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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1475 - (show annotations)
Fri Apr 25 11:59:19 2014 UTC (5 years, 6 months ago) by zherczeg
File size: 265004 byte(s)
Fixed an issue with nested table jumps.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.36 xx-xxx-2014
5 ------------------------
6
7 1. Got rid of some compiler warnings in the C++ modules that were shown up by
8 -Wmissing-field-initializers and -Wunused-parameter.
9
10 2. The tests for quantifiers being too big (greater than 65535) were being
11 applied after reading the number, and stupidly assuming that integer
12 overflow would give a negative number. The tests are now applied as the
13 numbers are read.
14
15 3. Tidy code in pcre_exec.c where two branches that used to be different are
16 now the same.
17
18 4. The JIT compiler did not generate match limit checks for certain
19 bracketed expressions with quantifiers. This may lead to exponential
20 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
21 issue should be resolved now.
22
23 5. Fixed an issue, which occures when nested alternatives are optimized
24 with table jumps.
25
26
27 Version 8.35 04-April-2014
28 --------------------------
29
30 1. A new flag is set, when property checks are present in an XCLASS.
31 When this flag is not set, PCRE can perform certain optimizations
32 such as studying these XCLASS-es.
33
34 2. The auto-possessification of character sets were improved: a normal
35 and an extended character set can be compared now. Furthermore
36 the JIT compiler optimizes more character set checks.
37
38 3. Got rid of some compiler warnings for potentially uninitialized variables
39 that show up only when compiled with -O2.
40
41 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
42 of a match later then the end of the match. The pcretest program was not
43 handling the case sensibly - it was outputting from the start to the next
44 binary zero. It now reports this situation in a message, and outputs the
45 text from the end to the start.
46
47 5. Fast forward search is improved in JIT. Instead of the first three
48 characters, any three characters with fixed position can be searched.
49 Search order: first, last, middle.
50
51 6. Improve character range checks in JIT. Characters are read by an inprecise
52 function now, which returns with an unknown value if the character code is
53 above a certain treshold (e.g: 256). The only limitation is that the value
54 must be bigger than the treshold as well. This function is useful, when
55 the characters above the treshold are handled in the same way.
56
57 7. The macros whose names start with RAWUCHAR are placeholders for a future
58 mode in which only the bottom 21 bits of 32-bit data items are used. To
59 make this more memorable for those maintaining the code, the names have
60 been changed to start with UCHAR21, and an extensive comment has been added
61 to their definition.
62
63 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
64 to the export list in Makefile.am (they were accidentally omitted from the
65 8.34 tarball).
66
67 9. The informational output from pcretest used the phrase "starting byte set"
68 which is inappropriate for the 16-bit and 32-bit libraries. As the output
69 for "first char" and "need char" really means "non-UTF-char", I've changed
70 "byte" to "char", and slightly reworded the output. The documentation about
71 these values has also been (I hope) clarified.
72
73 10. Another JIT related optimization: use table jumps for selecting the correct
74 backtracking path, when more than four alternatives are present inside a
75 bracket.
76
77 11. Empty match is not possible, when the minimum length is greater than zero,
78 and there is no \K in the pattern. JIT should avoid empty match checks in
79 such cases.
80
81 12. In a caseless character class with UCP support, when a character with more
82 than one alternative case was not the first character of a range, not all
83 the alternative cases were added to the class. For example, s and \x{17f}
84 are both alternative cases for S: the class [RST] was handled correctly,
85 but [R-T] was not.
86
87 13. The configure.ac file always checked for pthread support when JIT was
88 enabled. This is not used in Windows, so I have put this test inside a
89 check for the presence of windows.h (which was already tested for).
90
91 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
92 The algorithm provides a way to skip certain starting offsets, and usually
93 faster than linear prefix searches.
94
95 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
96 as for 'fr_FR' and 'french'. For some reason, however, it then used the
97 Windows-specific input and output files, which have 'french' screwed in.
98 So this could never have worked. One of the problems with locales is that
99 they aren't always the same. I have now updated RunTest so that it checks
100 the output of the locale test (test 3) against three different output
101 files, and it allows the test to pass if any one of them matches. With luck
102 this should make the test pass on some versions of Solaris where it was
103 failing. Because of the uncertainty, the script did not used to stop if
104 test 3 failed; it now does. If further versions of a French locale ever
105 come to light, they can now easily be added.
106
107 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
108 there was a message during ./configure, but it did not stop. This now
109 provokes an error. The invalid example in README has been corrected.
110 If a value less than the minimum is given, the minimum value has always
111 been used, but now a warning is given.
112
113 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
114 was a bug in the test system, which is now fixed. Also, the list of various
115 configurations that are tested for each release did not have one with both
116 16/32 bits and --enable-bar-anycrlf. It now does.
117
118 18. pcretest was missing "-C bsr" for displaying the \R default setting.
119
120 19. Little endian PowerPC systems are supported now by the JIT compiler.
121
122 20. The fast forward newline mechanism could enter to an infinite loop on
123 certain invalid UTF-8 input. Although we don't support these cases
124 this issue can be fixed by a performance optimization.
125
126 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
127 not take account if existing stack usage. There is now a new global
128 variable called pcre_stack_guard that can be set to point to an external
129 function to check stack availability. It is called at the start of
130 processing every parenthesized group.
131
132 22. A typo in the code meant that in ungreedy mode the max/min qualifier
133 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
134 not match "ab".
135
136 23. When UTF was disabled, the JIT program reported some incorrect compile
137 errors. These messages are silenced now.
138
139 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
140 compiler.
141
142 25. Change all the temporary files used in RunGrepTest to be different to those
143 used by RunTest so that the tests can be run simultaneously, for example by
144 "make -j check".
145
146
147 Version 8.34 15-December-2013
148 -----------------------------
149
150 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
151 executable memory. Patch inspired by Carsten Klein.
152
153 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
154 this macro is never tested and has no effect, because the work to support
155 coverage involves only compiling and linking options and special targets in
156 the Makefile. The comment in config.h implied that defining the macro would
157 enable coverage support, which is totally false. There was also support for
158 setting this macro in the CMake files (my fault, I just copied it from
159 configure). SUPPORT_GCOV has now been removed.
160
161 3. Make a small performance improvement in strlen16() and strlen32() in
162 pcretest.
163
164 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
165 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
166 these cases). There was also one in pcretest.c.
167
168 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
169
170 6. In UTF mode, the code for checking whether a group could match an empty
171 string (which is used for indefinitely repeated groups to allow for
172 breaking an infinite loop) was broken when the group contained a repeated
173 negated single-character class with a character that occupied more than one
174 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
175 UTF-8 mode). The effect was undefined: the group might or might not be
176 deemed as matching an empty string, or the program might have crashed.
177
178 7. The code for checking whether a group could match an empty string was not
179 recognizing that \h, \H, \v, \V, and \R must match a character.
180
181 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
182 an empty string. If it can, pcretest shows this in its information output.
183
184 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
185 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
186 matched by pcre_exec() without using JIT:
187
188 (a) If the rest of the pattern did not match after a maximal run of
189 grapheme clusters, the code for backing up to try with fewer of them
190 did not always back up over a full grapheme when characters that do not
191 have the modifier quality were involved, e.g. Hangul syllables.
192
193 (b) If the match point in a subject started with modifier character, and
194 there was no match, the code could incorrectly back up beyond the match
195 point, and potentially beyond the first character in the subject,
196 leading to a segfault or an incorrect match result.
197
198 10. A conditional group with an assertion condition could lead to PCRE
199 recording an incorrect first data item for a match if no other first data
200 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
201 first data item, and therefore matched "ca" after "c" instead of at the
202 start.
203
204 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
205 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
206
207 12. The source of pcregrep now includes z/OS-specific code so that it can be
208 compiled for z/OS as part of the special z/OS distribution.
209
210 13. Added the -T and -TM options to pcretest.
211
212 14. The code in pcre_compile.c for creating the table of named capturing groups
213 has been refactored. Instead of creating the table dynamically during the
214 actual compiling pass, the information is remembered during the pre-compile
215 pass (on the stack unless there are more than 20 named groups, in which
216 case malloc() is used) and the whole table is created before the actual
217 compile happens. This has simplified the code (it is now nearly 150 lines
218 shorter) and prepared the way for better handling of references to groups
219 with duplicate names.
220
221 15. A back reference to a named subpattern when there is more than one of the
222 same name now checks them in the order in which they appear in the pattern.
223 The first one that is set is used for the reference. Previously only the
224 first one was inspected. This change makes PCRE more compatible with Perl.
225
226 16. Unicode character properties were updated from Unicode 6.3.0.
227
228 17. The compile-time code for auto-possessification has been refactored, based
229 on a patch by Zoltan Herczeg. It now happens after instead of during
230 compilation. The code is cleaner, and more cases are handled. The option
231 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
232 options in pcretest are provided to set it. It can also be set by
233 (*NO_AUTO_POSSESS) at the start of a pattern.
234
235 18. The character VT has been added to the default ("C" locale) set of
236 characters that match \s and are generally treated as white space,
237 following this same change in Perl 5.18. There is now no difference between
238 "Perl space" and "POSIX space". Whether VT is treated as white space in
239 other locales depends on the locale.
240
241 19. The code for checking named groups as conditions, either for being set or
242 for being recursed, has been refactored (this is related to 14 and 15
243 above). Processing unduplicated named groups should now be as fast at
244 numerical groups, and processing duplicated groups should be faster than
245 before.
246
247 20. Two patches to the CMake build system, by Alexander Barkov:
248
249 (1) Replace the "source" command by "." in CMakeLists.txt because
250 "source" is a bash-ism.
251
252 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
253 without these the CMake build does not work on Solaris.
254
255 21. Perl has changed its handling of \8 and \9. If there is no previously
256 encountered capturing group of those numbers, they are treated as the
257 literal characters 8 and 9 instead of a binary zero followed by the
258 literals. PCRE now does the same.
259
260 22. Following Perl, added \o{} to specify codepoints in octal, making it
261 possible to specify values greater than 0777 and also making them
262 unambiguous.
263
264 23. Perl now gives an error for missing closing braces after \x{... instead of
265 treating the string as literal. PCRE now does the same.
266
267 24. RunTest used to grumble if an inappropriate test was selected explicitly,
268 but just skip it when running all tests. This make it awkward to run ranges
269 of tests when one of them was inappropriate. Now it just skips any
270 inappropriate tests, as it always did when running all tests.
271
272 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
273 character types such as \d or \w, too many callouts were inserted, and the
274 data that they returned was rubbish.
275
276 26. In UCP mode, \s was not matching two of the characters that Perl matches,
277 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
278 were matched by \h. The code has now been refactored so that the lists of
279 the horizontal and vertical whitespace characters used for \h and \v (which
280 are defined only in one place) are now also used for \s.
281
282 27. Add JIT support for the 64 bit TileGX architecture.
283 Patch by Jiong Wang (Tilera Corporation).
284
285 28. Possessive quantifiers for classes (both explicit and automatically
286 generated) now use special opcodes instead of wrapping in ONCE brackets.
287
288 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
289 (because it's meaningless), this was not happening when PCRE_CASELESS was
290 set. Not wrong, but inefficient.
291
292 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
293 properties for \w, \d, etc) is present in a test regex. Otherwise if the
294 test contains no characters greater than 255, Perl doesn't realise it
295 should be using Unicode semantics.
296
297 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
298 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
299 does in Unicode mode.
300
301 32. Added the "forbid" facility to pcretest so that putting tests into the
302 wrong test files can sometimes be quickly detected.
303
304 33. There is now a limit (default 250) on the depth of nesting of parentheses.
305 This limit is imposed to control the amount of system stack used at compile
306 time. It can be changed at build time by --with-parens-nest-limit=xxx or
307 the equivalent in CMake.
308
309 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
310 errors. Perl warns for these when in warning mode, but PCRE has no facility
311 for giving warnings.
312
313 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
314 However, this was not working for (?!) because it is optimized to (*FAIL),
315 for which PCRE does not allow quantifiers. The optimization is now disabled
316 when a quantifier follows (?!). I can't see any use for this, but it makes
317 things uniform.
318
319 36. Perl no longer allows group names to start with digits, so I have made this
320 change also in PCRE. It simplifies the code a bit.
321
322 37. In extended mode, Perl ignores spaces before a + that indicates a
323 possessive quantifier. PCRE allowed a space before the quantifier, but not
324 before the possessive +. It now does.
325
326 38. The use of \K (reset reported match start) within a repeated possessive
327 group such as (a\Kb)*+ was not working.
328
329 40. Document that the same character tables must be used at compile time and
330 run time, and that the facility to pass tables to pcre_exec() and
331 pcre_dfa_exec() is for use only with saved/restored patterns.
332
333 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
334 features for Builds with MSVC:
335
336 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
337 builds). The .rc files can be used to set FileDescription and many other
338 attributes.
339
340 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
341 This allows higher-level build scripts which want .pdb files to avoid
342 hard-coding the exact files needed."
343
344 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
345 mean "start of word" and "end of word", respectively, as a transition aid.
346
347 43. A minimizing repeat of a class containing codepoints greater than 255 in
348 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
349 compiled to use the heap for recursion.
350
351 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
352 is configured.
353
354
355 Version 8.33 28-May-2013
356 ------------------------
357
358 1. Added 'U' to some constants that are compared to unsigned integers, to
359 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
360 variables that are added to signed variables, to ensure the result is
361 signed and can be negated.
362
363 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
364 CMake config files.
365
366 3. Revise the creation of config.h.generic so that all boolean macros are
367 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
368 overriding via -D on the command line possible.
369
370 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
371 to unsigned int is reported to make a quite noticeable speed difference in
372 a specific Windows environment. Testing on Linux did also appear to show
373 some benefit (and it is clearly not harmful). Also fixed the definition of
374 Xop which should be unsigned.
375
376 5. Related to (4), changing the definition of the intermediate variable cc
377 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
378 improvements.
379
380 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
381 unnecessary spaces.
382
383 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
384 and later.
385
386 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
387
388 9. Optimizing fast_forward_start_bits in JIT.
389
390 10. Adding support for callouts in JIT, and fixing some issues revealed
391 during this work. Namely:
392
393 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
394
395 (b) Minimum length was not checked before the matching is started.
396
397 11. The value of capture_last that is passed to callouts was incorrect in some
398 cases when there was a capture on one path that was subsequently abandoned
399 after a backtrack. Also, the capture_last value is now reset after a
400 recursion, since all captures are also reset in this case.
401
402 12. The interpreter no longer returns the "too many substrings" error in the
403 case when an overflowing capture is in a branch that is subsequently
404 abandoned after a backtrack.
405
406 13. In the pathological case when an offset vector of size 2 is used, pcretest
407 now prints out the matched string after a yield of 0 or 1.
408
409 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
410 Only supported by the JIT compiler at the moment.
411
412 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
413
414 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
415 offset of the starting point of the matching process, provided the offsets
416 vector is large enough.
417
418 17. The \A escape now records a lookbehind value of 1, though its execution
419 does not actually inspect the previous character. This is to ensure that,
420 in partial multi-segment matching, at least one character from the old
421 segment is retained when a new segment is processed. Otherwise, if there
422 are no lookbehinds in the pattern, \A might match incorrectly at the start
423 of a new segment.
424
425 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
426
427 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
428 gives some modest performance improvement in 8-bit mode.
429
430 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
431 be expressed in certain programming languages using Universal Character
432 Names.
433
434 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
435 which points out that "non characters" are not "characters that may not
436 appear in Unicode strings" but rather "characters that are reserved for
437 internal use and have only local meaning".
438
439 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
440 there was a conditional group that depended on an assertion, if the
441 assertion was false, the callout that immediately followed the alternation
442 in the condition was skipped when pcre_exec() was used for matching.
443
444 23. Allow an explicit callout to be inserted before an assertion that is the
445 condition for a conditional group, for compatibility with automatic
446 callouts, which always insert a callout at this point.
447
448 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
449 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
450
451 25. (*PRUNE) is now supported by the JIT compiler.
452
453 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
454
455 27. Fix the case where there are two or more SKIPs with arguments that may be
456 ignored.
457
458 28. (*SKIP) is now supported by the JIT compiler.
459
460 29. (*THEN) is now supported by the JIT compiler.
461
462 30. Update RunTest with additional test selector options.
463
464 31. The way PCRE handles backtracking verbs has been changed in two ways.
465
466 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
467 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
468 backtracking. In some cases this makes it more Perl-compatible, but Perl's
469 rather obscure rules do not always do the same thing.
470
471 (2) Previously, backtracking verbs were confined within assertions. This is
472 no longer the case for positive assertions, except for (*ACCEPT). Again,
473 this sometimes improves Perl compatibility, and sometimes does not.
474
475 32. A number of tests that were in test 2 because Perl did things differently
476 have been moved to test 1, because either Perl or PCRE has changed, and
477 these tests are now compatible.
478
479 32. Backtracking control verbs are now handled in the same way in JIT and
480 interpreter.
481
482 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
483 contained a forward subroutine reference caused a compile error.
484
485 34. Auto-detect and optimize limited repetitions in JIT.
486
487 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
488 blocking (*UTF) etc.
489
490 36. In the interpreter, maximizing pattern repetitions for characters and
491 character types now use tail recursion, which reduces stack usage.
492
493 37. The value of the max lookbehind was not correctly preserved if a compiled
494 and saved regex was reloaded on a host of different endianness.
495
496 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
497 of the compiled pattern block, expand the flags field from 16 to 32 bits
498 because it was almost full.
499
500 39. Try madvise first before posix_madvise.
501
502 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
503 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
504 original change locked out all matches of empty strings. This has been
505 changed so that one match of an empty string per line is recognized.
506 Subsequent searches on the same line (for colouring or for --only-matching,
507 for example) do not recognize empty strings.
508
509 41. Applied a user patch to fix a number of spelling mistakes in comments.
510
511 42. Data lines longer than 65536 caused pcretest to crash.
512
513 43. Clarified the data type for length and startoffset arguments for pcre_exec
514 and pcre_dfa_exec in the function-specific man pages, where they were
515 explicitly stated to be in bytes, never having been updated. I also added
516 some clarification to the pcreapi man page.
517
518 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
519 a segmentation fault.
520
521
522 Version 8.32 30-November-2012
523 -----------------------------
524
525 1. Improved JIT compiler optimizations for first character search and single
526 character iterators.
527
528 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
529 Patch by Daniel Richard G.
530
531 3. Single character iterator optimizations in the JIT compiler.
532
533 4. Improved JIT compiler optimizations for character ranges.
534
535 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
536 Reported by Giuseppe D'Angelo.
537
538 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
539 of a line, was being set incorrectly in cases where .* appeared inside
540 atomic brackets at the start of a pattern, or where there was a subsequent
541 *PRUNE or *SKIP.
542
543 7. Improved instruction cache flush for POWER/PowerPC.
544 Patch by Daniel Richard G.
545
546 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
547 grep:
548
549 (a) There is now no limit to the number of patterns to be matched.
550
551 (b) An error is given if a pattern is too long.
552
553 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
554 are now supported.
555
556 (d) --exclude-from and --include-from (multiple use) have been added.
557
558 (e) Exclusions and inclusions now apply to all files and directories, not
559 just to those obtained from scanning a directory recursively.
560
561 (f) Multiple uses of -f and --file-list are now supported.
562
563 (g) In a Windows environment, the default for -d has been changed from
564 "read" (the GNU grep default) to "skip", because otherwise the presence
565 of a directory in the file list provokes an error.
566
567 (h) The documentation has been revised and clarified in places.
568
569 9. Improve the matching speed of capturing brackets.
570
571 10. Changed the meaning of \X so that it now matches a Unicode extended
572 grapheme cluster.
573
574 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
575 out POSIX threads when JIT support is configured.
576
577 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
578
579 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
580 structure could go wrong in environments where size_t is not the same size
581 as int.
582
583 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
584 set.
585
586 15. The EBCDIC support had decayed; later updates to the code had included
587 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
588 general tidy up of EBCDIC-related issues, and the documentation was also
589 not quite right. There is now a test that can be run on ASCII systems to
590 check some of the EBCDIC-related things (but is it not a full test).
591
592 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
593 in a small tidy to the code.
594
595 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
596
597 18. If the --only-matching (-o) option in pcregrep is specified multiple
598 times, each one causes appropriate output. For example, -o1 -o2 outputs the
599 substrings matched by the 1st and 2nd capturing parentheses. A separating
600 string can be specified by --om-separator (default empty).
601
602 19. Improving the first n character searches.
603
604 20. Turn case lists for horizontal and vertical white space into macros so that
605 they are defined only once.
606
607 21. This set of changes together give more compatible Unicode case-folding
608 behaviour for characters that have more than one other case when UCP
609 support is available.
610
611 (a) The Unicode property table now has offsets into a new table of sets of
612 three or more characters that are case-equivalent. The MultiStage2.py
613 script that generates these tables (the pcre_ucd.c file) now scans
614 CaseFolding.txt instead of UnicodeData.txt for character case
615 information.
616
617 (b) The code for adding characters or ranges of characters to a character
618 class has been abstracted into a generalized function that also handles
619 case-independence. In UTF-mode with UCP support, this uses the new data
620 to handle characters with more than one other case.
621
622 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
623 whose other case is greater than 256 are now correctly matched
624 caselessly. Previously, the high codepoint matched the low one, but not
625 vice versa.
626
627 (d) The processing of \h, \H, \v, and \ in character classes now makes use
628 of the new class addition function, using character lists defined as
629 macros alongside the case definitions of 20 above.
630
631 (e) Caseless back references now work with characters that have more than
632 one other case.
633
634 (f) General caseless matching of characters with more than one other case
635 is supported.
636
637 22. Unicode character properties were updated from Unicode 6.2.0
638
639 23. Improved CMake support under Windows. Patch by Daniel Richard G.
640
641 24. Add support for 32-bit character strings, and UTF-32
642
643 25. Major JIT compiler update (code refactoring and bugfixing).
644 Experimental Sparc 32 support is added.
645
646 26. Applied a modified version of Daniel Richard G's patch to create
647 pcre.h.generic and config.h.generic by "make" instead of in the
648 PrepareRelease script.
649
650 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
651 pcre_compile.c when checking for a zero character.
652
653 28. Introducing a native interface for JIT. Through this interface, the compiled
654 machine code can be directly executed. The purpose of this interface is to
655 provide fast pattern matching, so several sanity checks are not performed.
656 However, feature tests are still performed. The new interface provides
657 1.4x speedup compared to the old one.
658
659 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
660 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
661 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
662
663 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
664 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
665 the "old" RFC 2279). Instead, it ended up passing a negative length to
666 pcre_exec().
667
668 31. Add support for GCC's visibility feature to hide internal functions.
669
670 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
671 "unknown -C option" after outputting 0 or 1.
672
673 33. There is now support for generating a code coverage report for the test
674 suite in environments where gcc is the compiler and lcov is installed. This
675 is mainly for the benefit of the developers.
676
677 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
678 unaddressable using valgrind annotations, allowing valgrind to detect
679 invalid memory accesses. This is mainly for the benefit of the developers.
680
681 25. (*UTF) can now be used to start a pattern in any of the three libraries.
682
683 26. Give configure error if --enable-cpp but no C++ compiler found.
684
685
686 Version 8.31 06-July-2012
687 -------------------------
688
689 1. Fixing a wrong JIT test case and some compiler warnings.
690
691 2. Removed a bashism from the RunTest script.
692
693 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
694 to unsigned type, result still unsigned" that was given by an MS compiler
695 on encountering the code "-sizeof(xxx)".
696
697 4. Partial matching support is added to the JIT compiler.
698
699 5. Fixed several bugs concerned with partial matching of items that consist
700 of more than one character:
701
702 (a) /^(..)\1/ did not partially match "aba" because checking references was
703 done on an "all or nothing" basis. This also applied to repeated
704 references.
705
706 (b) \R did not give a hard partial match if \r was found at the end of the
707 subject.
708
709 (c) \X did not give a hard partial match after matching one or more
710 characters at the end of the subject.
711
712 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
713 a partial match for the string "\r".
714
715 (e) When newline was set to CRLF, the metacharacter "." did not recognize
716 a partial match for a CR character at the end of the subject string.
717
718 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
719 running pcretest, the text "(JIT)" added to the output whenever JIT is
720 actually used to run the match.
721
722 7. Individual JIT compile options can be set in pcretest by following -s+[+]
723 or /S+[+] with a digit between 1 and 7.
724
725 8. OP_NOT now supports any UTF character not just single-byte ones.
726
727 9. (*MARK) control verb is now supported by the JIT compiler.
728
729 10. The command "./RunTest list" lists the available tests without actually
730 running any of them. (Because I keep forgetting what they all are.)
731
732 11. Add PCRE_INFO_MAXLOOKBEHIND.
733
734 12. Applied a (slightly modified) user-supplied patch that improves performance
735 when the heap is used for recursion (compiled with --disable-stack-for-
736 recursion). Instead of malloc and free for each heap frame each time a
737 logical recursion happens, frames are retained on a chain and re-used where
738 possible. This sometimes gives as much as 30% improvement.
739
740 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
741 call.
742
743 14. As documented, (*COMMIT) is now confined to within a positive assertion.
744
745 15. It is now possible to link pcretest with libedit as an alternative to
746 libreadline.
747
748 16. (*COMMIT) control verb is now supported by the JIT compiler.
749
750 17. The Unicode data tables have been updated to Unicode 6.1.0.
751
752 18. Added --file-list option to pcregrep.
753
754 19. Added binary file support to pcregrep, including the -a, --binary-files,
755 -I, and --text options.
756
757 20. The madvise function is renamed for posix_madvise for QNX compatibility
758 reasons. Fixed by Giuseppe D'Angelo.
759
760 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
761
762 22. Changed the option for creating symbolic links for 16-bit man pages from
763 -s to -sf so that re-installing does not cause issues.
764
765 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
766
767 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
768 matches in certain environments (the workspace was not being correctly
769 retained). Also added to pcre_dfa_exec() a simple plausibility check on
770 some of the workspace data at the beginning of a restart.
771
772 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
773 was not doing so when it should - probably a typo introduced by SVN 528
774 (change 8.10/14).
775
776 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
777 \w+ when the character tables indicated that \x{c4} was a word character.
778 There were several related cases, all because the tests for doing a table
779 lookup were testing for characters less than 127 instead of 255.
780
781 27. If a pattern contains capturing parentheses that are not used in a match,
782 their slots in the ovector are set to -1. For those that are higher than
783 any matched groups, this happens at the end of processing. In the case when
784 there were back references that the ovector was too small to contain
785 (causing temporary malloc'd memory to be used during matching), and the
786 highest capturing number was not used, memory off the end of the ovector
787 was incorrectly being set to -1. (It was using the size of the temporary
788 memory instead of the true size.)
789
790 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
791 ovector size, it uses memory at the end of the block that it has got.
792
793 29. Check for an overlong MARK name and give an error at compile time. The
794 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
795
796 30. JIT compiler update.
797
798 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
799 Rill for the patch.
800
801 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
802
803 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
804
805 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
806 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
807 was enabled.
808
809 35. Improve JIT code generation for greedy plus quantifier.
810
811 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
812 1 to "aa" instead of to an empty string. The bug affected repeated groups
813 that could potentially match an empty string.
814
815 37. Optimizing single character iterators in JIT.
816
817 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
818 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
819 codepoints that are too big for the mode are faulted, and in a UTF mode,
820 disallowed codepoints are also faulted.
821
822 39. If PCRE was compiled with UTF support, in three places in the DFA
823 matcher there was code that should only have been obeyed in UTF mode, but
824 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
825 processing when bytes with values greater than 127 were present. In 16-bit
826 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
827 both cases the values are those that cannot be the first data item in a UTF
828 character. The three items that might have provoked this were recursions,
829 possessively repeated groups, and atomic groups.
830
831 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
832 and pcregrep, because some OS require shared objects to be explicitly
833 passed to ld, causing the link step to fail if they are not.
834
835 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
836 mode, patterns that started with \h* or \R* might be incorrectly matched.
837
838
839 Version 8.30 04-February-2012
840 -----------------------------
841
842 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
843 name is defined in ctype.h.
844
845 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
846 only in quite long subpatterns.
847
848 3. Removed the function pcre_info(), which has been obsolete and deprecated
849 since it was replaced by pcre_fullinfo() in February 2000.
850
851 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
852 match a (*MARK), and the match failed at the start of the subject, a
853 reference to memory before the start of the subject could occur. This bug
854 was introduced by fix 17 of release 8.21.
855
856 5. A reference to an unset group with zero minimum repetition was giving
857 totally wrong answers (in non-JavaScript-compatibility mode). For example,
858 /(another)?(\1?)test/ matched against "hello world test". This bug was
859 introduced in release 8.13.
860
861 6. Add support for 16-bit character strings (a large amount of work involving
862 many changes and refactorings).
863
864 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
865 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
866 from a file.
867
868 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
869 rounding is not applied in this particular case).
870
871 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
872 if they appear, or are escaped, in patterns.
873
874 10. Get rid of a number of -Wunused-but-set-variable warnings.
875
876 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
877 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
878 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
879 also returns the mark "x". This bug applied to capturing parentheses,
880 non-capturing parentheses, and atomic parentheses. It also applied to some
881 assertions.
882
883 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
884 information out of configure.ac instead of relying on pcre.h.generic, which
885 is not stored in the repository.
886
887 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
888 -lreadline.
889
890 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
891
892 15. Applied Graycode's patch to put the top-level frame on the stack rather
893 than the heap when not using the stack for recursion. This gives a
894 performance improvement in many cases when recursion is not deep.
895
896 16. Experimental code added to "pcretest -C" to output the stack frame size.
897
898
899 Version 8.21 12-Dec-2011
900 ------------------------
901
902 1. Updating the JIT compiler.
903
904 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
905 are added as well.
906
907 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
908 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
909 calling _pcre_jit_exec. Some extra comments are added.
910
911 4. (*MARK) settings inside atomic groups that do not contain any capturing
912 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
913 was introduced by change 18 for 8.20.
914
915 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
916 ECMA-262 standard.
917
918 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
919 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
920 This bug was probably introduced by change 9 of 8.13.
921
922 7. While fixing 6 above, I noticed that a number of other items were being
923 incorrectly rejected as "not fixed length". This arose partly because newer
924 opcodes had not been added to the fixed-length checking code. I have (a)
925 corrected the bug and added tests for these items, and (b) arranged for an
926 error to occur if an unknown opcode is encountered while checking for fixed
927 length instead of just assuming "not fixed length". The items that were
928 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
929 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
930 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
931
932 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
933 being incorrectly compiled and would have given unpredicatble results.
934
935 9. A possessively repeated subpattern with minimum repeat count greater than
936 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
937 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
938 the first (A) could occur when it should not.
939
940 10. Add a cast and remove a redundant test from the code.
941
942 11. JIT should use pcre_malloc/pcre_free for allocation.
943
944 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
945 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
946 is anything other than /usr, -L is still shown).
947
948 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
949
950 14. Perl does not support \N without a following name in a [] class; PCRE now
951 also gives an error.
952
953 15. If a forward reference was repeated with an upper limit of around 2000,
954 it caused the error "internal error: overran compiling workspace". The
955 maximum number of forward references (including repeats) was limited by the
956 internal workspace, and dependent on the LINK_SIZE. The code has been
957 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
958 the default depends on LINK_SIZE. There is a new upper limit (for safety)
959 of around 200,000 forward references. While doing this, I also speeded up
960 the filling in of repeated forward references.
961
962 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
963 incorrectly expecting the subject to contain another "a" after the start.
964
965 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
966 in the match, the SKIP should be ignored. This was not happening; instead
967 the SKIP was being treated as NOMATCH. For patterns such as
968 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
969 tested.
970
971 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
972 now much more compatible with Perl, in particular in cases where the result
973 is a non-match for a non-anchored pattern. For example, if
974 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
975 "m", where previously it did not return a name. A side effect of this
976 change is that for partial matches, the last encountered mark name is
977 returned, as for non matches. A number of tests that were previously not
978 Perl-compatible have been moved into the Perl-compatible test files. The
979 refactoring has had the pleasing side effect of removing one argument from
980 the match() function, thus reducing its stack requirements.
981
982 19. If the /S+ option was used in pcretest to study a pattern using JIT,
983 subsequent uses of /S (without +) incorrectly behaved like /S+.
984
985 21. Retrieve executable code size support for the JIT compiler and fixing
986 some warnings.
987
988 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
989 not work when the shorter character appeared right at the end of the
990 subject string.
991
992 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
993 systems.
994
995 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
996 output it when the /M option is used in pcretest.
997
998 25. The CheckMan script was not being included in the distribution. Also, added
999 an explicit "perl" to run Perl scripts from the PrepareRelease script
1000 because this is reportedly needed in Windows.
1001
1002 26. If study data was being save in a file and studying had not found a set of
1003 "starts with" bytes for the pattern, the data written to the file (though
1004 never used) was taken from uninitialized memory and so caused valgrind to
1005 complain.
1006
1007 27. Updated RunTest.bat as provided by Sheri Pierce.
1008
1009 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1010
1011 29. Computation of memory usage for the table of capturing group names was
1012 giving an unnecessarily large value.
1013
1014
1015 Version 8.20 21-Oct-2011
1016 ------------------------
1017
1018 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1019 a POSIX class. After further experiments with Perl, which convinced me that
1020 Perl has bugs and confusions, a closing square bracket is no longer allowed
1021 in a POSIX name. This bug also affected patterns with classes that started
1022 with full stops.
1023
1024 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1025 captured substring, but while checking the failing first alternative,
1026 substring 1 is temporarily captured. If the output vector supplied to
1027 pcre_exec() was not big enough for this capture, the yield of the function
1028 was still zero ("insufficient space for captured substrings"). This cannot
1029 be totally fixed without adding another stack variable, which seems a lot
1030 of expense for a edge case. However, I have improved the situation in cases
1031 such as /(a)(b)x|abc/ matched against "abc", where the return code
1032 indicates that fewer than the maximum number of slots in the ovector have
1033 been set.
1034
1035 3. Related to (2) above: when there are more back references in a pattern than
1036 slots in the output vector, pcre_exec() uses temporary memory during
1037 matching, and copies in the captures as far as possible afterwards. It was
1038 using the entire output vector, but this conflicts with the specification
1039 that only 2/3 is used for passing back captured substrings. Now it uses
1040 only the first 2/3, for compatibility. This is, of course, another edge
1041 case.
1042
1043 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1044 main code base, and can be used by building with --enable-jit. When this is
1045 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1046 runtime --no-jit option is given.
1047
1048 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1049 ovector, the return from the function was zero, implying that there were
1050 other matches that did not fit. The correct "exactly full" value is now
1051 returned.
1052
1053 6. If a subpattern that was called recursively or as a subroutine contained
1054 (*PRUNE) or any other control that caused it to give a non-standard return,
1055 invalid errors such as "Error -26 (nested recursion at the same subject
1056 position)" or even infinite loops could occur.
1057
1058 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1059 computing the minimum length on reaching *ACCEPT, and so ended up with the
1060 wrong value of 1 rather than 0. Further investigation indicates that
1061 computing a minimum subject length in the presence of *ACCEPT is difficult
1062 (think back references, subroutine calls), and so I have changed the code
1063 so that no minimum is registered for a pattern that contains *ACCEPT.
1064
1065 8. If (*THEN) was present in the first (true) branch of a conditional group,
1066 it was not handled as intended. [But see 16 below.]
1067
1068 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1069 Sheri Pierce.
1070
1071 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1072 the first byte in a match must be "a".
1073
1074 11. Change 17 for 8.13 increased the recursion depth for patterns like
1075 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1076 pattern contains any instances of (*THEN). If it does not, the old
1077 optimizations are restored. It would be nice to do this on a per-group
1078 basis, but at the moment that is not feasible.
1079
1080 12. In some environments, the output of pcretest -C is CRLF terminated. This
1081 broke RunTest's code that checks for the link size. A single white space
1082 character after the value is now allowed for.
1083
1084 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1085 For "fr", it uses the Windows-specific input and output files.
1086
1087 14. If (*THEN) appeared in a group that was called recursively or as a
1088 subroutine, it did not work as intended. [But see next item.]
1089
1090 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1091 pattern fragments (but not containing any | characters). If A and B are
1092 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1093 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1094 D. In other words, Perl considers parentheses that do not contain any |
1095 characters to be part of a surrounding alternative, whereas PCRE was
1096 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1097 differently. PCRE now behaves in the same way as Perl, except in the case
1098 of subroutine/recursion calls such as (?1) which have in any case always
1099 been different (but PCRE had them first :-).
1100
1101 16. Related to 15 above: Perl does not treat the | in a conditional group as
1102 creating alternatives. Such a group is treated in the same way as an
1103 ordinary group without any | characters when processing (*THEN). PCRE has
1104 been changed to match Perl's behaviour.
1105
1106 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1107 RunGrepTest script failed.
1108
1109 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1110 inevitable for groups that contain captures, but it can lead to a lot of
1111 stack use in large patterns. The old behaviour has been restored for atomic
1112 groups that do not contain any capturing parentheses.
1113
1114 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1115 suppress the check for a minimum subject length at run time. (If it was
1116 given to pcre_exec() or pcre_dfa_exec() it did work.)
1117
1118 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1119 fail to work when decoding hex characters in data strings in EBCDIC
1120 environments.
1121
1122 21. It appears that in at least one Mac OS environment, the isxdigit() function
1123 is implemented as a macro that evaluates to its argument more than once,
1124 contravening the C 90 Standard (I haven't checked a later standard). There
1125 was an instance in pcretest which caused it to go wrong when processing
1126 \x{...} escapes in subject strings. The has been rewritten to avoid using
1127 things like p++ in the argument of isxdigit().
1128
1129
1130 Version 8.13 16-Aug-2011
1131 ------------------------
1132
1133 1. The Unicode data tables have been updated to Unicode 6.0.0.
1134
1135 2. Two minor typos in pcre_internal.h have been fixed.
1136
1137 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1138 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1139 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1140
1141 4. There were a number of related bugs in the code for matching backrefences
1142 caselessly in UTF-8 mode when codes for the characters concerned were
1143 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1144 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1145 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1146 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1147 2-byte code at the end of the subject (it thought there wasn't enough data
1148 left).
1149
1150 5. Comprehensive information about what went wrong is now returned by
1151 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1152 as the output vector has at least 2 elements. The offset of the start of
1153 the failing character and a reason code are placed in the vector.
1154
1155 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1156 now returned is for the first byte of the failing character, instead of the
1157 last byte inspected. This is an incompatible change, but I hope it is small
1158 enough not to be a problem. It makes the returned offset consistent with
1159 pcre_exec() and pcre_dfa_exec().
1160
1161 7. pcretest now gives a text phrase as well as the error number when
1162 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1163 failure, the offset and reason code are output.
1164
1165 8. When \R was used with a maximizing quantifier it failed to skip backwards
1166 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1167 back over a single character (\n). This seems wrong (because it treated the
1168 two characters as a single entity when going forwards), conflicts with the
1169 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1170 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1171 has been changed.
1172
1173 9. Some internal refactoring has changed the processing so that the handling
1174 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1175 time (the PCRE_DOTALL option was changed this way some time ago: version
1176 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1177 which was always a bit of a fudge. It also means that there is one less
1178 argument for the match() function, which reduces its stack requirements
1179 slightly. This change also fixes an incompatibility with Perl: the pattern
1180 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1181
1182 10. More internal refactoring has drastically reduced the number of recursive
1183 calls to match() for possessively repeated groups such as (abc)++ when
1184 using pcre_exec().
1185
1186 11. While implementing 10, a number of bugs in the handling of groups were
1187 discovered and fixed:
1188
1189 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1190 (a|)*(?1) gave a compile-time internal error.
1191 ((a|)+)+ did not notice that the outer group could match an empty string.
1192 (^a|^)+ was not marked as anchored.
1193 (.*a|.*)+ was not marked as matching at start or after a newline.
1194
1195 12. Yet more internal refactoring has removed another argument from the match()
1196 function. Special calls to this function are now indicated by setting a
1197 value in a variable in the "match data" data block.
1198
1199 13. Be more explicit in pcre_study() instead of relying on "default" for
1200 opcodes that mean there is no starting character; this means that when new
1201 ones are added and accidentally left out of pcre_study(), testing should
1202 pick them up.
1203
1204 14. The -s option of pcretest has been documented for ages as being an old
1205 synonym of -m (show memory usage). I have changed it to mean "force study
1206 for every regex", that is, assume /S for every regex. This is similar to -i
1207 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1208 using it. It makes it easier to run collections of tests with and without
1209 study enabled, and thereby test pcre_study() more easily. All the standard
1210 tests are now run with and without -s (but some patterns can be marked as
1211 "never study" - see 20 below).
1212
1213 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1214 restoration of the capturing data to the outer values was not happening
1215 correctly.
1216
1217 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1218 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1219 pattern had matched an empty string, and so incorrectly returned a no
1220 match.
1221
1222 17. There was optimizing code for the last branch of non-capturing parentheses,
1223 and also for the obeyed branch of a conditional subexpression, which used
1224 tail recursion to cut down on stack usage. Unfortunately, now that there is
1225 the possibility of (*THEN) occurring in these branches, tail recursion is
1226 no longer possible because the return has to be checked for (*THEN). These
1227 two optimizations have therefore been removed. [But see 8.20/11 above.]
1228
1229 18. If a pattern containing \R was studied, it was assumed that \R always
1230 matched two bytes, thus causing the minimum subject length to be
1231 incorrectly computed because \R can also match just one byte.
1232
1233 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1234 was incorrectly computed.
1235
1236 20. If /S is present twice on a test pattern in pcretest input, it now
1237 *disables* studying, thereby overriding the use of -s on the command line
1238 (see 14 above). This is necessary for one or two tests to keep the output
1239 identical in both cases.
1240
1241 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1242 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1243
1244 22. When an atomic group that contained a capturing parenthesis was
1245 successfully matched, but the branch in which it appeared failed, the
1246 capturing was not being forgotten if a higher numbered group was later
1247 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1248 group 1 to "a", when in fact it should be unset. This applied to multi-
1249 branched capturing and non-capturing groups, repeated or not, and also to
1250 positive assertions (capturing in negative assertions does not happen
1251 in PCRE) and also to nested atomic groups.
1252
1253 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1254 subject after a captured substring, to make it easier to tell which of a
1255 number of identical substrings has been captured.
1256
1257 24. The way atomic groups are processed by pcre_exec() has been changed so that
1258 if they are repeated, backtracking one repetition now resets captured
1259 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1260 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1261 "aaa". Previously, it would have been "a". As part of this code
1262 refactoring, the way recursive calls are handled has also been changed.
1263
1264 25. If an assertion condition captured any substrings, they were not passed
1265 back unless some other capturing happened later. For example, if
1266 (?(?=(a))a) was matched against "a", no capturing was returned.
1267
1268 26. When studying a pattern that contained subroutine calls or assertions,
1269 the code for finding the minimum length of a possible match was handling
1270 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1271 group 1 called group 2 while simultaneously a separate group 2 called group
1272 1). A stack overflow occurred in this case. I have fixed this by limiting
1273 the recursion depth to 10.
1274
1275 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1276 Fortmann. This supports explicit test numbers on the command line, and has
1277 argument validation and error reporting.
1278
1279 28. An instance of \X with an unlimited repeat could fail if at any point the
1280 first character it looked at was a mark character.
1281
1282 29. Some minor code refactoring concerning Unicode properties and scripts
1283 should reduce the stack requirement of match() slightly.
1284
1285 30. Added the '=' option to pcretest to check the setting of unused capturing
1286 slots at the end of the pattern, which are documented as being -1, but are
1287 not included in the return count.
1288
1289 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1290 compiled something random. Now it gives a compile-time error (as does
1291 Perl).
1292
1293 32. A *MARK encountered during the processing of a positive assertion is now
1294 recorded and passed back (compatible with Perl).
1295
1296 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1297 had alternative anchored branches, the search for a second match in a line
1298 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1299 matched the line "0102" twice. The same bug affected patterns that started
1300 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1301 twice.
1302
1303 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1304 does, and because of capturing effects, quantifying parenthesized
1305 assertions may at times be useful. Quantifiers are now allowed for
1306 parenthesized assertions.
1307
1308 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1309
1310 36. \g was being checked for fancy things in a character class, when it should
1311 just be a literal "g".
1312
1313 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1314 appearance of a nested POSIX class supersedes an apparent external class.
1315 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1316 unescaped square brackets may also appear as part of class names. For
1317 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1318 more like Perl. (But see 8.20/1 above.)
1319
1320 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1321 was because it thought it was \N{name}, which is not supported).
1322
1323 39. Add minix to OS list not supporting the -S option in pcretest.
1324
1325 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1326 cannot analyze patterns in sufficient detail to catch mutual recursions
1327 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1328 subgroup is called recursively as a subpattern for a second time at the
1329 same position in the subject string. In previous releases this might have
1330 been caught by the recursion limit, or it might have run out of stack.
1331
1332 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1333 happen only once. PCRE was, however incorrectly giving a compile time error
1334 "recursive call could loop indefinitely" because it cannot analyze the
1335 pattern in sufficient detail. The compile time test no longer happens when
1336 PCRE is compiling a conditional subpattern, but actual runaway loops are
1337 now caught at runtime (see 40 above).
1338
1339 42. It seems that Perl allows any characters other than a closing parenthesis
1340 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1341 has been changed to be the same.
1342
1343 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1344 as not to get warnings when autogen.sh is called. Also changed
1345 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1346
1347 44. To help people who use pcregrep to scan files containing exceedingly long
1348 lines, the following changes have been made:
1349
1350 (a) The default value of the buffer size parameter has been increased from
1351 8K to 20K. (The actual buffer used is three times this size.)
1352
1353 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1354 PCRE is built.
1355
1356 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1357 to be set at run time.
1358
1359 (d) Numerical values in pcregrep options can be followed by K or M, for
1360 example --buffer-size=50K.
1361
1362 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1363 given and the return code is set to 2.
1364
1365 45. Add a pointer to the latest mark to the callout data block.
1366
1367 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1368 partial match of an empty string instead of no match. This was specific to
1369 the use of ".".
1370
1371 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1372 complete match instead of a partial match. This bug was dependent on both
1373 the PCRE_UTF8 and PCRE_DOTALL options being set.
1374
1375 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1376 starting byte set, because \b was not being ignored.
1377
1378
1379 Version 8.12 15-Jan-2011
1380 ------------------------
1381
1382 1. Fixed some typos in the markup of the man pages, and wrote a script that
1383 checks for such things as part of the documentation building process.
1384
1385 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1386 --match-limit and --recursion-limit options (added for 8.11). In
1387 particular, this made one of the standard tests fail. (The integer value
1388 went into the wrong half of a long int.)
1389
1390 3. If the --colour option was given to pcregrep with -v (invert match), it
1391 did strange things, either producing crazy output, or crashing. It should,
1392 of course, ignore a request for colour when reporting lines that do not
1393 match.
1394
1395 4. Another pcregrep bug caused similar problems if --colour was specified with
1396 -M (multiline) and the pattern match finished with a line ending.
1397
1398 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1399 matched in multiline mode, the following line was shown as part of the
1400 match. This seems wrong, so I have changed it.
1401
1402 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1403 the check for further matches in the same line (so they could be coloured)
1404 to overrun the end of the current line. If another match was found, it was
1405 incorrectly shown (and then shown again when found in the next line).
1406
1407 7. If pcregrep was compiled under Windows, there was a reference to the
1408 function pcregrep_exit() before it was defined. I am assuming this was
1409 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1410 reported by a user. I've moved the definition above the reference.
1411
1412
1413 Version 8.11 10-Dec-2010
1414 ------------------------
1415
1416 1. (*THEN) was not working properly if there were untried alternatives prior
1417 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1418 backtracked to try for "b" instead of moving to the next alternative branch
1419 at the same level (in this case, to look for "c"). The Perl documentation
1420 is clear that when (*THEN) is backtracked onto, it goes to the "next
1421 alternative in the innermost enclosing group".
1422
1423 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1424 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1425 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1426 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1427 (*THEN).
1428
1429 3. If \s appeared in a character class, it removed the VT character from
1430 the class, even if it had been included by some previous item, for example
1431 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1432 of \s, but is part of the POSIX "space" class.)
1433
1434 4. A partial match never returns an empty string (because you can always
1435 match an empty string at the end of the subject); however the checking for
1436 an empty string was starting at the "start of match" point. This has been
1437 changed to the "earliest inspected character" point, because the returned
1438 data for a partial match starts at this character. This means that, for
1439 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1440 (previously it gave "no match").
1441
1442 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1443 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1444 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1445 has an implication that the given string is incomplete (because a partial
1446 match is preferred over a full match). For this reason, these items now
1447 give a partial match in this situation. [Aside: previously, the one case
1448 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1449 match rather than a full match, which was wrong by the old rules, but is
1450 now correct.]
1451
1452 6. There was a bug in the handling of #-introduced comments, recognized when
1453 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1454 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1455 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1456 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1457 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1458 places in pcre_compile().
1459
1460 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1461 comments when looking ahead for named forward references to subpatterns,
1462 the only newline sequence it recognized was NL. It now handles newlines
1463 according to the set newline convention.
1464
1465 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1466 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1467 cater for a lack of strerror(). These oversights have been fixed.
1468
1469 9. Added --match-limit and --recursion-limit to pcregrep.
1470
1471 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1472
1473 11. When the -o option was used, pcregrep was setting a return code of 1, even
1474 when matches were found, and --line-buffered was not being honoured.
1475
1476 12. Added an optional parentheses number to the -o and --only-matching options
1477 of pcregrep.
1478
1479 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1480 can match an empty string. The code to do it in pcretest and pcredemo
1481 needed fixing:
1482
1483 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1484 only one byte after an empty string match just before CRLF (this case
1485 just got forgotten; "any" and "anycrlf" were OK).
1486
1487 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1488 mode when an empty string match preceded an ASCII character followed by
1489 a non-ASCII character. (The code for advancing by one character rather
1490 than one byte was nonsense.)
1491
1492 (c) The pcredemo.c sample program did not have any code at all to handle
1493 the cases when CRLF is a valid newline sequence.
1494
1495 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1496 as a starting offset was within the subject string. There is now a new
1497 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1498 negative or greater than the length of the string. In order to test this,
1499 pcretest is extended to allow the setting of negative starting offsets.
1500
1501 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1502 starting offset points to the beginning of a UTF-8 character was
1503 unnecessarily clumsy. I tidied it up.
1504
1505 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1506 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1507
1508 17. Nobody had reported that the --include_dir option, which was added in
1509 release 7.7 should have been called --include-dir (hyphen, not underscore)
1510 for compatibility with GNU grep. I have changed it to --include-dir, but
1511 left --include_dir as an undocumented synonym, and the same for
1512 --exclude-dir, though that is not available in GNU grep, at least as of
1513 release 2.5.4.
1514
1515 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1516 characters from a string of bytes) have been redefined so as not to use
1517 loops, in order to improve performance in some environments. At the same
1518 time, I abstracted some of the common code into auxiliary macros to save
1519 repetition (this should not affect the compiled code).
1520
1521 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1522 compile-time error is now given if \c is not followed by an ASCII
1523 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1524 different, and any byte value is allowed.)
1525
1526 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1527 START_OPTIMIZE option, which is now allowed at compile time - but just
1528 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1529 to pcregrep and other applications that have no direct access to PCRE
1530 options. The new /Y option in pcretest sets this option when calling
1531 pcre_compile().
1532
1533 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1534 back references. Groups containing recursive back references were forced to
1535 be atomic by that change, but in the case of named groups, the amount of
1536 memory required was incorrectly computed, leading to "Failed: internal
1537 error: code overflow". This has been fixed.
1538
1539 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1540 pcretest.c, to avoid build problems in some Borland environments.
1541
1542
1543 Version 8.10 25-Jun-2010
1544 ------------------------
1545
1546 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1547 THEN.
1548
1549 2. (*ACCEPT) was not working when inside an atomic group.
1550
1551 3. Inside a character class, \B is treated as a literal by default, but
1552 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1553 causes the error). The code is unchanged, but I tidied the documentation.
1554
1555 4. Inside a character class, PCRE always treated \R and \X as literals,
1556 whereas Perl faults them if its -w option is set. I have changed PCRE so
1557 that it faults them when PCRE_EXTRA is set.
1558
1559 5. Added support for \N, which always matches any character other than
1560 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1561
1562 6. When compiling pcregrep with newer versions of gcc which may have
1563 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1564 declared with attribute warn_unused_result" were given. Just casting the
1565 result to (void) does not stop the warnings; a more elaborate fudge is
1566 needed. I've used a macro to implement this.
1567
1568 7. Minor change to pcretest.c to avoid a compiler warning.
1569
1570 8. Added four artifical Unicode properties to help with an option to make
1571 \s etc use properties (see next item). The new properties are: Xan
1572 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1573
1574 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1575 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1576 this option. Modified pcretest to add /W to test this facility. Added
1577 REG_UCP to make it available via the POSIX interface.
1578
1579 10. Added --line-buffered to pcregrep.
1580
1581 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1582 studied, and the match started with a letter with a code point greater than
1583 127 whose first byte was different to the first byte of the other case of
1584 the letter, the other case of this starting letter was not recognized
1585 (#976).
1586
1587 12. If a pattern that was studied started with a repeated Unicode property
1588 test, for example, \p{Nd}+, there was the theoretical possibility of
1589 setting up an incorrect bitmap of starting bytes, but fortunately it could
1590 not have actually happened in practice until change 8 above was made (it
1591 added property types that matched character-matching opcodes).
1592
1593 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1594 possible starting bytes for non-anchored patterns.
1595
1596 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1597 \R, and also a number of cases that involve Unicode properties, both
1598 explicit and implicit when PCRE_UCP is set.
1599
1600 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1601 input, it could crash or give wrong results if characters with values
1602 greater than 0xc0 were present in the subject string. (Detail: it assumed
1603 UTF-8 input when processing these items.)
1604
1605 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1606 size_t is 64-bit (#991).
1607
1608 17. Added a check for running out of memory when PCRE is compiled with
1609 --disable-stack-for-recursion (#990).
1610
1611 18. If the last data line in a file for pcretest does not have a newline on
1612 the end, a newline was missing in the output.
1613
1614 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1615 less than 128) in its various bitmaps. However, there is a facility for
1616 generating tables according to the current locale when PCRE is compiled. It
1617 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1618 characters, are recognized by isspace() and therefore were getting set in
1619 these tables, and indeed these tables seem to approximate to ISO 8859. This
1620 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1621 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1622 which of course cannot start UTF-8 characters. I have changed the code so
1623 that only real ASCII characters (less than 128) and the correct starting
1624 bytes for UTF-8 encodings are set for characters greater than 127 when in
1625 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1626 altogether.)
1627
1628 20. Added the /T option to pcretest so as to be able to run tests with non-
1629 standard character tables, thus making it possible to include the tests
1630 used for 19 above in the standard set of tests.
1631
1632 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1633 reference to a subpattern the other side of a comment that contains an
1634 opening parenthesis caused either an internal compiling error, or a
1635 reference to the wrong subpattern.
1636
1637
1638 Version 8.02 19-Mar-2010
1639 ------------------------
1640
1641 1. The Unicode data tables have been updated to Unicode 5.2.0.
1642
1643 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1644 configured.
1645
1646 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1647 original author of that file, following a query about its status.
1648
1649 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1650 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1651
1652 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1653 quantifier applied to a forward-referencing subroutine call, could compile
1654 incorrect code or give the error "internal error: previously-checked
1655 referenced subpattern not found".
1656
1657 6. Both MS Visual Studio and Symbian OS have problems with initializing
1658 variables to point to external functions. For these systems, therefore,
1659 pcre_malloc etc. are now initialized to local functions that call the
1660 relevant global functions.
1661
1662 7. There were two entries missing in the vectors called coptable and poptable
1663 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1664 I've fixed the data, and added a kludgy way of testing at compile time that
1665 the lengths are correct (equal to the number of opcodes).
1666
1667 8. Following on from 7, I added a similar kludge to check the length of the
1668 eint vector in pcreposix.c.
1669
1670 9. Error texts for pcre_compile() are held as one long string to avoid too
1671 much relocation at load time. To find a text, the string is searched,
1672 counting zeros. There was no check for running off the end of the string,
1673 which could happen if a new error number was added without updating the
1674 string.
1675
1676 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1677
1678 11. \K was not working if it appeared in an atomic group or in a group that
1679 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1680 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1681 the assertion is positive, but not if it is negative.
1682
1683 12. Change 11 fortuitously reduced the size of the stack frame used in the
1684 "match()" function of pcre_exec.c by one pointer. Forthcoming
1685 implementation of support for (*MARK) will need an extra pointer on the
1686 stack; I have reserved it now, so that the stack frame size does not
1687 decrease.
1688
1689 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1690 item in branch that calls a recursion is a subroutine call - as in the
1691 second branch in the above example - was incorrectly given the compile-
1692 time error "recursive call could loop indefinitely" because pcre_compile()
1693 was not correctly checking the subroutine for matching a non-empty string.
1694
1695 14. The checks for overrunning compiling workspace could trigger after an
1696 overrun had occurred. This is a "should never occur" error, but it can be
1697 triggered by pathological patterns such as hundreds of nested parentheses.
1698 The checks now trigger 100 bytes before the end of the workspace.
1699
1700 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1701
1702
1703 Version 8.01 19-Jan-2010
1704 ------------------------
1705
1706 1. If a pattern contained a conditional subpattern with only one branch (in
1707 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1708 computed the wrong minimum data length (which is of course zero for such
1709 subpatterns). This could cause incorrect "no match" results.
1710
1711 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1712 the pattern is reset in the first branch, pcre_compile() failed with
1713 "internal error: code overflow at offset...". This happened only when
1714 the reset was to the original external option setting. (An optimization
1715 abstracts leading options settings into an external setting, which was the
1716 cause of this.)
1717
1718 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1719 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1720 assertion pattern did not match (meaning that the assertion was true), it
1721 was incorrectly treated as false if the SKIP had been reached during the
1722 matching. This also applied to assertions used as conditions.
1723
1724 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1725 assertion subpattern, including such a pattern used as a condition,
1726 unpredictable results occurred, instead of the error return
1727 PCRE_ERROR_DFA_UITEM.
1728
1729 5. The C++ GlobalReplace function was not working like Perl for the special
1730 situation when an empty string is matched. It now does the fancy magic
1731 stuff that is necessary.
1732
1733 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1734 removed. (These were left over from very, very early versions of PCRE.)
1735
1736 7. Some cosmetic changes to the code to make life easier when compiling it
1737 as part of something else:
1738
1739 (a) Change DEBUG to PCRE_DEBUG.
1740
1741 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1742 called "current" as "current_branch", to prevent a collision with the
1743 Linux macro when compiled as a kernel module.
1744
1745 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1746 prevent a collision with the Linux macro when compiled as a kernel
1747 module.
1748
1749 8. In pcre_compile() there are some checks for integer overflows that used to
1750 cast potentially large values to (double). This has been changed to that
1751 when building, a check for int64_t is made, and if it is found, it is used
1752 instead, thus avoiding the use of floating point arithmetic. (There is no
1753 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1754 double.
1755
1756 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1757 2005 (difference between two addresses compared to an unsigned value).
1758
1759 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1760 custom one, because of the following reported problem in Windows:
1761
1762 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1763 under Win32.
1764 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1765 therefore missing the function definition.
1766 - The compiler thus generates a "C" signature for the test function.
1767 - The linker fails to find the "C" function.
1768 - PCRE fails to configure if asked to do so against libbz2.
1769
1770 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1771 messages were output:
1772
1773 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1774 rerunning libtoolize, to keep the correct libtool macros in-tree.
1775 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1776
1777 I have done both of these things.
1778
1779 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1780 most of the time, it *can* run out if it is given a pattern that contains a
1781 runaway infinite recursion. I updated the discussion in the pcrestack man
1782 page.
1783
1784 13. Now that we have gone to the x.xx style of version numbers, the minor
1785 version may start with zero. Using 08 or 09 is a bad idea because users
1786 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1787 interpreted as invalid octal numbers. I've updated the previous comment in
1788 configure.ac, and also added a check that gives an error if 08 or 09 are
1789 used.
1790
1791 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1792 causing partial matching to fail when the end of the subject matched \W
1793 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1794
1795 15. There were some discrepancies between the declarations in pcre_internal.h
1796 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1797 their definitions. The declarations used "const uschar *" and the
1798 definitions used USPTR. Even though USPTR is normally defined as "const
1799 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1800 reported that: "This difference in casting confuses some C++ compilers, for
1801 example, SunCC recognizes above declarations as different functions and
1802 generates broken code for hbpcre." I have changed the declarations to use
1803 USPTR.
1804
1805 16. GNU libtool is named differently on some systems. The autogen.sh script now
1806 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1807 (FreeBSD).
1808
1809 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1810 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1811 comment: "Figure out how to create a longlong from a string: strtoll and
1812 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1813 instance, but it only takes 2 args instead of 3!"
1814
1815 18. A subtle bug concerned with back references has been fixed by a change of
1816 specification, with a corresponding code fix. A pattern such as
1817 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1818 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1819 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1820 same bug. Such groups have to be quantified to be useful, or contained
1821 inside another quantified group. (If there's no repetition, the reference
1822 can never match.) The problem arises because, having left the group and
1823 moved on to the rest of the pattern, a later failure that backtracks into
1824 the group uses the captured value from the final iteration of the group
1825 rather than the correct earlier one. I have fixed this in PCRE by forcing
1826 any group that contains a reference to itself to be an atomic group; that
1827 is, there cannot be any backtracking into it once it has completed. This is
1828 similar to recursive and subroutine calls.
1829
1830
1831 Version 8.00 19-Oct-09
1832 ----------------------
1833
1834 1. The table for translating pcre_compile() error codes into POSIX error codes
1835 was out-of-date, and there was no check on the pcre_compile() error code
1836 being within the table. This could lead to an OK return being given in
1837 error.
1838
1839 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1840 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1841 in a Windows environment.
1842
1843 3. The pcregrep --count option prints the count for each file even when it is
1844 zero, as does GNU grep. However, pcregrep was also printing all files when
1845 --files-with-matches was added. Now, when both options are given, it prints
1846 counts only for those files that have at least one match. (GNU grep just
1847 prints the file name in this circumstance, but including the count seems
1848 more useful - otherwise, why use --count?) Also ensured that the
1849 combination -clh just lists non-zero counts, with no names.
1850
1851 4. The long form of the pcregrep -F option was incorrectly implemented as
1852 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1853 but it seems right to fix it, and I didn't think it was worth preserving
1854 the old behaviour.
1855
1856 5. The command line items --regex=pattern and --regexp=pattern were not
1857 recognized by pcregrep, which required --regex pattern or --regexp pattern
1858 (with a space rather than an '='). The man page documented the '=' forms,
1859 which are compatible with GNU grep; these now work.
1860
1861 6. No libpcreposix.pc file was created for pkg-config; there was just
1862 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1863
1864 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1865 when UCP support is not needed, by modifying the Python script that
1866 generates it from Unicode data files. This should not matter if the module
1867 is correctly used as a library, but I received one complaint about 50K of
1868 unwanted data. My guess is that the person linked everything into his
1869 program rather than using a library. Anyway, it does no harm.
1870
1871 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1872 was a minimum greater than 1 for a wide character in a possessive
1873 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1874 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1875 character. Chaos in the form of incorrect output or a compiling loop could
1876 result.
1877
1878 9. The restrictions on what a pattern can contain when partial matching is
1879 requested for pcre_exec() have been removed. All patterns can now be
1880 partially matched by this function. In addition, if there are at least two
1881 slots in the offset vector, the offset of the earliest inspected character
1882 for the match and the offset of the end of the subject are set in them when
1883 PCRE_ERROR_PARTIAL is returned.
1884
1885 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1886 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1887 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1888 and may be more useful for multi-segment matching.
1889
1890 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1891 used to be given if ever the end of the subject was reached; now it is
1892 given only if matching could not proceed because another character was
1893 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1894 string "Z", which now yields "no match" instead of "partial match". In the
1895 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1896 final character ended with (*FAIL).
1897
1898 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1899 if the pattern had a "must contain" character that was already found in the
1900 earlier partial match, unless partial matching was again requested. For
1901 example, with the pattern /dog.(body)?/, the "must contain" character is
1902 "g". If the first part-match was for the string "dog", restarting with
1903 "sbody" failed. This bug has been fixed.
1904
1905 13. The string returned by pcre_dfa_exec() after a partial match has been
1906 changed so that it starts at the first inspected character rather than the
1907 first character of the match. This makes a difference only if the pattern
1908 starts with a lookbehind assertion or \b or \B (\K is not supported by
1909 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1910 matching functions compatible, and I think it's the right thing to do.
1911
1912 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1913 so that the demonstration program is easily available in environments where
1914 PCRE has not been installed from source.
1915
1916 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1917 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1918 library.
1919
1920 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1921 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1922 is not the first non-POSIX option to be added. Clearly some people find
1923 these options useful.
1924
1925 17. If a caller to the POSIX matching function regexec() passes a non-zero
1926 value for nmatch with a NULL value for pmatch, the value of
1927 nmatch is forced to zero.
1928
1929 18. RunGrepTest did not have a test for the availability of the -u option of
1930 the diff command, as RunTest does. It now checks in the same way as
1931 RunTest, and also checks for the -b option.
1932
1933 19. If an odd number of negated classes containing just a single character
1934 interposed, within parentheses, between a forward reference to a named
1935 subpattern and the definition of the subpattern, compilation crashed with
1936 an internal error, complaining that it could not find the referenced
1937 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1938 [The bug was that it was starting one character too far in when skipping
1939 over the character class, thus treating the ] as data rather than
1940 terminating the class. This meant it could skip too much.]
1941
1942 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1943 /g option in pcretest when the pattern contains \K, which makes it possible
1944 to have an empty string match not at the start, even when the pattern is
1945 anchored. Updated pcretest and pcredemo to use this option.
1946
1947 21. If the maximum number of capturing subpatterns in a recursion was greater
1948 than the maximum at the outer level, the higher number was returned, but
1949 with unset values at the outer level. The correct (outer level) value is
1950 now given.
1951
1952 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1953 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1954 make it do so. The string so far is captured, making this feature
1955 compatible with Perl.
1956
1957 23. The tests have been re-organized, adding tests 11 and 12, to make it
1958 possible to check the Perl 5.10 features against Perl 5.10.
1959
1960 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1961 pattern matches a fixed length string. PCRE did not allow this; now it
1962 does. Neither allows recursion.
1963
1964 25. I finally figured out how to implement a request to provide the minimum
1965 length of subject string that was needed in order to match a given pattern.
1966 (It was back references and recursion that I had previously got hung up
1967 on.) This code has now been added to pcre_study(); it finds a lower bound
1968 to the length of subject needed. It is not necessarily the greatest lower
1969 bound, but using it to avoid searching strings that are too short does give
1970 some useful speed-ups. The value is available to calling programs via
1971 pcre_fullinfo().
1972
1973 26. While implementing 25, I discovered to my embarrassment that pcretest had
1974 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1975 study optimizations had never been tested with that matching function.
1976 Oops. What is worse, even when it was passed study data, there was a bug in
1977 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1978 were also very few tests of studied patterns with pcre_dfa_exec().
1979
1980 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1981 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1982 on the other side of the coin, they are no longer allowed to have different
1983 names, because these cannot be distinguished in PCRE, and this has caused
1984 confusion. (This is a difference from Perl.)
1985
1986 28. When duplicate subpattern names are present (necessarily with different
1987 numbers, as required by 27 above), and a test is made by name in a
1988 conditional pattern, either for a subpattern having been matched, or for
1989 recursion in such a pattern, all the associated numbered subpatterns are
1990 tested, and the overall condition is true if the condition is true for any
1991 one of them. This is the way Perl works, and is also more like the way
1992 testing by number works.
1993
1994
1995 Version 7.9 11-Apr-09
1996 ---------------------
1997
1998 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1999 (pcretest), all targets were linked against these libraries. This included
2000 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2001 libraries. This caused unwanted dependencies to be created. This problem
2002 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2003 pcretest is linked with readline.
2004
2005 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2006 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2007 moved outside it again, because FALSE and TRUE are already defined in AIX,
2008 but BOOL is not.
2009
2010 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2011 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2012
2013 4. The pcregrep documentation said spaces were inserted as well as colons (or
2014 hyphens) following file names and line numbers when outputting matching
2015 lines. This is not true; no spaces are inserted. I have also clarified the
2016 wording for the --colour (or --color) option.
2017
2018 5. In pcregrep, when --colour was used with -o, the list of matching strings
2019 was not coloured; this is different to GNU grep, so I have changed it to be
2020 the same.
2021
2022 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2023 each matching line was coloured. Now it goes on to look for further matches
2024 of any of the test patterns, which is the same behaviour as GNU grep.
2025
2026 7. A pattern that could match an empty string could cause pcregrep to loop; it
2027 doesn't make sense to accept an empty string match in pcregrep, so I have
2028 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2029 seems to be how GNU grep behaves. [But see later change 40 for release
2030 8.33.]
2031
2032 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2033 start or after a newline", because the conditional assertion was not being
2034 correctly handled. The rule now is that both the assertion and what follows
2035 in the first alternative must satisfy the test.
2036
2037 9. If auto-callout was enabled in a pattern with a conditional group whose
2038 condition was an assertion, PCRE could crash during matching, both with
2039 pcre_exec() and pcre_dfa_exec().
2040
2041 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2042 used for matching.
2043
2044 11. Unicode property support in character classes was not working for
2045 characters (bytes) greater than 127 when not in UTF-8 mode.
2046
2047 12. Added the -M command line option to pcretest.
2048
2049 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2050
2051 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2052
2053 16. Added comments and documentation about mis-use of no_arg in the C++
2054 wrapper.
2055
2056 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2057 from Martin Jerabek that uses macro names for all relevant character and
2058 string constants.
2059
2060 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2061 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2062 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2063 these, but not everybody uses configure.
2064
2065 19. A conditional group that had only one branch was not being correctly
2066 recognized as an item that could match an empty string. This meant that an
2067 enclosing group might also not be so recognized, causing infinite looping
2068 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2069 with the subject "ab", where knowledge that the repeated group can match
2070 nothing is needed in order to break the loop.
2071
2072 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2073 exec(), but without supplying a callout function, matching went wrong.
2074
2075 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2076 leak if the size of the offset vector was greater than 30. When the vector
2077 is smaller, the saved offsets during recursion go onto a local stack
2078 vector, but for larger vectors malloc() is used. It was failing to free
2079 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2080 error, in fact).
2081
2082 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2083 heapframe that is used only when UTF-8 support is enabled. This caused no
2084 problem, but was untidy.
2085
2086 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2087 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2088 included within another project.
2089
2090 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2091 slightly modified by me:
2092
2093 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2094 not building pcregrep.
2095
2096 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2097 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2098
2099 25. Forward references, both numeric and by name, in patterns that made use of
2100 duplicate group numbers, could behave incorrectly or give incorrect errors,
2101 because when scanning forward to find the reference group, PCRE was not
2102 taking into account the duplicate group numbers. A pattern such as
2103 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2104
2105 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2106 the feature of a custom pointer more persuasive (as requested by a user).
2107
2108 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2109 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2110 pre-defined.
2111
2112 28. Added support for (*UTF8) at the start of a pattern.
2113
2114 29. Arrange for flags added by the "release type" setting in CMake to be shown
2115 in the configuration summary.
2116
2117
2118 Version 7.8 05-Sep-08
2119 ---------------------
2120
2121 1. Replaced UCP searching code with optimized version as implemented for Ad
2122 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2123 stage table and inline lookup instead of a function, giving speed ups of 2
2124 to 5 times on some simple patterns that I tested. Permission was given to
2125 distribute the MultiStage2.py script that generates the tables (it's not in
2126 the tarball, but is in the Subversion repository).
2127
2128 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2129 scripts.
2130
2131 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2132 a group with a zero qualifier. The result of the study could be incorrect,
2133 or the function might crash, depending on the pattern.
2134
2135 4. Caseless matching was not working for non-ASCII characters in back
2136 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2137 It now works when Unicode Property Support is available.
2138
2139 5. In pcretest, an escape such as \x{de} in the data was always generating
2140 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2141 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2142 truncation.
2143
2144 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2145
2146 7. Added two (int) casts to pcregrep when printing the difference of two
2147 pointers, in case they are 64-bit values.
2148
2149 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2150 test 2 if it fails.
2151
2152 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2153 and a #define of that name to empty if it is not externally set. This is to
2154 allow users of MSVC to set it if necessary.
2155
2156 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2157 the convenience functions in the pcre_get.c source file.
2158
2159 11. An option change at the start of a pattern that had top-level alternatives
2160 could cause overwriting and/or a crash. This command provoked a crash in
2161 some environments:
2162
2163 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2164
2165 This potential security problem was recorded as CVE-2008-2371.
2166
2167 12. For a pattern where the match had to start at the beginning or immediately
2168 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2169 pcre_dfa_exec() could read past the end of the passed subject if there was
2170 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2171 pcretest so that it places the subject at the end of its malloc-ed buffer.
2172
2173 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2174 exec() might read past the end of the data buffer in UTF-8 mode.
2175
2176 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2177 the data contained the byte 0x85 as part of a UTF-8 character within its
2178 first line. This applied both to normal and DFA matching.
2179
2180 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2181 /^[^d]*?$/8 failed to match "abc".
2182
2183 16. Added a missing copyright notice to pcrecpp_internal.h.
2184
2185 17. Make it more clear in the documentation that values returned from
2186 pcre_exec() in ovector are byte offsets, not character counts.
2187
2188 18. Tidied a few places to stop certain compilers from issuing warnings.
2189
2190 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2191 supplied by Stefan Weber. I made a further small update for 7.8 because
2192 there is a change of source arrangements: the pcre_searchfuncs.c module is
2193 replaced by pcre_ucd.c.
2194
2195
2196 Version 7.7 07-May-08
2197 ---------------------
2198
2199 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2200 a string to a long long, pretend we don't even have a long long." This is
2201 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2202
2203 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2204 pre-7.6 versions, which defined a global no_arg variable instead of putting
2205 it in the RE class. (See also #8 below.)
2206
2207 3. Remove a line of dead code, identified by coverity and reported by Nuno
2208 Lopes.
2209
2210 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2211
2212 (1) The include/exclude patterns were being applied to the whole pathnames
2213 of files, instead of just to the final components.
2214
2215 (2) If there was more than one level of directory, the subdirectories were
2216 skipped unless they satisfied the include/exclude conditions. This is
2217 inconsistent with GNU grep (and could even be seen as contrary to the
2218 pcregrep specification - which I improved to make it absolutely clear).
2219 The action now is always to scan all levels of directory, and just
2220 apply the include/exclude patterns to regular files.
2221
2222 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2223 --exclude_dir in the tests to avoid scanning .svn directories.
2224
2225 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2226 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2227 doesn't support NULs in patterns.
2228
2229 7. Added some missing "const"s to declarations of static tables in
2230 pcre_compile.c and pcre_dfa_exec.c.
2231
2232 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2233 caused by fix #2 above. (Subsequently also a second patch to fix the
2234 first patch. And a third patch - this was a messy problem.)
2235
2236 9. Applied Craig's patch to remove the use of push_back().
2237
2238 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2239 matching function regexec().
2240
2241 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2242 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2243 references. PCRE supports relative numbers with this syntax (I don't think
2244 Oniguruma does).
2245
2246 12. Previously, a group with a zero repeat such as (...){0} was completely
2247 omitted from the compiled regex. However, this means that if the group
2248 was called as a subroutine from elsewhere in the pattern, things went wrong
2249 (an internal error was given). Such groups are now left in the compiled
2250 pattern, with a new opcode that causes them to be skipped at execution
2251 time.
2252
2253 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2254 to the way PCRE behaves:
2255
2256 (a) A lone ] character is dis-allowed (Perl treats it as data).
2257
2258 (b) A back reference to an unmatched subpattern matches an empty string
2259 (Perl fails the current match path).
2260
2261 (c) A data ] in a character class must be notated as \] because if the
2262 first data character in a class is ], it defines an empty class. (In
2263 Perl it is not possible to have an empty class.) The empty class []
2264 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2265 The negative empty class [^] matches any one character, independently
2266 of the DOTALL setting.
2267
2268 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2269 non-existent subpattern following a character class starting with ']' and
2270 containing () gave an internal compiling error instead of "reference to
2271 non-existent subpattern". Fortunately, when the pattern did exist, the
2272 compiled code was correct. (When scanning forwards to check for the
2273 existence of the subpattern, it was treating the data ']' as terminating
2274 the class, so got the count wrong. When actually compiling, the reference
2275 was subsequently set up correctly.)
2276
2277 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2278 it was being rejected as not supported by pcre_dfa_exec(), even though
2279 other assertions are supported. I have made pcre_dfa_exec() support
2280 (*FAIL).
2281
2282 16. The implementation of 13c above involved the invention of a new opcode,
2283 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2284 cannot be changed at match time, I realized I could make a small
2285 improvement to matching performance by compiling OP_ALLANY instead of
2286 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2287 on the OP_ANY path.
2288
2289 17. Compiling pcretest on Windows with readline support failed without the
2290 following two fixes: (1) Make the unistd.h include conditional on
2291 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2292
2293 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2294 ncurses library to be included for pcretest when ReadLine support is
2295 requested, but also to allow for it to be overridden. This patch came from
2296 Daniel Bergström.
2297
2298 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2299 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2300 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2301 spotting this.
2302
2303
2304 Version 7.6 28-Jan-08
2305 ---------------------
2306
2307 1. A character class containing a very large number of characters with
2308 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2309 overflow.
2310
2311 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2312 HAVE_LONG_LONG is not defined.
2313
2314 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2315 bring them up to date and include new features. This patch includes:
2316
2317 - Fixed PH's badly added libz and libbz2 support.
2318 - Fixed a problem with static linking.
2319 - Added pcredemo. [But later removed - see 7 below.]
2320 - Fixed dftables problem and added an option.
2321 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2322 HAVE_LONG_LONG.
2323 - Added readline support for pcretest.
2324 - Added an listing of the option settings after cmake has run.
2325
2326 4. A user submitted a patch to Makefile that makes it easy to create
2327 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2328 Makefile.am that cause it to include this special target, without
2329 affecting anything else. Note that the same mingw target plus all
2330 the other distribution libraries and programs are now supported
2331 when configuring with CMake (see 6 below) instead of with
2332 Configure/Make.
2333
2334 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2335 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2336 exported in the Windows port". It has not yet been confirmed that the patch
2337 solves the problem, but it does no harm.
2338
2339 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2340 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2341 with CMake, and also correct the comment about stack recursion.
2342
2343 7. Remove the automatic building of pcredemo from the ./configure system and
2344 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2345 of a program that users should build themselves after PCRE is installed, so
2346 building it automatically is not really right. What is more, it gave
2347 trouble in some build environments.
2348
2349 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2350
2351
2352 Version 7.5 10-Jan-08
2353 ---------------------
2354
2355 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2356 values in parens when parsing an RE using the C++ wrapper."
2357
2358 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2359 Characters greater than 255 were excluded from the class instead of being
2360 included.
2361
2362 3. The same bug as (2) above applied to negated POSIX classes such as
2363 [:^space:].
2364
2365 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2366 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2367 I have changed it.
2368
2369 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2370 first named subpattern) and a construct such as (?&a) would reference the
2371 first named subpattern whose name started with "a" (in other words, the
2372 length check was missing). Both these problems are fixed. "Subpattern name
2373 expected" is now given for (?&) (a zero-length name), and this patch also
2374 makes it give the same error for \k'' (previously it complained that that
2375 was a reference to a non-existent subpattern).
2376
2377 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2378 this is right because (?- can be followed by option settings as well as by
2379 digits. I have, however, made the messages clearer.
2380
2381 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2382 than the number used in the conditional) now cause a compile-time error.
2383 This is actually not compatible with Perl, which accepts such patterns, but
2384 treats the conditional as always being FALSE (as PCRE used to), but it
2385 seems to me that giving a diagnostic is better.
2386
2387 8. Change "alphameric" to the more common word "alphanumeric" in comments
2388 and messages.
2389
2390 9. Fix two occurrences of "backslash" in comments that should have been
2391 "backspace".
2392
2393 10. Remove two redundant lines of code that can never be obeyed (their function
2394 was moved elsewhere).
2395
2396 11. The program that makes PCRE's Unicode character property table had a bug
2397 which caused it to generate incorrect table entries for sequences of
2398 characters that have the same character type, but are in different scripts.
2399 It amalgamated them into a single range, with the script of the first of
2400 them. In other words, some characters were in the wrong script. There were
2401 thirteen such cases, affecting characters in the following ranges:
2402
2403 U+002b0 - U+002c1
2404 U+0060c - U+0060d
2405 U+0061e - U+00612
2406 U+0064b - U+0065e
2407 U+0074d - U+0076d
2408 U+01800 - U+01805
2409 U+01d00 - U+01d77
2410 U+01d9b - U+01dbf
2411 U+0200b - U+0200f
2412 U+030fc - U+030fe
2413 U+03260 - U+0327f
2414 U+0fb46 - U+0fbb1
2415 U+10450 - U+1049d
2416
2417 12. The -o option (show only the matching part of a line) for pcregrep was not
2418 compatible with GNU grep in that, if there was more than one match in a
2419 line, it showed only the first of them. It now behaves in the same way as
2420 GNU grep.
2421
2422 13. If the -o and -v options were combined for pcregrep, it printed a blank
2423 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2424 does the same. The return code can be used to tell if there were any
2425 non-matching lines.
2426
2427 14. Added --file-offsets and --line-offsets to pcregrep.
2428
2429 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2430 infinitely looping recursion. The bug was that positive lookaheads were not
2431 being skipped when checking for a possible empty match (negative lookaheads
2432 and both kinds of lookbehind were skipped).
2433
2434 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2435 inclusion of <windows.h> to before rather than after the definition of
2436 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2437
2438 17. Specifying a possessive quantifier with a specific limit for a Unicode
2439 character property caused pcre_compile() to compile bad code, which led at
2440 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2441 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2442 caused the error; without that there was no problem.
2443
2444 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2445
2446 19. Added --enable-pcretest-libreadline.
2447
2448 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2449 RE::GlobalReplace(). As a result, the number of replacements returned was
2450 double what it should be. I removed one of the increments, but Craig sent a
2451 later patch that removed the other one (the right fix) and added unit tests
2452 that check the return values (which was not done before).
2453
2454 21. Several CMake things:
2455
2456 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2457 the names libpcre and libpcreposix, not just pcre and pcreposix.
2458
2459 (2) The above change means that pcretest and pcregrep are now correctly
2460 linked with the newly-built libraries, not previously installed ones.
2461
2462 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2463
2464 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2465 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2466 UTF-8 newline character). The key issue is that the pattern starts .*;
2467 this means that the match must be either at the beginning, or after a
2468 newline. The bug was in the code for advancing after a failed match and
2469 checking that the new position followed a newline. It was not taking
2470 account of UTF-8 characters correctly.
2471
2472 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2473 character classes. PCRE was not treating the sequence [:...:] as a
2474 character class unless the ... were all letters. Perl, however, seems to
2475 allow any characters between [: and :], though of course it rejects as
2476 unknown any "names" that contain non-letters, because all the known class
2477 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2478 for example, whereas PCRE did not - it did not recognize a POSIX character
2479 class. This seemed a bit dangerous, so the code has been changed to be
2480 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2481 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2482 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2483 Perl does, and where it didn't before.
2484
2485 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2486 Windows environments %n is disabled by default.
2487
2488
2489 Version 7.4 21-Sep-07
2490 ---------------------
2491
2492 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2493 means that a class such as [\s] counted as "explicit reference to CR or
2494 LF". That isn't really right - the whole point of the change was to try to
2495 help when there was an actual mention of one of the two characters. So now
2496 the change happens only if \r or \n (or a literal CR or LF) character is
2497 encountered.
2498
2499 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2500 of both had grown to the point where there were only 3 bits left.
2501 Fortunately, there was spare space in the data structure, and so I have
2502 moved the internal flags into a new 16-bit field to free up more option
2503 bits.
2504
2505 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2506 but did not set the internal JCHANGED flag - either of these is enough to
2507 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2508 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2509 start sets both bits.
2510
2511 4. Added options (at build time, compile time, exec time) to change \R from
2512 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2513
2514 5. doc/pcresyntax.html was missing from the distribution.
2515
2516 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2517 compatibility, even though it is no longer used.
2518
2519 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2520 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2521 windows.h file is present (where different names are used). [This was
2522 reversed later after testing - see 16 below.]
2523
2524 8. Changed all #include <config.h> to #include "config.h". There were also
2525 some further <pcre.h> cases that I changed to "pcre.h".
2526
2527 9. When pcregrep was used with the --colour option, it missed the line ending
2528 sequence off the lines that it output.
2529
2530 10. It was pointed out to me that arrays of string pointers cause lots of
2531 relocations when a shared library is dynamically loaded. A technique of
2532 using a single long string with a table of offsets can drastically reduce
2533 these. I have refactored PCRE in four places to do this. The result is
2534 dramatic:
2535
2536 Originally: 290
2537 After changing UCP table: 187
2538 After changing error message table: 43
2539 After changing table of "verbs" 36
2540 After changing table of Posix names 22
2541
2542 Thanks to the folks working on Gregex for glib for this insight.
2543
2544 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2545 unicode-properties was also set.
2546
2547 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2548
2549 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2550 checked only for CRLF.
2551
2552 14. Added casts to pcretest.c to avoid compiler warnings.
2553
2554 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2555
2556 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2557 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2558 entirely. This removes changes made in 7 above.
2559
2560 17. The CMake files have been updated, and there is now more information about
2561 building with CMake in the NON-UNIX-USE document.
2562
2563
2564 Version 7.3 28-Aug-07
2565 ---------------------
2566
2567 1. In the rejigging of the build system that eventually resulted in 7.1, the
2568 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2569 brackets there is not right, since it causes compilers to look for an
2570 installed pcre.h, not the version that is in the source that is being
2571 compiled (which of course may be different). I have changed it back to:
2572
2573 #include "pcre.h"
2574
2575 I have a vague recollection that the change was concerned with compiling in
2576 different directories, but in the new build system, that is taken care of
2577 by the VPATH setting the Makefile.
2578
2579 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2580 when the subject happened to end in the byte 0x85 (e.g. if the last
2581 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2582 characters but of course it shouldn't be taken as a newline when it is part
2583 of another character. The bug was that, for an unlimited repeat of . in
2584 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2585 characters when looking for a newline.
2586
2587 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2588
2589 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2590 in debug output.
2591
2592 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2593 long printing in the pcrecpp unittest when running under MinGW.
2594
2595 6. ESC_K was left out of the EBCDIC table.
2596
2597 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2598 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2599 limit also applies to "virtual nesting" when a pattern is recursive, and in
2600 this case 1000 isn't so big. I have been able to remove this limit at the
2601 expense of backing off one optimization in certain circumstances. Normally,
2602 when pcre_exec() would call its internal match() function recursively and
2603 immediately return the result unconditionally, it uses a "tail recursion"
2604 feature to save stack. However, when a subpattern that can match an empty
2605 string has an unlimited repetition quantifier, it no longer makes this
2606 optimization. That gives it a stack frame in which to save the data for
2607 checking that an empty string has been matched. Previously this was taken
2608 from the 1000-entry workspace that had been reserved. So now there is no
2609 explicit limit, but more stack is used.
2610
2611 8. Applied Daniel's patches to solve problems with the import/export magic
2612 syntax that is required for Windows, and which was going wrong for the
2613 pcreposix and pcrecpp parts of the library. These were overlooked when this
2614 problem was solved for the main library.
2615
2616 9. There were some crude static tests to avoid integer overflow when computing
2617 the size of patterns that contain repeated groups with explicit upper
2618 limits. As the maximum quantifier is 65535, the maximum group length was
2619 set at 30,000 so that the product of these two numbers did not overflow a
2620 32-bit integer. However, it turns out that people want to use groups that
2621 are longer than 30,000 bytes (though not repeat them that many times).
2622 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2623 made it possible to implement the integer overflow checks in a much more
2624 dynamic way, which I have now done. The artificial limitation on group
2625 length has been removed - we now have only the limit on the total length of
2626 the compiled pattern, which depends on the LINK_SIZE setting.
2627
2628 10. Fixed a bug in the documentation for get/copy named substring when
2629 duplicate names are permitted. If none of the named substrings are set, the
2630 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2631 empty string.
2632
2633 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2634 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2635 because the ] is interpreted as the first data character and the
2636 terminating ] is not found. PCRE has been made compatible with Perl in this
2637 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2638 cause memory overwriting.
2639
2640 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2641 string has been matched (to stop an infinite loop). It was not recognizing
2642 a conditional subpattern that could match an empty string if that
2643 subpattern was within another subpattern. For example, it looped when
2644 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2645 condition was not nested. This bug has been fixed.
2646
2647 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2648 past the start of the subject in the presence of bytes with the top bit
2649 set, for example "\x8aBCD".
2650
2651 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2652 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2653
2654 14. Optimized (?!) to (*FAIL).
2655
2656 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2657 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2658 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2659 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2660 does: it's just the validity check that is more restrictive.
2661
2662 16. Inserted checks for integer overflows during escape sequence (backslash)
2663 processing, and also fixed erroneous offset values for syntax errors during
2664 backslash processing.
2665
2666 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2667 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2668
2669 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2670 caused an overrun.
2671
2672 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2673 something other than just ASCII characters) inside a group that had an
2674 unlimited repeat caused a loop at compile time (while checking to see
2675 whether the group could match an empty string).
2676
2677 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2678 [\P{Any}] did so. (Error in the code for printing property names.)
2679
2680 21. An orphan \E inside a character class could cause a crash.
2681
2682 22. A repeated capturing bracket such as (A)? could cause a wild memory
2683 reference during compilation.
2684
2685 23. There are several functions in pcre_compile() that scan along a compiled
2686 expression for various reasons (e.g. to see if it's fixed length for look
2687 behind). There were bugs in these functions when a repeated \p or \P was
2688 present in the pattern. These operators have additional parameters compared
2689 with \d, etc, and these were not being taken into account when moving along
2690 the compiled data. Specifically:
2691
2692 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2693 length.
2694
2695 (b) An item such as \pL+ within a repeated group could cause crashes or
2696 loops.
2697
2698 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2699 "reference to non-existent subpattern" error.
2700
2701 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2702
2703 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2704 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2705
2706 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2707
2708 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2709 character were causing crashes (broken optimization).
2710
2711 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2712 \p or \P) caused a compile-time loop.
2713
2714 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2715 break. For example, the unstudied pattern [\r\n]A does not match the string
2716 "\r\nA" because change 7.0/46 below moves the current point on by two
2717 characters after failing to match at the start. However, the pattern \nA
2718 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2719 the same is true. There doesn't seem any very clean way out of this, but
2720 what I have chosen to do makes the common cases work: PCRE now takes note
2721 of whether there can be an explicit match for \r or \n anywhere in the
2722 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2723 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2724 pattern has explicit CR or LF references.
2725
2726 29. Added (*CR) etc for changing newline setting at start of pattern.
2727
2728
2729 Version 7.2 19-Jun-07
2730 ---------------------
2731
2732 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2733 which is apparently normally available under Windows.
2734
2735 2. Re-jig the pcregrep tests with different newline settings in an attempt
2736 to make them independent of the local environment's newline setting.
2737
2738 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2739
2740 4. Some of the "internals" tests were previously cut out when the link size
2741 was not 2, because the output contained actual offsets. The recent new
2742 "Z" feature of pcretest means that these can be cut out, making the tests
2743 usable with all link sizes.
2744
2745 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2746 stack recursion. This gives a massive performance boost under BSD, but just
2747 a small improvement under Linux. However, it saves one field in the frame
2748 in all cases.
2749
2750 6. Added more features from the forthcoming Perl 5.10:
2751
2752 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2753 recursion call. It refers to the nth most recently opened parentheses.
2754
2755 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2756 to be opened parentheses.
2757
2758 (c) Conditions that refer to capturing parentheses can be specified
2759 relatively, for example, (?(-2)... or (?(+3)...
2760
2761 (d) \K resets the start of the current match so that everything before
2762 is not part of it.
2763
2764 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2765
2766 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2767 reference syntax.
2768
2769 (g) (?| introduces a group in which the numbering of parentheses in each
2770 alternative starts with the same number.
2771
2772 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2773
2774 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2775 PCRE_INFO_JCHANGED.
2776
2777 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2778 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2779 for detecting groups that can match an empty string.
2780
2781 9. A pattern with a very large number of alternatives (more than several
2782 hundred) was running out of internal workspace during the pre-compile
2783 phase, where pcre_compile() figures out how much memory will be needed. A
2784 bit of new cunning has reduced the workspace needed for groups with
2785 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2786 workspace instead of running out of the 4096 that are available.
2787
2788 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2789
2790 11. Applied patch from Google to remove an optimization that didn't quite work.
2791 The report of the bug said:
2792
2793 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2794 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2795 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2796
2797 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2798 it matched the wrong number of bytes.
2799
2800
2801 Version 7.1 24-Apr-07
2802 ---------------------
2803
2804 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2805 that is more "standard", making use of automake and other Autotools. There
2806 is some re-arrangement of the files and adjustment of comments consequent
2807 on this.
2808
2809 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2810 for recursive directory scanning broke on some systems because the files
2811 are not scanned in any specific order and on different systems the order
2812 was different. A call to "sort" has been inserted into RunGrepTest for the
2813 approprate test as a short-term fix. In the longer term there may be an
2814 alternative.
2815
2816 3. I had an email from Eric Raymond about problems translating some of PCRE's
2817 man pages to HTML (despite the fact that I distribute HTML pages, some
2818 people do their own conversions for various reasons). The problems
2819 concerned the use of low-level troff macros .br and .in. I have therefore
2820 removed all such uses from the man pages (some were redundant, some could
2821 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2822 HTML has been updated to handle .nf/.fi and to complain if it encounters
2823 .br or .in.
2824
2825 4. Updated comments in configure.ac that get placed in config.h.in and also
2826 arranged for config.h to be included in the distribution, with the name
2827 config.h.generic, for the benefit of those who have to compile without
2828 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2829
2830 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2831 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2832 makevp.bat for the current PCRE, using the additional files
2833 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2834
2835 6. A Windows user reported a minor discrepancy with test 2, which turned out
2836 to be caused by a trailing space on an input line that had got lost in his
2837 copy. The trailing space was an accident, so I've just removed it.
2838
2839 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2840 that is needed.
2841
2842 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2843 as "const" (a) because they are and (b) because it helps the PHP
2844 maintainers who have recently made a script to detect big data structures
2845 in the php code that should be moved to the .rodata section. I remembered
2846 to update Builducptable as well, so it won't revert if ucptable.h is ever
2847 re-created.
2848
2849 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2850 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2851 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2852 support is not required. This saves 1.5-2K of code, which is important in
2853 some applications.
2854
2855 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2856 so as not to refer to the tables, even though these functions will never be
2857 called when UTF-8 support is disabled. Otherwise there are problems with a
2858 shared library.
2859
2860 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2861
2862 (a) It was defining its arguments as char * instead of void *.
2863
2864 (b) It was assuming that all moves were upwards in memory; this was true
2865 a long time ago when I wrote it, but is no longer the case.
2866
2867 The emulated memove() is provided for those environments that have neither
2868 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2869 is clearly not the case, as these two bugs were recently reported.
2870
2871 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2872 and Detrail to create the HTML documentation, the .txt form of the man
2873 pages, and it removes trailing spaces from listed files. It also creates
2874 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2875 case, it wraps all the #defines with #ifndefs. This script should be run
2876 before "make dist".
2877
2878 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2879 with Unicode property support.
2880
2881 (a) For a maximizing quantifier, if the two different cases of the
2882 character were of different lengths in their UTF-8 codings (there are
2883 some cases like this - I found 11), and the matching function had to
2884 back up over a mixture of the two cases, it incorrectly assumed they
2885 were both the same length.
2886
2887 (b) When PCRE was configured to use the heap rather than the stack for
2888 recursion during matching, it was not correctly preserving the data for
2889 the other case of a UTF-8 character when checking ahead for a match
2890 while processing a minimizing repeat. If the check also involved
2891 matching a wide character, but failed, corruption could cause an
2892 erroneous result when trying to check for a repeat of the original
2893 character.
2894
2895 13. Some tidying changes to the testing mechanism:
2896
2897 (a) The RunTest script now detects the internal link size and whether there
2898 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2899 values substituted by "configure". (The RunGrepTest script already did
2900 this for UTF-8.) The configure.ac script no longer substitutes the
2901 relevant variables.
2902
2903 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2904 with length and offset values. This means that the output is different
2905 for different internal link sizes. Test 2 is skipped for link sizes
2906 other than 2 because of this, bypassing the problem. Unfortunately,
2907 there was also a test in test 3 (the locale tests) that used /B and
2908 failed for link sizes other than 2. Rather than cut the whole test out,
2909 I have added a new /Z option to pcretest that replaces the length and
2910 offset values with spaces. This is now used to make test 3 independent
2911 of link size. (Test 2 will be tidied up later.)
2912
2913 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2914 segmentation fault instead of returning the appropriate error message.
2915
2916 15. In multiline mode when the newline sequence was set to "any", the pattern
2917 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2918 This doesn't seem right; it now treats the CRLF combination as the line
2919 ending, and so does not match in that case. It's only a pattern such as ^$
2920 that would hit this one: something like ^ABC$ would have failed after \r
2921 and then tried again after \r\n.
2922
2923 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2924 in an attempt to make files that differ only in their line terminators
2925 compare equal. This works on Linux.
2926
2927 17. Under certain error circumstances pcregrep might try to free random memory
2928 as it exited. This is now fixed, thanks to valgrind.
2929
2930 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2931 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2932 was because its rules for how to advance for /g after matching an empty
2933 string at the end of a line did not allow for this case. They now check for
2934 it specially.
2935
2936 20. pcretest is supposed to handle patterns and data of any length, by
2937 extending its buffers when necessary. It was getting this wrong when the
2938 buffer for a data line had to be extended.
2939
2940 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2941 CRLF as a newline sequence.
2942
2943 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2944 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2945 I have nevertheless tidied it up.
2946
2947 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2948
2949 24. Added a man page for pcre-config.
2950
2951
2952 Version 7.0 19-Dec-06
2953 ---------------------
2954
2955 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2956 moving to gcc 4.1.1.
2957
2958 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2959 sys/time.h, which is documented as needed for this function. It doesn't
2960 seem to matter on Linux, but it showed up on some releases of OS X.
2961
2962 3. It seems that there are systems where bytes whose values are greater than
2963 127 match isprint() in the "C" locale. The "C" locale should be the
2964 default when a C program starts up. In most systems, only ASCII printing
2965 characters match isprint(). This difference caused the output from pcretest
2966 to vary, making some of the tests fail. I have changed pcretest so that:
2967
2968 (a) When it is outputting text in the compiled version of a pattern, bytes
2969 other than 32-126 are always shown as hex escapes.
2970
2971 (b) When it is outputting text that is a matched part of a subject string,
2972 it does the same, unless a different locale has been set for the match
2973 (using the /L modifier). In this case, it uses isprint() to decide.
2974
2975 4. Fixed a major bug that caused incorrect computation of the amount of memory
2976 required for a compiled pattern when options that changed within the
2977 pattern affected the logic of the preliminary scan that determines the
2978 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2979 that the computed length was too small. The symptoms of this bug were
2980 either the PCRE error "internal error: code overflow" from pcre_compile(),
2981 or a glibc crash with a message such as "pcretest: free(): invalid next
2982 size (fast)". Examples of patterns that provoked this bug (shown in
2983 pcretest format) are:
2984
2985 /(?-x: )/x
2986 /(?x)(?-x: \s*#\s*)/
2987 /((?i)[\x{c0}])/8
2988 /(?i:[\x{c0}])/8
2989
2990 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2991 is now done differently.
2992
2993 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2994 wrapper classes; (b) implement a new function in the C++ scanner that is
2995 more efficient than the old way of doing things because it avoids levels of
2996 recursion in the regex matching; (c) add a paragraph to the documentation
2997 for the FullMatch() function.
2998
2999 6. The escape sequence \n was being treated as whatever was defined as
3000 "newline". Not only was this contrary to the documentation, which states
3001 that \n is character 10 (hex 0A), but it also went horribly wrong when
3002 "newline" was defined as CRLF. This has been fixed.
3003
3004 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3005 was being set to -1 for the "end of line" case (supposedly a value that no
3006 character can have). Though this value is never used (the check for end of
3007 line is "zero bytes in current character"), it caused compiler complaints.
3008 I've changed it to 0xffffffff.
3009
3010 8. In pcre_version.c, the version string was being built by a sequence of
3011 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3012 string (as it is for production releases) called a macro with an empty
3013 argument. The C standard says the result of this is undefined. The gcc
3014 compiler treats it as an empty string (which was what was wanted) but it is
3015 reported that Visual C gives an error. The source has been hacked around to
3016 avoid this problem.
3017
3018 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3019 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3020 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3021 of them did).
3022
3023 10. Originally, pcretest opened its input and output without "b"; then I was
3024 told that "b" was needed in some environments, so it was added for release
3025 5.0 to both the input and output. (It makes no difference on Unix-like
3026 systems.) Later I was told that it is wrong for the input on Windows. I've
3027 now abstracted the modes into two macros, to make it easier to fiddle with
3028 them, and removed "b" from the input mode under Windows.
3029
3030 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3031
3032 12. Added -help and --help to pcretest as an official way of being reminded
3033 of the options.
3034
3035 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3036 and pcrecpp.cc because they annoy compilers at high warning levels.
3037
3038 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3039
3040 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3041 scripts are not C programs :-) and which was not noticed because it works
3042 on Linux.
3043
3044 16. pcretest is supposed to handle any length of pattern and data line (as one
3045 line or as a continued sequence of lines) by extending its input buffer if
3046 necessary. This feature was broken for very long pattern lines, leading to
3047 a string of junk being passed to pcre_compile() if the pattern was longer
3048 than about 50K.
3049
3050 17. I have done a major re-factoring of the way pcre_compile() computes the
3051 amount of memory needed for a compiled pattern. Previously, there was code
3052 that made a preliminary scan of the pattern in order to do this. That was
3053 OK when PCRE was new, but as the facilities have expanded, it has become
3054 harder and harder to keep it in step with the real compile phase, and there
3055 have been a number of bugs (see for example, 4 above). I have now found a
3056 cunning way of running the real compile function in a "fake" mode that
3057 enables it to compute how much memory it would need, while actually only
3058 ever using a few hundred bytes of working memory and without too many
3059 tests of the mode. This should make future maintenance and development
3060 easier. A side effect of this work is that the limit of 200 on the nesting
3061 depth of parentheses has been removed (though this was never a serious
3062 limitation, I suspect). However, there is a downside: pcre_compile() now
3063 runs more slowly than before (30% or more, depending on the pattern). I
3064 hope this isn't a big issue. There is no effect on runtime performance.
3065
3066 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3067 newline (only possible for the last line of a file) and it was a
3068 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3069
3070 19. Added additional timing features to pcretest. (1) The -tm option now times
3071 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3072 separate command line item, by a number that specifies the number of
3073 repeats to use when timing. The default is 50000; this gives better
3074 precision, but takes uncomfortably long for very large patterns.
3075
3076 20. Extended pcre_study() to be more clever in cases where a branch of a
3077 subpattern has no definite first character. For example, (a*|b*)[cd] would
3078 previously give no result from pcre_study(). Now it recognizes that the
3079 first character must be a, b, c, or d.
3080
3081 21. There was an incorrect error "recursive call could loop indefinitely" if
3082 a subpattern (or the entire pattern) that was being tested for matching an
3083 empty string contained only one non-empty item after a nested subpattern.
3084 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3085 incorrectly, because the \d was being skipped in the check.
3086
3087 22. The pcretest program now has a new pattern option /B and a command line
3088 option -b, which is equivalent to adding /B to every pattern. This causes
3089 it to show the compiled bytecode, without the additional information that
3090 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3091 is the same as /B/I).
3092
3093 23. A new optimization is now able automatically to treat some sequences such
3094 as a*b as a*+b. More specifically, if something simple (such as a character
3095 or a simple class like \d) has an unlimited quantifier, and is followed by
3096 something that cannot possibly match the quantified thing, the quantifier
3097 is automatically "possessified".
3098
3099 24. A recursive reference to a subpattern whose number was greater than 39
3100 went wrong under certain circumstances in UTF-8 mode. This bug could also
3101 have affected the operation of pcre_study().
3102
3103 25. Realized that a little bit of performance could be had by replacing
3104 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3105
3106 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3107
3108 27. Possessive quantifiers such as a++ were previously implemented by turning
3109 them into atomic groups such as ($>a+). Now they have their own opcodes,
3110 which improves performance. This includes the automatically created ones
3111 from 23 above.
3112
3113 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3114 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3115 the first matched character to be a colon. This applied both to named and
3116 numbered groups.
3117
3118 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3119
3120 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3121 building PCRE on Windows easier, so I have included it in the distribution.
3122
3123 31. There is now a check in pcretest against a ridiculously large number being
3124 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3125 loop, the loop is abandoned.
3126
3127 32. Forward references to subpatterns in conditions such as (?(2)...) where
3128 subpattern 2 is defined later cause pcre_compile() to search forwards in
3129 the pattern for the relevant set of parentheses. This search went wrong
3130 when there were unescaped parentheses in a character class, parentheses
3131 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3132
3133 33. "Subroutine" calls and backreferences were previously restricted to
3134 referencing subpatterns earlier in the regex. This restriction has now
3135 been removed.
3136
3137 34. Added a number of extra features that are going to be in Perl 5.10. On the
3138 whole, these are just syntactic alternatives for features that PCRE had
3139 previously implemented using the Python syntax or my own invention. The
3140 other formats are all retained for compatibility.
3141
3142 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3143 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3144 also .NET compatible.
3145
3146 (b) A recursion or subroutine call to a named group can now be defined as
3147 (?&name) as well as (?P>name).
3148
3149 (c) A backreference to a named group can now be defined as \k<name> or
3150 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3151 5.10, are also .NET compatible.
3152
3153 (d) A conditional reference to a named group can now use the syntax
3154 (?(<name>) or (?('name') as well as (?(name).
3155
3156 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3157 groups (named and numbered) that are never evaluated inline, but can be
3158 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3159 is always false. There may be only one alternative in such a group.
3160
3161 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3162 as the simple (?(R). The condition is true only if the most recent
3163 recursion is that of the given number or name. It does not search out
3164 through the entire recursion stack.
3165
3166 (g) The escape \gN or \g{N} has been added, where N is a positive or
3167 negative number, specifying an absolute or relative reference.
3168
3169 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3170 some "unreachable code" warnings.
3171
3172 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3173 things, this adds five new scripts.
3174
3175 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3176 There were also incompatibilities regarding the handling of \Q..\E inside
3177 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3178 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3179
3180 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3181 matches an empty string, and forcibly breaks the loop. There were bugs in
3182 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3183 against aaaa the result was just "a" rather than "aaaa", for example. Two
3184 separate and independent bugs (that affected different cases) have been
3185 fixed.
3186
3187 39. Refactored the code to abolish the use of different opcodes for small
3188 capturing bracket numbers. This is a tidy that I avoided doing when I
3189 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3190 The new approach is not only tidier, it makes it possible to reduce the
3191 memory needed to fix the previous bug (38).
3192
3193 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3194 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3195 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3196 mode.
3197
3198 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3199 report.
3200
3201 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3202 copy construction and assignment in the C++ wrapper.
3203
3204 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3205 couple of bugs that could have given wrong results in the "--newline=crlf"
3206 case.
3207
3208 44. Added a number of casts and did some reorganization of signed/unsigned int
3209 variables following suggestions from Dair Grant. Also renamed the variable
3210 "this" as "item" because it is a C++ keyword.
3211
3212 45. Arranged for dftables to add
3213
3214 #include "pcre_internal.h"
3215
3216 to pcre_chartables.c because without it, gcc 4.x may remove the array
3217 definition from the final binary if PCRE is built into a static library and
3218 dead code stripping is activated.
3219
3220 46. For an unanchored pattern, if a match attempt fails at the start of a
3221 newline sequence, and the newline setting is CRLF or ANY, and the next two
3222 characters are CRLF, advance by two characters instead of one.
3223
3224
3225 Version 6.7 04-Jul-06
3226 ---------------------
3227
3228 1. In order to handle tests when input lines are enormously long, pcretest has
3229 been re-factored so that it automatically extends its buffers when
3230 necessary. The code is crude, but this _is_ just a test program. The
3231 default size has been increased from 32K to 50K.
3232
3233 2. The code in pcre_study() was using the value of the re argument before
3234 testing it for NULL. (Of course, in any sensible call of the function, it
3235 won't be NULL.)
3236
3237 3. The memmove() emulation function in pcre_internal.h, which is used on
3238 systems that lack both memmove() and bcopy() - that is, hardly ever -
3239 was missing a "static" storage class specifier.
3240
3241 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3242 containing an extended class (one that cannot be represented by a bitmap
3243 because it contains high-valued characters or Unicode property items, e.g.
3244 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3245 pattern, but PCRE should not loop if you do not (it no longer does).
3246 [Detail: two cases were found: (a) a repeated subpattern containing an
3247 extended class; (b) a recursive reference to a subpattern that followed a
3248 previous extended class. It wasn't skipping over the extended class
3249 correctly when UTF-8 mode was not set.]
3250
3251 5. A negated single-character class was not being recognized as fixed-length
3252 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3253 compile error "lookbehind assertion is not fixed length".
3254
3255 6. The RunPerlTest auxiliary script was showing an unexpected difference
3256 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3257 write a Perl script that can interpret lines of an input file either as
3258 byte characters or as UTF-8, which is what "perltest" was being required to
3259 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3260 can't do is switch easily at run time between having the "use utf8;" pragma
3261 or not. In the end, I fudged it by using the RunPerlTest script to insert
3262 "use utf8;" explicitly for the UTF-8 tests.
3263
3264 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3265 the end of the subject string, contrary to the documentation and to what
3266 Perl does. This was true of both matching functions. Now it matches only at
3267 the start of the subject and immediately after *internal* newlines.
3268
3269 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3270 a pointer to an int instead of a pointer to an unsigned long int. This
3271 caused problems on 64-bit systems.
3272
3273 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3274 instance of the 'standard' template library not being so standard".
3275
3276 10. There was no check on the number of named subpatterns nor the maximum
3277 length of a subpattern name. The product of these values is used to compute
3278 the size of the memory block for a compiled pattern. By supplying a very
3279 long subpattern name and a large number of named subpatterns, the size
3280 computation could be caused to overflow. This is now prevented by limiting
3281 the length of names to 32 characters, and the number of named subpatterns
3282 to 10,000.
3283
3284 11. Subpatterns that are repeated with specific counts have to be replicated in
3285 the compiled pattern. The size of memory for this was computed from the
3286 length of the subpattern and the repeat count. The latter is limited to
3287 65535, but there was no limit on the former, meaning that integer overflow
3288 could in principle occur. The compiled length of a repeated subpattern is
3289 now limited to 30,000 bytes in order to prevent this.
3290
3291 12. Added the optional facility to have named substrings with the same name.
3292
3293 13. Added the ability to use a named substring as a condition, using the
3294 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3295 are numbers (not recommended). Forward references are permitted.
3296
3297 14. Added forward references in named backreferences (if you see what I mean).
3298
3299 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3300 pattern could run off the end of the subject. For example, the pattern
3301 "(?s)(.{1,5})"8 did this with the subject "ab".
3302
3303 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3304 PCRE_CASELESS was set when matching characters that were quantified with ?
3305 or *.
3306
3307 17. A character class other than a single negated character that had a minimum
3308 but no maximum quantifier - for example [ab]{6,} - was not handled
3309 correctly by pce_dfa_exec(). It would match only one character.
3310
3311 18. A valid (though odd) pattern that looked like a POSIX character
3312 class but used an invalid character after [ (for example [[,abc,]]) caused
3313 pcre_compile() to give the error "Failed: internal error: code overflow" or
3314 in some cases to crash with a glibc free() error. This could even happen if
3315 the pattern terminated after [[ but there just happened to be a sequence of
3316 letters, a binary zero, and a closing ] in the memory that followed.
3317
3318 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3319 over the years. Originally (before any Unicode support), just the bottom 8
3320 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3321 output from "man perlunicode" includes this:
3322
3323 The regular expression compiler produces polymorphic opcodes. That
3324 is, the pattern adapts to the data and automatically switches to
3325 the Unicode character scheme when presented with Unicode data--or
3326 instead uses a traditional byte scheme when presented with byte
3327 data.
3328
3329 Sadly, a wide octal escape does not cause a switch, and in a string with
3330 no other multibyte characters, these octal escapes are treated as before.
3331 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3332 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3333 Unicode string.
3334
3335 I have not perpetrated such confusion in PCRE. Up till now, it took just
3336 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3337 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3338 translate to the appropriate multibyte character.
3339
3340 29. Applied some refactoring to reduce the number of warnings from Microsoft
3341 and Borland compilers. This has included removing the fudge introduced
3342 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3343 a warning about an unused variable.
3344
3345 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3346 characters since release 4.0, because Perl (from release 5.004) does not.
3347 [Or at least, is documented not to: some releases seem to be in conflict
3348 with the documentation.] However, when a pattern was studied with
3349 pcre_study() and all its branches started with \s, PCRE still included VT
3350 as a possible starting character. Of course, this did no harm; it just
3351 caused an unnecessary match attempt.
3352
3353 22. Removed a now-redundant internal flag bit that recorded the fact that case
3354 dependency changed within the pattern. This was once needed for "required
3355 byte" processing, but is no longer used. This recovers a now-scarce options
3356 bit. Also moved the least significant internal flag bit to the most-
3357 significant bit of the word, which was not previously used (hangover from
3358 the days when it was an int rather than a uint) to free up another bit for
3359 the future.
3360
3361 23. Added support for CRLF line endings as well as CR and LF. As well as the
3362 default being selectable at build time, it can now be changed at runtime
3363 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3364 specify that it is scanning data with non-default line endings.
3365
3366 24. Changed the definition of CXXLINK to make it agree with the definition of
3367 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3368
3369 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3370 recursions. This makes a big different to stack usage for some patterns.
3371
3372 26. If a subpattern containing a named recursion or subroutine reference such
3373 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3374 the space required for the compiled pattern went wrong and gave too small a
3375 value. Depending on the environment, this could lead to "Failed: internal
3376 error: code overflow at offset 49" or "glibc detected double free or
3377 corruption" errors.
3378
3379 27. Applied patches from Google (a) to support the new newline modes and (b) to
3380 advance over multibyte UTF-8 characters in GlobalReplace.
3381
3382 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3383 difference for some implementation of PCRE in some Windows version.
3384
3385 29. Added some extra testing facilities to pcretest:
3386
3387 \q<number> in a data line sets the "match limit" value
3388 \Q<number> in a data line sets the "match recursion limt" value
3389 -S <number> sets the stack size, where <number> is in megabytes
3390
3391 The -S option isn't available for Windows.
3392
3393
3394 Version 6.6 06-Feb-06
3395 ---------------------
3396
3397 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3398 in pcreposix.h. I have copied the definition from pcre.h.
3399
3400 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3401 because pcre.h is no longer a built file.
3402
3403 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3404 not normally included in the compiled code.
3405
3406
3407 Version 6.5 01-Feb-06
3408 ---------------------
3409
3410 1. When using the partial match feature with pcre_dfa_exec(), it was not
3411 anchoring the second and subsequent partial matches at the new starting
3412 point. This could lead to incorrect results. For example, with the pattern
3413 /1234/, partially matching against "123" and then "a4" gave a match.
3414
3415 2. Changes to pcregrep:
3416
3417 (a) All non-match returns from pcre_exec() were being treated as failures
3418 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3419 error message is output. Some extra information is given for the
3420 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3421 probably the only errors that are likely to be caused by users (by
3422 specifying a regex that has nested indefinite repeats, for instance).
3423 If there are more than 20 of these errors, pcregrep is abandoned.
3424
3425 (b) A binary zero was treated as data while matching, but terminated the
3426 output line if it was written out. This has been fixed: binary zeroes
3427 are now no different to any other data bytes.
3428
3429 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3430 used to set a locale for matching. The --locale=xxxx long option has
3431 been added (no short equivalent) to specify a locale explicitly on the
3432 pcregrep command, overriding the environment variables.
3433
3434 (d) When -B was used with -n, some line numbers in the output were one less
3435 than they should have been.
3436
3437 (e) Added the -o (--only-matching) option.
3438
3439 (f) If -A or -C was used with -c (count only), some lines of context were
3440 accidentally printed for the final match.
3441
3442 (g) Added the -H (--with-filename) option.
3443
3444 (h) The combination of options -rh failed to suppress file names for files
3445 that were found from directory arguments.
3446
3447 (i) Added the -D (--devices) and -d (--directories) options.
3448
3449 (j) Added the -F (--fixed-strings) option.
3450
3451 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3452
3453 (l) Added the --colo(u)r option.
3454
3455 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3456 is not present by default.
3457
3458 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3459 items such as (?R) or (?1), when the recursion could match a number of
3460 alternatives. If it matched one of the alternatives, but subsequently,
3461 outside the recursion, there was a failure, the code tried to back up into
3462 the recursion. However, because of the way PCRE is implemented, this is not
3463 possible, and the result was an incorrect result from the match.
3464
3465 In order to prevent this happening, the specification of recursion has
3466 been changed so that all such subpatterns are automatically treated as
3467 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3468
3469 4. I had overlooked the fact that, in some locales, there are characters for
3470 which isalpha() is true but neither isupper() nor islower() are true. In
3471 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3472 and ordfeminine) are like this. This affected the treatment of \w and \W
3473 when they appeared in character classes, but not when they appeared outside
3474 a character class. The bit map for "word" characters is now created
3475 separately from the results of isalnum() instead of just taking it from the
3476 upper, lower, and digit maps. (Plus the underscore character, of course.)
3477
3478 5. The above bug also affected the handling of POSIX character classes such as
3479 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3480 permanent tables. Instead, the bit maps for such a class were previously
3481 created as the appropriate unions of the upper, lower, and digit bitmaps.
3482 Now they are created by subtraction from the [[:word:]] class, which has
3483 its own bitmap.
3484
3485 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3486 It is created by subtracting the vertical space characters (\x09, \x0a,
3487 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3488 subtraction was done in the overall bitmap for a character class, meaning
3489 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3490 be recognized. This bug has been fixed.
3491
3492 7. Patches from the folks at Google:
3493
3494 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3495 real life, but is still worth protecting against".
3496
3497 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3498 regular expressions".
3499
3500 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3501 have it.
3502
3503 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3504 "configure" and the latter not, in order to fix a problem somebody had
3505 with compiling the Arg class on HP-UX.
3506
3507 (e) Improve the error-handling of the C++ wrapper a little bit.
3508
3509 (f) New tests for checking recursion limiting.
3510
3511 8. The pcre_memmove() function, which is used only if the environment does not
3512 have a standard memmove() function (and is therefore rarely compiled),
3513 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3514 returning a result (though PCRE never actually uses the result).
3515
3516 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3517 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3518 returned instead of calling malloc() with an overflowing number that would
3519 most likely cause subsequent chaos.
3520
3521 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3522
3523 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3524 with this option is matched, the nmatch and pmatch options of regexec() are
3525 ignored.
3526
3527 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3528 provided in case anyone wants to the the POSIX interface with UTF-8
3529 strings.
3530
3531 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3532 C++ linking (needed for some HP-UX environments).
3533
3534 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3535 (unused parameter) and in the pcre_printint() function (omitted "default"
3536 switch label when the default is to do nothing).
3537
3538 15. Added some code to make it possible, when PCRE is compiled as a C++
3539 library, to replace subject pointers for pcre_exec() with a smart pointer
3540 class, thus making it possible to process discontinuous strings.
3541
3542 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3543 much the same function. They were added by different people who were trying
3544 to make PCRE easy to compile on non-Unix systems. It has been suggested
3545 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3546 for compiling on Windows systems. I have therefore replaced it with
3547 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3548 defaults to "extern" for C or "extern C" for C++, which works fine on
3549 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3550 SCOPE with something explicit in config.h. In addition:
3551
3552 (a) pcreposix.h still had just "extern" instead of either of these macros;
3553 I have replaced it with PCRE_DATA_SCOPE.
3554
3555 (b) Functions such as _pcre_xclass(), which are internal to the library,
3556 but external in the C sense, all had PCRE_EXPORT in their definitions.
3557 This is apparently wrong for the Windows case, so I have removed it.
3558 (It makes no difference on Unix-like systems.)
3559
3560 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3561 of recursive calls to match(). This is different to MATCH_LIMIT because
3562 that limits the total number of calls to match(), not all of which increase
3563 the depth of recursion. Limiting the recursion depth limits the amount of
3564 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3565 when PCRE is compiled, and changed at run time. A patch from Google adds
3566 this functionality to the C++ interface.
3567
3568 18. Changes to the handling of Unicode character properties:
3569
3570 (a) Updated the table to Unicode 4.1.0.
3571
3572 (b) Recognize characters that are not in the table as "Cn" (undefined).
3573
3574 (c) I revised the way the table is implemented to a much improved format
3575 which includes recognition of ranges. It now supports the ranges that
3576 are defined in UnicodeData.txt, and it also amalgamates other
3577 characters into ranges. This has reduced the number of entries in the
3578 table from around 16,000 to around 3,000, thus reducing its size
3579 considerably. I realized I did not need to use a tree structure after
3580 all - a binary chop search is just as efficient. Having reduced the
3581 number of entries, I extended their size from 6 bytes to 8 bytes to
3582 allow for more data.
3583
3584 (d) Added support for Unicode script names via properties such as \p{Han}.
3585
3586 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3587 matching that character.
3588
3589 20. When matching a repeated Unicode property with a minimum greater than zero,
3590 (for example \pL{2,}), PCRE could look past the end of the subject if it
3591 reached it while seeking the minimum number of characters. This could
3592 happen only if some of the characters were more than one byte long, because
3593 there is a check for at least the minimum number of bytes.
3594
3595 21. Refactored the implementation of \p and \P so as to be more general, to
3596 allow for more different types of property in future. This has changed the
3597 compiled form incompatibly. Anybody with saved compiled patterns that use
3598 \p or \P will have to recompile them.
3599
3600 22. Added "Any" and "L&" to the supported property types.
3601
3602 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3603 but give a compile time error if the value is greater than 0xff.
3604
3605 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3606 accidentally not being installed or uninstalled.
3607
3608 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3609 made were to insert the current release number. This seemed silly, because
3610 it made things harder for people building PCRE on systems that don't run
3611 "configure". I have turned pcre.h into a distributed file, no longer built
3612 by "configure", with the version identification directly included. There is
3613 no longer a pcre.h.in file.
3614
3615 However, this change necessitated a change to the pcre-config script as
3616 well. It is built from pcre-config.in, and one of the substitutions was the
3617 release number. I have updated configure.ac so that ./configure now finds
3618 the release number by grepping pcre.h.
3619
3620 26. Added the ability to run the tests under valgrind.
3621
3622
3623 Version 6.4 05-Sep-05
3624 ---------------------
3625
3626 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3627 "--" to be printed when multiple files were scanned, even when none of the
3628 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3629 consider it to be a bug, and have restored the previous behaviour.
3630
3631 2. A couple of code tidies to get rid of compiler warnings.
3632
3633 3. The pcretest program used to cheat by referring to symbols in the library
3634 whose names begin with _pcre_. These are internal symbols that are not
3635 really supposed to be visible externally, and in some environments it is
3636 possible to suppress them. The cheating is now confined to including
3637 certain files from the library's source, which is a bit cleaner.
3638
3639 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3640 file's purpose clearer.
3641
3642 5. Reorganized pcre_ucp_findchar().
3643
3644
3645 Version 6.3 15-Aug-05
3646 ---------------------
3647
3648 1. The file libpcre.pc.in did not have general read permission in the tarball.
3649
3650 2. There were some problems when building without C++ support:
3651
3652 (a) If C++ support was not built, "make install" and "make test" still
3653 tried to test it.
3654
3655 (b) There were problems when the value of CXX was explicitly set. Some
3656 changes have been made to try to fix these, and ...
3657
3658 (c) --disable-cpp can now be used to explicitly disable C++ support.
3659
3660 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3661 backslash in a target when C++ was disabled. This confuses some
3662 versions of "make", apparently. Using an intermediate variable solves
3663 this. (Same for CPP_LOBJ.)
3664
3665 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3666 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3667 necessary on certain architectures.
3668
3669 4. Added a setting of -export-symbols-regex to the link command to remove
3670 those symbols that are exported in the C sense, but actually are local
3671 within the library, and not documented. Their names all begin with
3672 "_pcre_". This is not a perfect job, because (a) we have to except some
3673 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3674 available (and never for static libraries). I have made a note to try to
3675 find a way round (a) in the future.
3676
3677
3678 Version 6.2 01-Aug-05
3679 ---------------------
3680
3681 1. There was no test for integer overflow of quantifier values. A construction
3682 such as {1111111111111111} would give undefined results. What is worse, if
3683 a minimum quantifier for a parenthesized subpattern overflowed and became
3684 negative, the calculation of the memory size went wrong. This could have
3685 led to memory overwriting.
3686
3687 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3688
3689 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3690 operating environments where this matters.
3691
3692 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3693 PCRE options from within the C++ wrapper.
3694
3695 5. Named capturing subpatterns were not being correctly counted when a pattern
3696 was compiled. This caused two problems: (a) If there were more than 100
3697 such subpatterns, the calculation of the memory needed for the whole
3698 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3699 back references of the form \12, where the number was greater than 9, were
3700 not recognized as back references, even though there were sufficient
3701 previous subpatterns.
3702
3703 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3704 versions of gcc, e.g. 2.95.4.
3705
3706
3707 Version 6.1 21-Jun-05
3708 ---------------------
3709
3710 1. There was one reference to the variable "posix" in pcretest.c that was not
3711 surrounded by "#if !defined NOPOSIX".
3712
3713 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3714 the cross-check on the old pcre_info() function, for the benefit of the
3715 cut-down version of PCRE that is currently imported into Exim.
3716
3717 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3718 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3719 patterns that start (?i) but I don't think that matters. The use of (?i) is
3720 just an example; this all applies to the other options as well.
3721
3722 4. Since libtool seems to echo the compile commands it is issuing, the output
3723 from "make" can be reduced a bit by putting "@" in front of each libtool
3724 compile command.
3725
3726 5. Patch from the folks at Google for configure.in to be a bit more thorough
3727 in checking for a suitable C++ installation before trying to compile the
3728 C++ stuff. This should fix a reported problem when a compiler was present,
3729 but no suitable headers.
3730
3731 6. The man pages all had just "PCRE" as their title. I have changed them to
3732 be the relevant file name. I have also arranged that these names are
3733 retained in the file doc/pcre.txt, which is a concatenation in text format
3734 of all the man pages except the little individual ones for each function.
3735
3736 7. The NON-UNIX-USE file had not been updated for the different set of source
3737 files that come with release 6. I also added a few comments about the C++
3738 wrapper.
3739
3740
3741 Version 6.0 07-Jun-05
3742 ---------------------
3743
3744 1. Some minor internal re-organization to help with my DFA experiments.
3745
3746 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3747 didn't matter for the library itself when fully configured, but did matter
3748 when compiling without UCP support, or within Exim, where the ucp files are
3749 not imported.
3750
3751 3. Refactoring of the library code to split up the various functions into
3752 different source modules. The addition of the new DFA matching code (see
3753 below) to a single monolithic source would have made it really too
3754 unwieldy, quite apart from causing all the code to be include in a
3755 statically linked application, when only some functions are used. This is
3756 relevant even without the DFA addition now that patterns can be compiled in
3757 one application and matched in another.
3758
3759 The downside of splitting up is that there have to be some external
3760 functions and data tables that are used internally in different modules of
3761 the library but which are not part of the API. These have all had their
3762 names changed to start with "_pcre_" so that they are unlikely to clash
3763 with other external names.
3764
3765 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3766 a different (DFA) algorithm. Although it is slower than the original
3767 function, it does have some advantages for certain types of matching
3768 problem.
3769
3770 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3771 including restarting after a partial match.
3772
3773 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3774 defined when compiling for Windows was sent to me. I have put it into the
3775 code, though I have no means of testing or verifying it.
3776
3777 7. Added the pcre_refcount() auxiliary function.
3778
3779 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3780 match before or at the first newline in the subject string. In pcretest,
3781 the /f option on a pattern can be used to set this.
3782
3783 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3784 would behave wrongly. This has been present in PCRE since release 4.0.
3785
3786 10. A number of changes to the pcregrep command:
3787
3788 (a) Refactored how -x works; insert ^(...)$ instead of setting
3789 PCRE_ANCHORED and checking the length, in preparation for adding
3790 something similar for -w.
3791
3792 (b) Added the -w (match as a word) option.
3793
3794 (c) Refactored the way lines are read and buffered so as to have more
3795 than one at a time available.
3796
3797 (d) Implemented a pcregrep test script.
3798
3799 (e) Added the -M (multiline match) option. This allows patterns to match
3800 over several lines of the subject. The buffering ensures that at least
3801 8K, or the rest of the document (whichever is the shorter) is available
3802 for matching (and similarly the previous 8K for lookbehind assertions).
3803
3804 (f) Changed the --help output so that it now says
3805
3806 -w, --word-regex(p)
3807
3808 instead of two lines, one with "regex" and the other with "regexp"
3809 because that confused at least one person since the short forms are the
3810 same. (This required a bit of code, as the output is generated
3811 automatically from a table. It wasn't just a text change.)
3812
3813 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3814 option but starts with a hyphen. Could be a pattern or a path name
3815 starting with a hyphen, for instance.
3816
3817 (h) "-" can be given as a file name to represent stdin.
3818
3819 (i) When file names are being printed, "(standard input)" is used for
3820 the standard input, for compatibility with GNU grep. Previously
3821 "<stdin>" was used.
3822
3823 (j) The option --label=xxx can be used to supply a name to be used for
3824 stdin when file names are being printed. There is no short form.
3825
3826 (k) Re-factored the options decoding logic because we are going to add
3827 two more options that take data. Such options can now be given in four
3828 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3829
3830 (l) Added the -A, -B, and -C options for requesting that lines of context
3831 around matches be printed.
3832
3833 (m) Added the -L option to print the names of files that do not contain
3834 any matching lines, that is, the complement of -l.
3835
3836 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3837 continue to scan other files.
3838
3839 (o) The -s option was incorrectly implemented. For compatibility with other
3840 greps, it now suppresses the error message for a non-existent or non-
3841 accessible file (but not the return code). There is a new option called
3842 -q that suppresses the output of matching lines, which was what -s was
3843 previously doing.
3844
3845 (p) Added --include and --exclude options to specify files for inclusion
3846 and exclusion when recursing.
3847
3848 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3849 Hopefully, it now does.
3850
3851 12. Missing cast in pcre_study().
3852
3853 13. Added an "uninstall" target to the makefile.
3854
3855 14. Replaced "extern" in the function prototypes in Makefile.in with
3856 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3857 world, but is set differently for Windows.
3858
3859 15. Added a second compiling function called pcre_compile2(). The only
3860 difference is that it has an extra argument, which is a pointer to an
3861 integer error code. When there is a compile-time failure, this is set
3862 non-zero, in addition to the error test pointer being set to point to an
3863 error message. The new argument may be NULL if no error number is required
3864 (but then you may as well call pcre_compile(), which is now just a
3865 wrapper). This facility is provided because some applications need a
3866 numeric error indication, but it has also enabled me to tidy up the way
3867 compile-time errors are handled in the POSIX wrapper.
3868
3869 16. Added VPATH=.libs to the makefile; this should help when building with one
3870 prefix path and installing with another. (Or so I'm told by someone who
3871 knows more about this stuff than I do.)
3872
3873 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3874 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3875 match everything, including newlines. This is not POSIX-compatible, but
3876 somebody wanted the feature. From pcretest it can be activated by using
3877 both the P and the s flags.
3878
3879 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3880
3881 19. libpcre.pc was being incorrectly installed as executable.
3882
3883 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3884 it now also looks for '\r' so that it will work unmodified on Windows.
3885
3886 21. Added Google's contributed C++ wrapper to the distribution.
3887
3888 22. Added some untidy missing memory free() calls in pcretest, to keep
3889 Electric Fence happy when testing.
3890
3891
3892
3893 Version 5.0 13-Sep-04
3894 ---------------------
3895
3896 1. Internal change: literal characters are no longer packed up into items
3897 containing multiple characters in a single byte-string. Each character
3898 is now matched using a separate opcode. However, there may be more than one
3899 byte in the character in UTF-8 mode.
3900
3901 2. The pcre_callout_block structure has two new fields: pattern_position and
3902 next_item_length. These contain the offset in the pattern to the next match
3903 item, and its length, respectively.
3904
3905 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3906 insertion of callouts before each pattern item. Added the /C option to
3907 pcretest to make use of this.
3908
3909 4. On the advice of a Windows user, the lines
3910
3911 #if defined(_WIN32) || defined(WIN32)
3912 _setmode( _fileno( stdout ), 0x8000 );
3913 #endif /* defined(_WIN32) || defined(WIN32) */
3914
3915 have been added to the source of pcretest. This apparently does useful
3916 magic in relation to line terminators.
3917
3918 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3919 for the benefit of those environments where the "b" makes a difference.
3920
3921 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3922 to know about it. I have put a hack into configure.in that adds in code
3923 to set GCC=yes if CC=icc. This seems to end up at a point in the
3924 generated configure script that is early enough to affect the setting of
3925 compiler options, which is what is needed, but I have no means of testing
3926 whether it really works. (The user who reported this had patched the
3927 generated configure script, which of course I cannot do.)
3928
3929 LATER: After change 22 below (new libtool files), the configure script
3930 seems to know about icc (and also ecc). Therefore, I have commented out
3931 this hack in configure.in.
3932
3933 7. Added support for pkg-config (2 patches were sent in).
3934
3935 8. Negated POSIX character classes that used a combination of internal tables
3936 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3937 [[:^ascii]]. Typically, they would match almost any characters. The other
3938 POSIX classes were not broken in this way.
3939
3940 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3941 to find the match, as PCRE was deluded into thinking that the match had to
3942 start at the start point or following a newline. The same bug applied to
3943 patterns with negative forward assertions or any backward assertions
3944 preceding ".*" at the start, unless the pattern required a fixed first
3945 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3946
3947 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3948 starting at the last subject character, bytes beyond the end of the subject
3949 string were read.
3950
3951 11. Renamed the variable "class" as "classbits" to make life easier for C++
3952 users. (Previously there was a macro definition, but it apparently wasn't
3953 enough.)
3954
3955 12. Added the new field "tables" to the extra data so that tables can be passed
3956 in at exec time, or the internal tables can be re-selected. This allows
3957 a compiled regex to be saved and re-used at a later time by a different
3958 program that might have everything at different addresses.
3959
3960 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3961 -R library as well as a -L library.
3962
3963 14. The debugging options of pcretest (-d on the command line or D on a
3964 pattern) showed incorrect output for anything following an extended class
3965 that contained multibyte characters and which was followed by a quantifier.
3966
3967 15. Added optional support for general category Unicode character properties
3968 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3969 support. It adds about 90K to the size of the library. The meanings of the
3970 inbuilt class escapes such as \d and \s have NOT been changed.
3971
3972 16. Updated pcredemo.c to include calls to free() to release the memory for the
3973 compiled pattern.
3974
3975 17. The generated file chartables.c was being created in the source directory
3976 instead of in the building directory. This caused the build to fail if the
3977 source directory was different from the building directory, and was
3978 read-only.
3979
3980 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3981 file. No doubt somebody will tell me if they don't make sense... Also added
3982 Dan Mooney's comments about building on OpenVMS.
3983
3984 19. Added support for partial matching via the PCRE_PARTIAL option for
3985 pcre_exec() and the \P data escape in pcretest.
3986
3987 20. Extended pcretest with 3 new pattern features:
3988
3989 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3990 write the compiled pattern to the file whose name is "rest-of-line".
3991 This is a straight binary dump of the data, with the saved pointer to
3992 the character tables forced to be NULL. The study data, if any, is
3993 written too. After writing, pcretest reads a new pattern.
3994
3995 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3996 compiled pattern from the given file. There must not be any
3997 occurrences of "<" in the file name (pretty unlikely); if there are,
3998 pcretest will instead treat the initial "<" as a pattern delimiter.
3999 After reading in the pattern, pcretest goes on to read data lines as
4000 usual.
4001
4002 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4003 and 16-bit fields in a compiled pattern, to simulate a pattern that
4004 was compiled on a host of opposite endianness.
4005
4006 21. The pcre-exec() function can now cope with patterns that were compiled on
4007 hosts of opposite endianness, with this restriction:
4008
4009 As for any compiled expression that is saved and used later, the tables
4010 pointer field cannot be preserved; the extra_data field in the arguments
4011 to pcre_exec() should be used to pass in a tables address if a value
4012 other than the default internal tables were used at compile time.
4013
4014 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4015 now diagnosed as an error. Previously, most of the time, a negative number
4016 would have been treated as zero, but if in addition "ovector" was passed as
4017 NULL, a crash could occur.
4018
4019 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4020 new versions from the libtool 1.5 distribution (the last one is a copy of
4021 a file called libtool.m4). This seems to have fixed the need to patch
4022 "configure" to support Darwin 1.3 (which I used to do). However, I still
4023 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4024 workstation).
4025
4026 24. Changed the PCRE licence to be the more standard "BSD" licence.
4027
4028
4029 Version 4.5 01-Dec-03
4030 ---------------------
4031
4032 1. There has been some re-arrangement of the code for the match() function so
4033 that it can be compiled in a version that does not call itself recursively.
4034 Instead, it keeps those local variables that need separate instances for
4035 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4036 needs to "recurse". Keeping track of where control must go is done by means
4037 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4038 hide most of the details from the main code, and operates only if
4039 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4040 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4041 operating.
4042
4043 To make it easier for callers to provide specially tailored get/free
4044 functions for this usage, two new functions, pcre_stack_malloc, and
4045 pcre_stack_free, are used. They are always called in strict stacking order,
4046 and the size of block requested is always the same.
4047
4048 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4049 PCRE has been compiled to use the stack or the heap for recursion. The
4050 -C option of pcretest uses this to show which version is compiled.
4051
4052 A new data escape \S, is added to pcretest; it causes the amounts of store
4053 obtained and freed by both kinds of malloc/free at match time to be added
4054 to the output.
4055
4056 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4057 what's available on my current Linux desktop machine.
4058
4059 3. When matching a UTF-8 string, the test for a valid string at the start has
4060 been extended. If start_offset is not zero, PCRE now checks that it points
4061 to a byte that is the start of a UTF-8 character. If not, it returns
4062 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4063 this is necessary because there may be backward assertions in the pattern.
4064 When matching the same subject several times, it may save resources to use
4065 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4066
4067 4. The code for checking the validity of UTF-8 strings has been tightened so
4068 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4069 containing "overlong sequences".
4070
4071 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4072 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4073 should have been "&", but it just so happened that all the cases this let
4074 through by mistake were picked up later in the function.
4075
4076 6. I had used a variable called "isblank" - this is a C99 function, causing
4077 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4078
4079 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4080 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4081 so the version is shown; (c) stop "make test" echoing "./RunTest".
4082
4083 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4084
4085 9. The prototype for memmove() for systems that don't have it was using
4086 size_t, but the inclusion of the header that defines size_t was later. I've
4087 moved the #includes for the C headers earlier to avoid this.
4088
4089 10. Added some adjustments to the code to make it easier to compiler on certain
4090 special systems:
4091
4092 (a) Some "const" qualifiers were missing.
4093 (b) Added the macro EXPORT before all exported functions; by default this
4094 is defined to be empty.
4095 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4096 that it reads its output file name as an argument instead of writing
4097 to the standard output and assuming this can be redirected.
4098
4099 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4100 class containing characters with values greater than 255, PCRE compilation
4101 went into a loop.
4102
4103 12. A recursive reference to a subpattern that was within another subpattern
4104 that had a minimum quantifier of zero caused PCRE to crash. For example,
4105 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4106 recursion. If the recursively-called subpattern itself had a zero repeat,
4107 that was OK.
4108
4109 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4110 buffer into which it was copied (for escape processing) was still set at
4111 1024, so long lines caused crashes.
4112
4113 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4114 "internal error: code overflow...". This applied to any character class
4115 that was followed by a possessive quantifier.
4116
4117 15. Modified the Makefile to add libpcre.la as a prerequisite for
4118 libpcreposix.la because I was told this is needed for a parallel build to
4119 work.
4120
4121 16. If a pattern that contained .* following optional items at the start was
4122 studied, the wrong optimizing data was generated, leading to matching
4123 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4124 matching string must start with a or b or c. The correct conclusion for
4125 this pattern is that a match can start with any character.
4126
4127
4128 Version 4.4 13-Aug-03
4129 ---------------------
4130
4131 1. In UTF-8 mode, a character class containing characters with values between
4132 127 and 255 was not handled correctly if the compiled pattern was studied.
4133 In fixing this, I have also improved the studying algorithm for such
4134 classes (slightly).
4135
4136 2. Three internal functions had redundant arguments passed to them. Removal
4137 might give a very teeny performance improvement.
4138
4139 3. Documentation bug: the value of the capture_top field in a callout is *one
4140 more than* the number of the hightest numbered captured substring.
4141
4142 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4143 in incorrectly linking with a previously installed version. They now link
4144 explicitly with libpcre.la.
4145
4146 5. configure.in no longer needs to recognize Cygwin specially.
4147
4148 6. A problem in pcre.in for Windows platforms is fixed.
4149
4150 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4151 pcretest, it used to include the size of the study block as part of its
4152 output. Unfortunately, the structure contains a field that has a different
4153 size on different hardware architectures. This meant that the tests that
4154 showed this size failed. As the block is currently always of a fixed size,
4155 this information isn't actually particularly useful in pcretest output, so
4156 I have just removed it.
4157
4158 8. Three pre-processor statements accidentally did not start in column 1.
4159 Sadly, there are *still* compilers around that complain, even though
4160 standard C has not required this for well over a decade. Sigh.
4161
4162 9. In pcretest, the code for checking callouts passed small integers in the
4163 callout_data field, which is a void * field. However, some picky compilers
4164 complained about the casts involved for this on 64-bit systems. Now
4165 pcretest passes the address of the small integer instead, which should get
4166 rid of the warnings.
4167
4168 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4169 both compile and run time, and gives an error if an invalid UTF-8 sequence
4170 is found. There is a option for disabling this check in cases where the
4171 string is known to be correct and/or the maximum performance is wanted.
4172
4173 11. In response to a bug report, I changed one line in Makefile.in from
4174
4175 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4176 to
4177 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4178
4179 to look similar to other lines, but I have no way of telling whether this
4180 is the right thing to do, as I do not use Windows. No doubt I'll get told
4181 if it's wrong...
4182
4183
4184 Version 4.3 21-May-03
4185 ---------------------
4186
4187 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4188 Makefile.
4189
4190 2. Some refactoring to improve the quality of the code:
4191
4192 (i) The utf8_table... variables are now declared "const".
4193
4194 (ii) The code for \cx, which used the "case flipping" table to upper case
4195 lower case letters, now just substracts 32. This is ASCII-specific,
4196 but the whole concept of \cx is ASCII-specific, so it seems
4197 reasonable.
4198
4199 (iii) PCRE was using its character types table to recognize decimal and
4200 hexadecimal digits in the pattern. This is silly, because it handles
4201 only 0-9, a-f, and A-F, but the character types table is locale-
4202 specific, which means strange things might happen. A private
4203 table is now used for this - though it costs 256 bytes, a table is
4204 much faster than multiple explicit tests. Of course, the standard
4205 character types table is still used for matching digits in subject
4206 strings against \d.
4207
4208 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4209 ending in _t are). So I've renamed it as ESC_tee.
4210
4211 3. The first argument for regexec() in the POSIX wrapper should have been
4212 defined as "const".
4213
4214 4. Changed pcretest to use malloc() for its buffers so that they can be
4215 Electric Fenced for debugging.
4216
4217 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4218 to read one or more bytes before the start of the subject string. Often this
4219 had no effect on PCRE's behaviour, but in some circumstances it could
4220 provoke a segmentation fault.
4221
4222 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4223 to try to read one or more bytes before the start of the subject string.
4224
4225 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4226 UTF-8 support could misbehave in various ways if the subject string
4227 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4228 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4229 back over UTF-8 characters.)
4230
4231
4232 Version 4.2 14-Apr-03
4233 ---------------------
4234
4235 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4236
4237 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4238 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4239 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4240 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4241 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4242 and BUILD_EXEEXT
4243 Note: automatic setting of the BUILD variables is not yet working
4244 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4245 compile-time but not at link-time
4246 [LINK]: use for linking executables only
4247 make different versions for Windows and non-Windows
4248 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4249 libraries
4250 [LINK_FOR_BUILD]: new variable
4251 [OBJEXT]: use throughout
4252 [EXEEXT]: use throughout
4253 <winshared>: new target
4254 <wininstall>: new target
4255 <dftables.o>: use native compiler
4256 <dftables>: use native linker
4257 <install>: handle Windows platform correctly
4258 <clean>: ditto
4259 <check>: ditto
4260 copy DLL to top builddir before testing
4261
4262 As part of these changes, -no-undefined was removed again. This was reported
4263 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4264 in any case.
4265
4266 3. Some tidies to get rid of compiler warnings:
4267
4268 . In the match_data structure, match_limit was an unsigned long int, whereas
4269 match_call_count was an int. I've made them both unsigned long ints.
4270
4271 . In pcretest the fact that a const uschar * doesn't automatically cast to
4272 a void * provoked a warning.
4273
4274 . Turning on some more compiler warnings threw up some "shadow" variables
4275 and a few more missing casts.
4276
4277 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4278 option, a class that contained a single character with a value between 128
4279 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4280
4281 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4282 option, a class that contained several characters, but with at least one
4283 whose value was between 128 and 255 caused PCRE to crash.
4284
4285
4286 Version 4.1 12-Mar-03
4287 ---------------------
4288
4289 1. Compiling with gcc -pedantic found a couple of places where casts were
4290 needed, and a string in dftables.c that was longer than standard compilers are
4291 required to support.
4292
4293 2. Compiling with Sun's compiler found a few more places where the code could
4294 be tidied up in order to avoid warnings.
4295
4296 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4297 first of these names is deprecated in the latest Autoconf in favour of the name
4298 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4299 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4300 analogy I have changed it to CFLAGS_FOR_BUILD.
4301
4302 4. Added -no-undefined to the linking command in the Makefile, because this is
4303 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4304 linking step for the pcreposix library.
4305
4306 5. PCRE was failing to diagnose the case of two named groups with the same
4307 name.
4308
4309 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4310 literal character that is needed in the subject for a match, and scans along to
4311 ensure that it is present before embarking on the full matching process. This
4312 saves time in cases of nested unlimited repeats that are never going to match.
4313 Problem: the scan can take a lot of time if the subject is very long (e.g.
4314 megabytes), thus penalizing straightforward matches. It is now done only if the
4315 amount of subject to be scanned is less than 1000 bytes.
4316
4317 7. A lesser problem with the same optimization is that it was recording the
4318 first character of an anchored pattern as "needed", thus provoking a search
4319 right along the subject, even when the first match of the pattern was going to
4320 fail. The "needed" character is now not set for anchored patterns, unless it
4321 follows something in the pattern that is of non-fixed length. Thus, it still
4322 fulfils its original purpose of finding quick non-matches in cases of nested
4323 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4324
4325
4326 Version 4.0 17-Feb-03
4327 ---------------------
4328
4329 1. If a comment in an extended regex that started immediately after a meta-item
4330 extended to the end of string, PCRE compiled incorrect data. This could lead to
4331 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4332
4333 2. Moved to autoconf 2.53 and libtool 1.4.2.
4334
4335 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4336 the special perltest8 script is no longer needed - all the tests can be run
4337 from a single perltest script.
4338
4339 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4340 by \s. It has now been removed in PCRE. This means it isn't recognized as
4341 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4342 class [:space:] *does* include VT, thereby creating a mess.
4343
4344 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4345 space and tab.
4346
4347 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4348 its new features into the main test script, reducing the number of scripts.
4349
4350 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4351 were backward compatible, and made the (?i) apply to the whole pattern, as if
4352 /i were given. Now it behaves more logically, and applies the option setting
4353 only to what follows. PCRE has been changed to follow suit. However, if it
4354 finds options settings right at the start of the pattern, it extracts them into
4355 the global options, as before. Thus, they show up in the info data.
4356
4357 8. Added support for the \Q...\E escape sequence. Characters in between are
4358 treated as literals. This is slightly different from Perl in that $ and @ are
4359 also handled as literals inside the quotes. In Perl, they will cause variable
4360 interpolation. Note the following examples:
4361
4362 Pattern PCRE matches Perl matches
4363
4364 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4365 \Qabc\$xyz\E abc\$xyz abc\$xyz
4366 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4367
4368 For compatibility with Perl, \Q...\E sequences are recognized inside character
4369 classes as well as outside them.
4370
4371 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4372 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4373 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4374 signed/unsigned warnings.
4375
4376 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4377 option for pcretest, so I've replaced it by a simple function that does just
4378 that job.
4379
4380 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4381 "pcregrep -".
4382
4383 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4384 Java package. This provides some syntactic sugar for simple cases of what my
4385 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4386 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4387 item, you can use this simplified notation. Note that only makes sense with
4388 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4389 greediness, whatever the setting of the PCRE_UNGREEDY option.
4390
4391 13. A change of greediness default within a pattern was not taking effect at
4392 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4393 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4394 was abstracted outside.
4395
4396 14. PCRE now supports the \G assertion. It is true when the current matching
4397 position is at the start point of the match. This differs from \A when the
4398 starting offset is non-zero. Used with the /g option of pcretest (or similar
4399 code), it works in the same way as it does for Perl's /g option. If all
4400 alternatives of a regex begin with \G, the expression is anchored to the start
4401 match position, and the "anchored" flag is set in the compiled expression.
4402
4403 15. Some bugs concerning the handling of certain option changes within patterns
4404 have been fixed. These applied to options other than (?ims). For example,
4405 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4406 been the other way round. Some of this was related to change 7 above.
4407
4408 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4409 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4410 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4411 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4412
4413 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4414 mode. Unlike ".", it always matches newline, whatever the setting of
4415 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4416 assertions. Perl allows it, but it doesn't (in general) work because it can't
4417 calculate the length of the lookbehind. At least, that's the case for Perl
4418 5.8.0 - I've been told they are going to document that it doesn't work in
4419 future.
4420
4421 18. Added an error diagnosis for escapes that PCRE does not support: these are
4422 \L, \l, \N, \P, \p, \U, \u, and \X.
4423
4424 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4425 reading past the end of the pattern in cases such as /[abcd/.
4426
4427 20. PCRE was getting more memory than necessary for patterns with classes that
4428 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4429
4430 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4431 compiling PCRE for use with Virtual Pascal.
4432
4433 22. Small fix to the Makefile to make it work properly if the build is done
4434 outside the source tree.
4435
4436 23. Added a new extension: a condition to go with recursion. If a conditional
4437 subpattern starts with (?(R) the "true" branch is used if recursion has
4438 happened, whereas the "false" branch is used only at the top level.
4439
4440 24. When there was a very long string of literal characters (over 255 bytes
4441 without UTF support, over 250 bytes with UTF support), the computation of how
4442 much memory was required could be incorrect, leading to segfaults or other
4443 strange effects.
4444
4445 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4446 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4447 there was a subsequent back reference to those brackets. This meant that, for
4448 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4449 possible to check for precisely this case. All we can do is abandon the
4450 optimization if .* occurs inside capturing brackets when there are any back
4451 references whatsoever. (See below for a better fix that came later.)
4452
4453 26. The handling of the optimization for finding the first character of a
4454 non-anchored pattern, and for finding a character that is required later in the
4455 match were failing in some cases. This didn't break the matching; it just
4456 failed to optimize when it could. The way this is done has been re-implemented.
4457
4458 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4459
4460 28. Added a new feature that provides some of the functionality that Perl
4461 provides with (?{...}). The facility is termed a "callout". The way it is done
4462 in PCRE is for the caller to provide an optional function, by setting
4463 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4464 global variable. By default it is unset, which disables all calling out. To get
4465 the function called, the regex must include (?C) at appropriate points. This
4466 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4467 This provides a means of identifying different callout points. When PCRE
4468 reaches such a point in the regex, if pcre_callout has been set, the external
4469 function is called. It is provided with data in a structure called
4470 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4471 matching continues; if it returns a non-zero value, the match at the current
4472 point fails. However, backtracking will occur if possible. [This was changed
4473 later and other features added - see item 49 below.]
4474
4475 29. pcretest is upgraded to test the callout functionality. It provides a
4476 callout function that displays information. By default, it shows the start of
4477 the match and the current position in the text. There are some new data escapes
4478 to vary what happens:
4479
4480 \C+ in addition, show current contents of captured substrings
4481 \C- do not supply a callout function
4482 \C!n return 1 when callout number n is reached
4483 \C!n!m return 1 when callout number n is reached for the mth time
4484
4485 30. If pcregrep was called with the -l option and just a single file name, it
4486 output "<stdin>" if a match was found, instead of the file name.
4487
4488 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4489 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4490 pcre_exec(). This saves a malloc/free per call. The default value of
4491 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4492 when configuring.
4493
4494 32. The default maximum size of a compiled pattern is 64K. There have been a
4495 few cases of people hitting this limit. The code now uses macros to handle the
4496 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4497 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4498 configuring. Tests 2 and 5 work only with 2-byte links because they output
4499 debugging information about compiled patterns.
4500
4501 33. Internal code re-arrangements:
4502
4503 (a) Moved the debugging function for printing out a compiled regex into
4504 its own source file (printint.c) and used #include to pull it into
4505 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4506 separate copies.
4507
4508 (b) Defined the list of op-code names for debugging as a macro in
4509 internal.h so that it is next to the definition of the opcodes.
4510
4511 (c) Defined a table of op-code lengths for simpler skipping along compiled
4512 code. This is again a macro in internal.h so that it is next to the
4513 definition of the opcodes.
4514
4515 34. Added support for recursive calls to individual subpatterns, along the
4516 lines of Robin Houston's patch (but implemented somewhat differently).
4517
4518 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4519 allow it to read and process whole directories in Win32. This code was
4520 contributed by Lionel Fourquaux; it has not been tested by me.
4521
4522 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4523 used to name a group. Names consist of alphanumerics and underscores, and must
4524 be unique. Back references use the syntax (?P=name) and recursive calls use
4525 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4526 numbers. The function pcre_fullinfo() can be used after compilation to extract
4527 a name/number map. There are three relevant calls:
4528
4529 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4530 PCRE_INFO_NAMECOUNT yields the number of entries
4531 PCRE_INFO_NAMETABLE yields a pointer to the map.
4532
4533 The map is a vector of fixed-size entries. The size of each entry depends on
4534 the length of the longest name used. The first two bytes of each entry are the
4535 group number, most significant byte first. There follows the corresponding
4536 name, zero terminated. The names are in alphabetical order.
4537
4538 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4539 case instead of 255. Making it the same both with and without UTF-8 support
4540 means that the same test output works with both.
4541
4542 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4543 calling malloc() with a zero argument.
4544
4545 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4546 optimization. I've improved things by keeping a bitmap of backreferences with
4547 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4548 fact referenced, the optimization can be applied. It is unlikely that a
4549 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4550 the match to follow \n) will appear inside brackets with a number greater than
4551 31, but if it does, any back reference > 31 suppresses the optimization.
4552
4553 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4554 of disabling numbered capturing parentheses. Any opening parenthesis that is
4555 not followed by ? behaves as if it were followed by ?: but named parentheses
4556 can still be used for capturing (and they will acquire numbers in the usual
4557 way).
4558
4559 41. Redesigned the return codes from the match() function into yes/no/error so
4560 that errors can be passed back from deep inside the nested calls. A malloc
4561 failure while inside a recursive subpattern call now causes the
4562 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4563
4564 42. It is now possible to set a limit on the number of times the match()
4565 function is called in a call to pcre_exec(). This facility makes it possible to
4566 limit the amount of recursion and backtracking, though not in a directly
4567 obvious way, because the match() function is used in a number of different
4568 circumstances. The count starts from zero for each position in the subject
4569 string (for non-anchored patterns). The default limit is, for compatibility, a
4570 large number, namely 10 000 000. You can change this in two ways:
4571
4572 (a) When configuring PCRE before making, you can use --with-match-limit=n
4573 to set a default value for the compiled library.
4574
4575 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4576 a different value is set. See 45 below.
4577
4578 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4579
4580 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4581 of things that can be changed at compile time. The first argument specifies
4582 what is wanted and the second points to where the information is to be placed.
4583 The current list of available information is:
4584
4585 PCRE_CONFIG_UTF8
4586
4587 The output is an integer that is set to one if UTF-8 support is available;
4588 otherwise it is set to zero.
4589
4590 PCRE_CONFIG_NEWLINE
4591
4592 The output is an integer that it set to the value of the code that is used for
4593 newline. It is either LF (10) or CR (13).
4594
4595 PCRE_CONFIG_LINK_SIZE
4596
4597 The output is an integer that contains the number of bytes used for internal
4598 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4599
4600 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4601
4602 The output is an integer that contains the threshold above which the POSIX
4603 interface uses malloc() for output vectors. See item 31 above.
4604
4605 PCRE_CONFIG_MATCH_LIMIT
4606
4607 The output is an unsigned integer that contains the default limit of the number
4608 of match() calls in a pcre_exec() execution. See 42 above.
4609
4610 44. pcretest has been upgraded by the addition of the -C option. This causes it
4611 to extract all the available output from the new pcre_config() function, and to
4612 output it. The program then exits immediately.
4613
4614 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4615 order to support additional features. One way would have been to define
4616 pcre_exec2() (for example) with extra arguments, but this would not have been
4617 extensible, and would also have required all calls to the original function to
4618 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4619 is used for passing in "extra" data from pcre_study().
4620
4621 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4622 contains the following fields:
4623
4624 flags a bitmap indicating which of the following fields are set
4625 study_data opaque data from pcre_study()
4626 match_limit a way of specifying a limit on match() calls for a specific
4627 call to pcre_exec()
4628 callout_data data for callouts (see 49 below)
4629
4630 The flag bits are also defined in pcre.h, and are
4631
4632 PCRE_EXTRA_STUDY_DATA
4633 PCRE_EXTRA_MATCH_LIMIT
4634 PCRE_EXTRA_CALLOUT_DATA
4635
4636 The pcre_study() function now returns one of these new pcre_extra blocks, with
4637 the actual study data pointed to by the study_data field, and the
4638 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4639 before. That is, this change is entirely upwards-compatible and requires no
4640 change to existing code.
4641
4642 If you want to pass in additional data to pcre_exec(), you can either place it
4643 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4644 block.
4645
4646 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4647 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4648 times with different match limits, until it finds the minimum value needed for
4649 pcre_exec() to complete. The value is then output. This can be instructive; for
4650 most simple matches the number is quite small, but for pathological cases it
4651 gets very large very quickly.
4652
4653 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4654 returns the size of the data block pointed to by the study_data field in a
4655 pcre_extra block, that is, the value that was passed as the argument to
4656 pcre_malloc() when PCRE was getting memory in which to place the information
4657 created by pcre_study(). The fourth argument should point to a size_t variable.
4658 pcretest has been extended so that this information is shown after a successful
4659 pcre_study() call when information about the compiled regex is being displayed.
4660
4661 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4662 because what follows is always an absolute path. (Later: it turns out that this
4663 is more than cosmetic for MinGW, because it doesn't like empty path
4664 components.)
4665
4666 49. Some changes have been made to the callout feature (see 28 above):
4667
4668 (i) A callout function now has three choices for what it returns:
4669
4670 0 => success, carry on matching
4671 > 0 => failure at this point, but backtrack if possible
4672 < 0 => serious error, return this value from pcre_exec()
4673
4674 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4675 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4676 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4677 use by callout functions. It will never be used by PCRE itself.
4678
4679 (ii) The pcre_extra structure (see 45 above) has a void * field called
4680 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4681 pcre_callout_block structure has a field of the same name. The contents of
4682 the field passed in the pcre_extra structure are passed to the callout
4683 function in the corresponding field in the callout block. This makes it
4684 easier to use the same callout-containing regex from multiple threads. For
4685 testing, the pcretest program has a new data escape
4686
4687 \C*n pass the number n (may be negative) as callout_data
4688
4689 If the callout function in pcretest receives a non-zero value as
4690 callout_data, it returns that value.
4691
4692 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4693 there were some redundant $(CFLAGS) in commands that are now specified as
4694 $(LINK), which already includes $(CFLAGS).
4695
4696 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4697 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4698 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4699 one-byte characters throughout. Note that case-insensitive matching applies
4700 only to characters whose values are less than 256. PCRE doesn't support the
4701 notion of cases for higher-valued characters.
4702
4703 (i) A character class whose characters are all within 0-255 is handled as
4704 a bit map, and the map is inverted for negative classes. Previously, a
4705 character > 255 always failed to match such a class; however it should
4706 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4707
4708 (ii) A negated character class with a single character < 255 is coded as
4709 "not this character" (OP_NOT). This wasn't working properly when the test
4710 character was multibyte, either singly or repeated.
4711
4712 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4713 mode, for example: \x{100}{2,3}.
4714
4715 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4716 singly or repeated) now correctly test multibyte characters. However,
4717 PCRE doesn't recognize any characters with values greater than 255 as
4718 digits, spaces, or word characters. Such characters always match \D, \S,
4719 and \W, and never match \d, \s, or \w.
4720
4721 (v) Classes may now contain characters and character ranges with values
4722 greater than 255. For example: [ab\x{100}-\x{400}].
4723
4724 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4725 PCRE in UTF-8 mode.
4726
4727 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4728 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4729 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4730 value.)
4731
4732 53. The single man page has become too large. I have therefore split it up into
4733 a number of separate man pages. These also give rise to individual HTML pages;
4734 these are now put in a separate directory, and there is an index.html page that
4735 lists them all. Some hyperlinking between the pages has been installed.
4736
4737 54. Added convenience functions for handling named capturing parentheses.
4738
4739 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4740 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4741 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4742 are faulted.
4743
4744 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4745 calling configure. These values are used when compiling the dftables.c program
4746 which is run to generate the source of the default character tables. They
4747 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4748 you will need to set these values.
4749
4750 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4751
4752
4753 Version 3.9 02-Jan-02
4754 ---------------------
4755
4756 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4757
4758 2. If --disable-static was given, the building process failed when trying to
4759 build pcretest and pcregrep. (For some reason it was using libtool to compile
4760 them, which is not right, as they aren't part of the library.)
4761
4762
4763 Version 3.8 18-Dec-01
4764 ---------------------
4765
4766 1. The experimental UTF-8 code was completely screwed up. It was packing the
4767 bytes in the wrong order. How dumb can you get?
4768
4769
4770 Version 3.7 29-Oct-01
4771 ---------------------
4772
4773 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4774 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4775 this didn't happen under Solaris 8, where I normally test things.
4776
4777 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4778 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4779 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4780 this fix an hour or so after the initial 3.7 release.)
4781
4782
4783 Version 3.6 23-Oct-01
4784 ---------------------
4785
4786 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4787 offsets passed as NULL with zero offset count.
4788
4789 2. The config.guess and config.sub files had not been updated when I moved to
4790 the latest autoconf.
4791
4792
4793 Version 3.5 15-Aug-01
4794 ---------------------
4795
4796 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4797 had been forgotten.
4798
4799 2. By using declared but undefined structures, we can avoid using "void"
4800 definitions in pcre.h while keeping the internal definitions of the structures
4801 private.
4802
4803 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4804 user point of view, this means that both static and shared libraries are built
4805 by default, but this can be individually controlled. More of the work of
4806 handling this static/shared cases is now inside libtool instead of PCRE's make
4807 file.
4808
4809 4. The pcretest utility is now installed along with pcregrep because it is
4810 useful for users (to test regexs) and by doing this, it automatically gets
4811 relinked by libtool. The documentation has been turned into a man page, so
4812 there are now .1, .txt, and .html versions in /doc.
4813
4814 5. Upgrades to pcregrep:
4815 (i) Added long-form option names like gnu grep.
4816 (ii) Added --help to list all options with an explanatory phrase.
4817 (iii) Added -r, --recursive to recurse into sub-directories.
4818 (iv) Added -f, --file to read patterns from a file.
4819
4820 6. pcre_exec() was referring to its "code" argument before testing that
4821 argument for NULL (and giving an error if it was NULL).
4822
4823 7. Upgraded Makefile.in to allow for compiling in a different directory from
4824 the source directory.
4825
4826 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4827 options bits, the pointer it was passed was to an int instead of to an unsigned
4828 long int. This mattered only on 64-bit systems.
4829
4830 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4831 generated) instead of pcre.in, which it its source. Also made the same change
4832 in several of the .c files.
4833
4834 10. A new release of gcc defines printf() as a macro, which broke pcretest
4835 because it had an ifdef in the middle of a string argument for printf(). Fixed
4836 by using separate calls to printf().
4837
4838 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4839 script, to force use of CR or LF instead of \n in the source. On non-Unix
4840 systems, the value can be set in config.h.
4841
4842 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4843 absolute limit. Changed the text of the error message to make this clear, and
4844 likewise updated the man page.
4845
4846 13. The limit of 99 on the number of capturing subpatterns has been removed.
4847 The new limit is 65535, which I hope will not be a "real" limit.
4848
4849
4850 Version 3.4 22-Aug-00
4851 ---------------------
4852
4853 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4854
4855 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4856
4857
4858 Version 3.3 01-Aug-00
4859 ---------------------
4860
4861 1. If an octal character was given, but the value was greater than \377, it
4862 was not getting masked to the least significant bits, as documented. This could
4863 lead to crashes in some systems.
4864
4865 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4866 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4867
4868 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4869 These just pass their arguments on to (pcre_free)(), but they are provided
4870 because some uses of PCRE bind it to non-C systems that can call its functions,
4871 but cannot call free() or pcre_free() directly.
4872
4873 4. Add "make test" as a synonym for "make check". Corrected some comments in
4874 the Makefile.
4875
4876 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4877 Makefile.
4878
4879 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4880 command called pgrep for grepping around the active processes.
4881
4882 7. Added the beginnings of support for UTF-8 character strings.
4883
4884 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4885 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4886 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4887 out for the ar command.)
4888
4889
4890 Version 3.2 12-May-00
4891 ---------------------
4892
4893 This is purely a bug fixing release.
4894
4895 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4896 of ZA. This was just one example of several cases that could provoke this bug,
4897 which was introduced by change 9 of version 2.00. The code for breaking
4898 infinite loops after an iteration that matches an empty string was't working
4899 correctly.
4900
4901 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4902 when matched against abbab (for example). After matching an empty string, it
4903 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4904 caused it to match further down the string than it should.
4905
4906 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4907 was there because it doesn't seem to be needed, and it causes trouble on some
4908 systems, as it is not a Standard C header. It has been removed.
4909
4910 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4911 were reported on the Macintosh. The changes were from
4912
4913 while ((c = *(++ptr)) != 0 && c != '\n');
4914 to
4915 while ((c = *(++ptr)) != 0 && c != '\n') ;
4916
4917 Totally extraordinary, but if that's what it takes...
4918
4919 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4920 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4921 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4922 assumes the way PCRE uses memmove() (always moving upwards).
4923
4924 6. PCRE is being used in one environment where strchr() is not available. There
4925 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4926 faster code anyway.
4927
4928
4929 Version 3.1 09-Feb-00
4930 ---------------------
4931
4932 The only change in this release is the fixing of some bugs in Makefile.in for
4933 the "install" target:
4934
4935 (1) It was failing to install pcreposix.h.
4936
4937 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4938
4939
4940 Version 3.0 01-Feb-00
4941 ---------------------
4942
4943 1. Add support for the /+ modifier to perltest (to output $` like it does in
4944 pcretest).
4945
4946 2. Add support for the /g modifier to perltest.
4947
4948 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4949 matches null strings.
4950
4951 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4952 pattern. Perl treats empty patterns specially - it reuses the most recent
4953 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4954 effect.
4955
4956 5. The POSIX interface was broken in that it was just handing over the POSIX
4957 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4958 required a bigger vector, with some working space on the end. This means that
4959 the POSIX wrapper now has to get and free some memory, and copy the results.
4960
4961 6. Added some simple autoconf support, placing the test data and the
4962 documentation in separate directories, re-organizing some of the
4963 information files, and making it build pcre-config (a GNU standard). Also added
4964 libtool support for building PCRE as a shared library, which is now the
4965 default.
4966
4967 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4968 09 are not valid octal constants. Single digits will be used for minor values
4969 less than 10.
4970
4971 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4972 existing programs that set these in the POSIX interface can use PCRE without
4973 modification.
4974
4975 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4976 return all that pcre_info() returns, plus additional data. The pcre_info()
4977 function is retained for compatibility, but is considered to be obsolete.
4978
4979 10. Added experimental recursion feature (?R) to handle one common case that
4980 Perl 5.6 will be able to do with (?p{...}).
4981
4982 11. Added support for POSIX character classes like [:alpha:], which Perl is
4983 adopting.
4984
4985
4986 Version 2.08 31-Aug-99
4987 ----------------------
4988
4989 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4990 trying to match at the startoffset position, but instead was moving forward to
4991 the next newline as if a previous match had failed.
4992
4993 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4994 and could get into a loop if a null string was matched other than at the start
4995 of the subject.
4996
4997 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4998 be distinguished at compile time, and for completeness also added PCRE_DATE.
4999
5000 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5001 in GnuWin32 environments.
5002
5003
5004 Version 2.07 29-Jul-99
5005 ----------------------
5006
5007 1. The documentation is now supplied in plain text form and HTML as well as in
5008 the form of man page sources.
5009
5010 2. C++ compilers don't like assigning (void *) values to other pointer types.
5011 In particular this affects malloc(). Although there is no problem in Standard
5012 C, I've put in casts to keep C++ compilers happy.
5013
5014 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5015 should be (const char *).
5016
5017 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5018 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5019 However, I haven't made this a standard facility. The documentation doesn't
5020 mention it, and the Makefile doesn't support it.
5021
5022 5. The Makefile now contains an "install" target, with editable destinations at
5023 the top of the file. The pcretest program is not installed.
5024
5025 6. pgrep -V now gives the PCRE version number and date.
5026
5027 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5028 causing the entire string to be ignored, instead of just the last character.
5029
5030 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5031 non-matching string, it can take a very, very long time, even for strings of
5032 quite modest length, because of the nested recursion. PCRE now does better in
5033 some of these cases. It does this by remembering the last required literal
5034 character in the pattern, and pre-searching the subject to ensure it is present
5035 before running the real match. In other words, it applies a heuristic to detect
5036 some types of certain failure quickly, and in the above example, if presented
5037 with a string that has no trailing " it gives "no match" very quickly.
5038
5039 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5040 other alternatives are tried instead.
5041
5042
5043 Version 2.06 09-Jun-99
5044 ----------------------
5045
5046 1. Change pcretest's output for amount of store used to show just the code
5047 space, because the remainder (the data block) varies in size between 32-bit and
5048 64-bit systems.
5049
5050 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5051 start matching at. This allows lookbehinds to work when searching for multiple
5052 occurrences in a string.
5053
5054 3. Added additional options to pcretest for testing multiple occurrences:
5055
5056 /+ outputs the rest of the string that follows a match
5057 /g loops for multiple occurrences, using the new startoffset argument
5058 /G loops for multiple occurrences by passing an incremented pointer
5059
5060 4. PCRE wasn't doing the "first character" optimization for patterns starting
5061 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5062 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5063 the letter 'x'. On long subject strings, this gives a significant speed-up.
5064
5065
5066 Version 2.05 21-Apr-99
5067 ----------------------
5068
5069 1. Changed the type of magic_number from int to long int so that it works
5070 properly on 16-bit systems.
5071
5072 2. Fixed a bug which caused patterns starting with .* not to work correctly
5073 when the subject string contained newline characters. PCRE was assuming
5074 anchoring for such patterns in all cases, which is not correct because .* will
5075 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5076 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5077 must be retried after every newline in the subject.
5078
5079
5080 Version 2.04 18-Feb-99
5081 ----------------------
5082
5083 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5084 computation of the store needed to hold the pattern was incorrect (too large).
5085 If such patterns were nested a few deep, this could multiply and become a real
5086 problem.
5087
5088 2. Added /M option to pcretest to show the memory requirement of a specific
5089 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5090
5091 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5092 compiled in such a way that the backtracking after subsequent failure was
5093 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5094 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5095
5096
5097 Version 2.03 02-Feb-99
5098 ----------------------
5099
5100 1. Fixed typo and small mistake in man page.
5101
5102 2. Added 4th condition (GPL supersedes if conflict) and created separate
5103 LICENCE file containing the conditions.
5104
5105 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5106 Perl, that is the internal \ allows the delimiter to be included in the
5107 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5108 the final delimiter, add \ to the end of the pattern (to test the error).
5109
5110 4. Added the convenience functions for extracting substrings after a successful
5111 match. Updated pcretest to make it able to test these functions.
5112
5113
5114 Version 2.02 14-Jan-99
5115 ----------------------
5116
5117 1. Initialized the working variables associated with each extraction so that
5118 their saving and restoring doesn't refer to uninitialized store.
5119
5120 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5121 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5122 fix the problem.
5123
5124 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5125 calls, and wasn't printing the correct value for compiling calls. Increased the
5126 default value of LOOPREPEAT, and the number of significant figures in the
5127 times.
5128
5129 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5130
5131 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5132 a building problem on Windows NT with a FAT file system.
5133
5134
5135 Version 2.01 21-Oct-98
5136 ----------------------
5137
5138 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5139 to character tables built by pcre_maketables() in the current locale. If NULL
5140 is passed, the default tables are used.
5141
5142
5143 Version 2.00 24-Sep-98
5144 ----------------------
5145
5146 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5147 it any more.
5148
5149 2. Allow quantification of (?>) groups, and make it work correctly.
5150
5151 3. The first character computation wasn't working for (?>) groups.
5152
5153 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5154 end of the subject) and add 5.005's \z, which really does match only at the
5155 very end of the subject.
5156
5157 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5158
5159 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5160 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5161 localized options. All options to pcre_study() were also removed.
5162
5163 7. Add other new features from 5.005:
5164
5165 $(?<= positive lookbehind
5166 $(?<! negative lookbehind
5167 (?imsx-imsx) added the unsetting capability
5168 such a setting is global if at outer level; local otherwise
5169 (?imsx-imsx:) non-capturing groups with option setting
5170 (?(cond)re|re) conditional pattern matching
5171
5172 A backreference to itself in a repeated group matches the previous
5173 captured string.
5174
5175 8. General tidying up of studying (both automatic and via "study")
5176 consequential on the addition of new assertions.
5177
5178 9. As in 5.005, unlimited repeated groups that could match an empty substring
5179 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5180 runtime if any iteration does actually match an empty substring.
5181
5182 10. Include the RunTest script in the distribution.
5183
5184 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5185 discrepancies, some of which were old and were also with respect to 5.004. They
5186 have now been fixed.
5187
5188
5189 Version 1.09 28-Apr-98
5190 ----------------------
5191
5192 1. A negated single character class followed by a quantifier with a minimum
5193 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5194 program crashes, or just wrong answers. This did not apply to negated classes
5195 containing more than one character, or to minima other than one.
5196
5197
5198 Version 1.08 27-Mar-98
5199 ----------------------
5200
5201 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5202
5203 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5204 latter must appear before anything that relies on it in the pattern.
5205
5206
5207 Version 1.07 16-Feb-98
5208 ----------------------
5209
5210 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5211 repeat of a potentially empty string).
5212
5213
5214 Version 1.06 23-Jan-98
5215 ----------------------
5216
5217 1. Added Markus Oberhumer's little patches for C++.
5218
5219 2. Literal strings longer than 255 characters were broken.
5220
5221
5222 Version 1.05 23-Dec-97
5223 ----------------------
5224
5225 1. Negated character classes containing more than one character were failing if
5226 PCRE_CASELESS was set at run time.
5227
5228
5229 Version 1.04 19-Dec-97
5230 ----------------------
5231
5232 1. Corrected the man page, where some "const" qualifiers had been omitted.
5233
5234 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5235 input syntax.
5236
5237 3. Fixed memory leak which occurred when a regex with back references was
5238 matched with an offsets vector that wasn't big enough. The temporary memory
5239 that is used in this case wasn't being freed if the match failed.
5240
5241 4. Tidied pcretest to ensure it frees memory that it gets.
5242
5243 5. Temporary memory was being obtained in the case where the passed offsets
5244 vector was exactly big enough.
5245
5246 6. Corrected definition of offsetof() from change 5 below.
5247
5248 7. I had screwed up change 6 below and broken the rules for the use of
5249 setjmp(). Now fixed.
5250
5251
5252 Version 1.03 18-Dec-97
5253 ----------------------
5254
5255 1. A erroneous regex with a missing opening parenthesis was correctly
5256 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5257 on some systems.
5258
5259 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5260 it was reported that one broken compiler failed on the former because "code" is
5261 also an independent variable.
5262
5263 3. The erroneous regex a[]b caused an array overrun reference.
5264
5265 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5266 fail on data ending with that character. (It was going on too far, and checking
5267 the next character, typically a binary zero.) This was specific to the
5268 optimized code for single-character negative classes.
5269
5270 5. Added a contributed patch from the TIN world which does the following:
5271
5272 + Add an undef for memmove, in case the the system defines a macro for it.
5273
5274 + Add a definition of offsetof(), in case there isn't one. (I don't know
5275 the reason behind this - offsetof() is part of the ANSI standard - but
5276 it does no harm).
5277
5278 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5279 most of the places where whitespace preceded '#'. I have given up and
5280 allowed the remaining 2 cases to be at the margin.
5281
5282 + Rename some variables in pcre to eliminate shadowing. This seems very
5283 pedantic, but does no harm, of course.
5284
5285 6. Moved the call to setjmp() into its own function, to get rid of warnings
5286 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5287
5288 7. Constructs such as \d{8,} were compiling into the equivalent of
5289 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5290 outcome, but in this particular case used more store than had been allocated,
5291 which caused the bug to be discovered because it threw up an internal error.
5292
5293 8. The debugging code in both pcre and pcretest for outputting the compiled
5294 form of a regex was going wrong in the case of back references followed by
5295 curly-bracketed repeats.
5296
5297
5298 Version 1.02 12-Dec-97
5299 ----------------------
5300
5301 1. Typos in pcre.3 and comments in the source fixed.
5302
5303 2. Applied a contributed patch to get rid of places where it used to remove
5304 'const' from variables, and fixed some signed/unsigned and uninitialized
5305 variable warnings.
5306
5307 3. Added the "runtest" target to Makefile.
5308
5309 4. Set default compiler flag to -O2 rather than just -O.
5310
5311
5312 Version 1.01 19-Nov-97
5313 ----------------------
5314
5315 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5316 like /([ab]*)*/, that is, for classes with more than one character in them.
5317
5318 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5319 as /((?>a*))*/ (a PCRE_EXTRA facility).
5320
5321
5322 Version 1.00 18-Nov-97
5323 ----------------------
5324
5325 1. Added compile-time macros to support systems such as SunOS4 which don't have
5326 memmove() or strerror() but have other things that can be used instead.
5327
5328 2. Arranged that "make clean" removes the executables.
5329
5330
5331 Version 0.99 27-Oct-97
5332 ----------------------
5333
5334 1. Fixed bug in code for optimizing classes with only one character. It was
5335 initializing a 32-byte map regardless, which could cause it to run off the end
5336 of the memory it had got.
5337
5338 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5339
5340
5341 Version 0.98 22-Oct-97
5342 ----------------------
5343
5344 1. Fixed bug in code for handling temporary memory usage when there are more
5345 back references than supplied space in the ovector. This could cause segfaults.
5346
5347
5348 Version 0.97 21-Oct-97
5349 ----------------------
5350
5351 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5352
5353 2. Optimized negated single characters not to use a bit map.
5354
5355 3. Brought error texts together as macro definitions; clarified some of them;
5356 fixed one that was wrong - it said "range out of order" when it meant "invalid
5357 escape sequence".
5358
5359 4. Changed some char * arguments to const char *.
5360
5361 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5362
5363 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5364 pcretest.
5365
5366
5367 Version 0.96 16-Oct-97
5368 ----------------------
5369
5370 1. Added a simple "pgrep" utility to the distribution.
5371
5372 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5373 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5374 where "ddd" means "one or more decimal digits".
5375
5376 3. Fixed serious bug. If a pattern had a back reference, but the call to
5377 pcre_exec() didn't supply a large enough ovector to record the related
5378 identifying subpattern, the match always failed. PCRE now remembers the number
5379 of the largest back reference, and gets some temporary memory in which to save
5380 the offsets during matching if necessary, in order to ensure that
5381 backreferences always work.
5382
5383 4. Increased the compatibility with Perl in a number of ways:
5384
5385 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5386 to request this handling. The option can be set at compile or exec time.
5387
5388 (b) $ matches before a terminating newline by default; an option
5389 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5390 mode). The option can be set at compile or exec time.
5391
5392 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5393 the same as Perl's. If the decimal number it represents is less than 10
5394 or there aren't that many previous left capturing parentheses, an octal
5395 escape is read. Inside a character class, it's always an octal escape,
5396 even if it is a single digit.
5397
5398 (d) An escaped but undefined alphabetic character is taken as a literal,
5399 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5400 escapes.
5401
5402 (e) {0} is now permitted. (The previous item is removed from the compiled
5403 pattern).
5404
5405 5. Changed all the names of code files so that the basic parts are no longer
5406 than 10 characters, and abolished the teeny "globals.c" file.
5407
5408 6. Changed the handling of character classes; they are now done with a 32-byte
5409 bit map always.
5410
5411 7. Added the -d and /D options to pcretest to make it possible to look at the
5412 internals of compilation without having to recompile pcre.
5413
5414
5415 Version 0.95 23-Sep-97
5416 ----------------------
5417
5418 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5419 \x20 at the start of a run of normal characters. These were being treated as
5420 real characters, instead of the source characters being re-checked.
5421
5422
5423 Version 0.94 18-Sep-97
5424 ----------------------
5425
5426 1. The functions are now thread-safe, with the caveat that the global variables
5427 containing pointers to malloc() and free() or alternative functions are the
5428 same for all threads.
5429
5430 2. Get pcre_study() to generate a bitmap of initial characters for non-
5431 anchored patterns when this is possible, and use it if passed to pcre_exec().
5432
5433
5434 Version 0.93 15-Sep-97
5435 ----------------------
5436
5437 1. /(b)|(:+)/ was computing an incorrect first character.
5438
5439 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5440 but not actually doing anything yet.
5441
5442 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5443 as Perl does (e.g. [-az] or [az-]).
5444
5445 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5446 all possible positions.
5447
5448 5. Split up into different modules to avoid including unneeded functions in a
5449 compiled binary. However, compile and exec are still in one module. The "study"
5450 function is split off.
5451
5452 6. The character tables are now in a separate module whose source is generated
5453 by an auxiliary program - but can then be edited by hand if required. There are
5454 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5455 toupper() in the code.
5456
5457 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5458 make them global. Abolish the function for setting them, as the caller can now
5459 set them directly.
5460
5461
5462 Version 0.92 11-Sep-97
5463 ----------------------
5464
5465 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5466 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5467
5468 2. Caseless matching was not working in character classes if the characters in
5469 the pattern were in upper case.
5470
5471 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5472
5473 4. Make PCRE_ANCHORED public and accept as a compile option.
5474
5475 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5476 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5477 pass them.
5478
5479 6. Give an error if bad option bits passed at compile or run time.
5480
5481 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5482 pcretest to cause it to pass that flag.
5483
5484 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5485 options, and the first character, if set.
5486
5487 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5488
5489
5490 Version 0.91 10-Sep-97
5491 ----------------------
5492
5493 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5494 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5495
5496 2. PCRE was looping on encountering an indefinitely repeated back reference to
5497 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5498 Perl does - treats the match as successful.
5499
5500 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5