of the source characters being re-checked.
5176
5177
5178 Version 0.94 18-Sep-97
5179 ----------------------
5180
5181 1. The functions are now thread-safe, with the caveat that the global variables
5182 containing pointers to malloc() and free() or alternative functions are the
5183 same for all threads.
5184
5185 2. Get pcre_study() to generate a bitmap of initial characters for non-
5186 anchored patterns when this is possible, and use it if passed to pcre_exec().
5187
5188
5189 Version 0.93 15-Sep-97
5190 ----------------------
5191
5192 1. /(b)|(:+)/ was computing an incorrect first character.
5193
5194 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5195 but not actually doing anything yet.
5196
5197 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5198 as Perl does (e.g. [-az] or [az-]).
5199
5200 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5201 all possible positions.
5202
5203 5. Split up into different modules to avoid including unneeded functions in a
5204 compiled binary. However, compile and exec are still in one module. The "study"
5205 function is split off.
5206
5207 6. The character tables are now in a separate module whose source is generated
5208 by an auxiliary program - but can then be edited by hand if required. There are
5209 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5210 toupper() in the code.
5211
5212 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5213 make them global. Abolish the function for setting them, as the caller can now
5214 set them directly.
5215
5216
5217 Version 0.92 11-Sep-97
5218 ----------------------
5219
5220 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5221 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5222
5223 2. Caseless matching was not working in character classes if the characters in
5224 the pattern were in upper case.
5225
5226 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5227
5228 4. Make PCRE_ANCHORED public and accept as a compile option.
5229
5230 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5231 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5232 pass them.
5233
5234 6. Give an error if bad option bits passed at compile or run time.
5235
5236 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5237 pcretest to cause it to pass that flag.
5238
5239 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5240 options, and the first character, if set.
5241
5242 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5243
5244
5245 Version 0.91 10-Sep-97
5246 ----------------------
5247
5248 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5249 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5250
5251 2. PCRE was looping on encountering an indefinitely repeated back reference to
5252 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5253 Perl does - treats the match as successful.
5254
5255 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5