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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1443 - (show annotations)
Sun Jan 12 19:20:27 2014 UTC (5 years, 9 months ago) by ph10
File size: 262272 byte(s)
Error occurred while calculating annotation data.
Check alternative outputs for the locale test in RunTest. It should now work 
for the 'fr' locale (which was broken).
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.35-RC1 xx-xxxx-201x
5 -----------------------------
6
7 1. A new flag is set, when property checks are present in an XCLASS.
8 When this flag is not set, PCRE can perform certain optimizations
9 such as studying these XCLASS-es.
10
11 2. The auto-possessification of character sets were improved: a normal
12 and an extended character set can be compared now. Furthermore
13 the JIT compiler optimizes more character set checks.
14
15 3. Got rid of some compiler warnings for potentially uninitialized variables
16 that show up only when compiled with -O2.
17
18 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
19 of a match later then the end of the match. The pcretest program was not
20 handling the case sensibly - it was outputting from the start to the next
21 binary zero. It now reports this situation in a message, and outputs the
22 text from the end to the start.
23
24 5. Fast forward search is improved in JIT. Instead of the first three
25 characters, any three characters with fixed position can be searched.
26 Search order: first, last, middle.
27
28 6. Improve character range checks in JIT. Characters are read by an inprecise
29 function now, which returns with an unknown value if the character code is
30 above a certain treshold (e.g: 256). The only limitation is that the value
31 must be bigger than the treshold as well. This function is useful, when
32 the characters above the treshold are handled in the same way.
33
34 7. The macros whose names start with RAWUCHAR are placeholders for a future
35 mode in which only the bottom 21 bits of 32-bit data items are used. To
36 make this more memorable for those maintaining the code, the names have
37 been changed to start with UCHAR21, and an extensive comment has been added
38 to their definition.
39
40 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
41 to the export list in Makefile.am (they were accidentally omitted from the
42 8.34 tarball).
43
44 9. The informational output from pcretest used the phrase "starting byte set"
45 which is inappropriate for the 16-bit and 32-bit libraries. As the output
46 for "first char" and "need char" really means "non-UTF-char", I've changed
47 "byte" to "char", and slightly reworded the output. The documentation about
48 these values has also been (I hope) clarified.
49
50 10. Another JIT related optimization: use table jumps for selecting the correct
51 backtracking path, when more than four alternatives are present inside a
52 bracket.
53
54 11. Empty match is not possible, when the minimum length is greater than zero,
55 and there is no \K in the pattern. JIT should avoid empty match checks in
56 such cases.
57
58 12. In a caseless character class with UCP support, when a character with more
59 than one alternative case was not the first character of a range, not all
60 the alternative cases were added to the class. For example, s and \x{17f}
61 are both alternative cases for S: the class [RST] was handled correctly,
62 but [R-T] was not.
63
64 13. The configure.ac file always checked for pthread support when JIT was
65 enabled. This is not used in Windows, so I have put this test inside a
66 check for the presence of windows.h (which was already tested for).
67
68 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
69 The algorithm provides a way to skip certain starting offsets, and usually
70 faster than linear prefix searches.
71
72 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
73 as for 'fr_FR' and 'french'. For some reason, however, it then used the
74 Windows-specific input and output files, which have 'french' screwed in.
75 So this could never have worked. One of the problems with locales is that
76 they aren't always the same. I have now updated RunTest so that it checks
77 the output of the locale test (test 3) against three different output
78 files, and it allows the test to pass if any one of them matches. With luck
79 this should make the test pass on some versions of Solaris where it was
80 failing. Because of the uncertainty, the script did not used to stop if
81 test 3 failed; it now does. If further versions of a French locale ever
82 come to light, they can now easily be added.
83
84
85 Version 8.34 15-December-2013
86 -----------------------------
87
88 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
89 executable memory. Patch inspired by Carsten Klein.
90
91 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
92 this macro is never tested and has no effect, because the work to support
93 coverage involves only compiling and linking options and special targets in
94 the Makefile. The comment in config.h implied that defining the macro would
95 enable coverage support, which is totally false. There was also support for
96 setting this macro in the CMake files (my fault, I just copied it from
97 configure). SUPPORT_GCOV has now been removed.
98
99 3. Make a small performance improvement in strlen16() and strlen32() in
100 pcretest.
101
102 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
103 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
104 these cases). There was also one in pcretest.c.
105
106 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
107
108 6. In UTF mode, the code for checking whether a group could match an empty
109 string (which is used for indefinitely repeated groups to allow for
110 breaking an infinite loop) was broken when the group contained a repeated
111 negated single-character class with a character that occupied more than one
112 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
113 UTF-8 mode). The effect was undefined: the group might or might not be
114 deemed as matching an empty string, or the program might have crashed.
115
116 7. The code for checking whether a group could match an empty string was not
117 recognizing that \h, \H, \v, \V, and \R must match a character.
118
119 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
120 an empty string. If it can, pcretest shows this in its information output.
121
122 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
123 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
124 matched by pcre_exec() without using JIT:
125
126 (a) If the rest of the pattern did not match after a maximal run of
127 grapheme clusters, the code for backing up to try with fewer of them
128 did not always back up over a full grapheme when characters that do not
129 have the modifier quality were involved, e.g. Hangul syllables.
130
131 (b) If the match point in a subject started with modifier character, and
132 there was no match, the code could incorrectly back up beyond the match
133 point, and potentially beyond the first character in the subject,
134 leading to a segfault or an incorrect match result.
135
136 10. A conditional group with an assertion condition could lead to PCRE
137 recording an incorrect first data item for a match if no other first data
138 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
139 first data item, and therefore matched "ca" after "c" instead of at the
140 start.
141
142 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
143 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
144
145 12. The source of pcregrep now includes z/OS-specific code so that it can be
146 compiled for z/OS as part of the special z/OS distribution.
147
148 13. Added the -T and -TM options to pcretest.
149
150 14. The code in pcre_compile.c for creating the table of named capturing groups
151 has been refactored. Instead of creating the table dynamically during the
152 actual compiling pass, the information is remembered during the pre-compile
153 pass (on the stack unless there are more than 20 named groups, in which
154 case malloc() is used) and the whole table is created before the actual
155 compile happens. This has simplified the code (it is now nearly 150 lines
156 shorter) and prepared the way for better handling of references to groups
157 with duplicate names.
158
159 15. A back reference to a named subpattern when there is more than one of the
160 same name now checks them in the order in which they appear in the pattern.
161 The first one that is set is used for the reference. Previously only the
162 first one was inspected. This change makes PCRE more compatible with Perl.
163
164 16. Unicode character properties were updated from Unicode 6.3.0.
165
166 17. The compile-time code for auto-possessification has been refactored, based
167 on a patch by Zoltan Herczeg. It now happens after instead of during
168 compilation. The code is cleaner, and more cases are handled. The option
169 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
170 options in pcretest are provided to set it. It can also be set by
171 (*NO_AUTO_POSSESS) at the start of a pattern.
172
173 18. The character VT has been added to the default ("C" locale) set of
174 characters that match \s and are generally treated as white space,
175 following this same change in Perl 5.18. There is now no difference between
176 "Perl space" and "POSIX space". Whether VT is treated as white space in
177 other locales depends on the locale.
178
179 19. The code for checking named groups as conditions, either for being set or
180 for being recursed, has been refactored (this is related to 14 and 15
181 above). Processing unduplicated named groups should now be as fast at
182 numerical groups, and processing duplicated groups should be faster than
183 before.
184
185 20. Two patches to the CMake build system, by Alexander Barkov:
186
187 (1) Replace the "source" command by "." in CMakeLists.txt because
188 "source" is a bash-ism.
189
190 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
191 without these the CMake build does not work on Solaris.
192
193 21. Perl has changed its handling of \8 and \9. If there is no previously
194 encountered capturing group of those numbers, they are treated as the
195 literal characters 8 and 9 instead of a binary zero followed by the
196 literals. PCRE now does the same.
197
198 22. Following Perl, added \o{} to specify codepoints in octal, making it
199 possible to specify values greater than 0777 and also making them
200 unambiguous.
201
202 23. Perl now gives an error for missing closing braces after \x{... instead of
203 treating the string as literal. PCRE now does the same.
204
205 24. RunTest used to grumble if an inappropriate test was selected explicitly,
206 but just skip it when running all tests. This make it awkward to run ranges
207 of tests when one of them was inappropriate. Now it just skips any
208 inappropriate tests, as it always did when running all tests.
209
210 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
211 character types such as \d or \w, too many callouts were inserted, and the
212 data that they returned was rubbish.
213
214 26. In UCP mode, \s was not matching two of the characters that Perl matches,
215 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
216 were matched by \h. The code has now been refactored so that the lists of
217 the horizontal and vertical whitespace characters used for \h and \v (which
218 are defined only in one place) are now also used for \s.
219
220 27. Add JIT support for the 64 bit TileGX architecture.
221 Patch by Jiong Wang (Tilera Corporation).
222
223 28. Possessive quantifiers for classes (both explicit and automatically
224 generated) now use special opcodes instead of wrapping in ONCE brackets.
225
226 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
227 (because it's meaningless), this was not happening when PCRE_CASELESS was
228 set. Not wrong, but inefficient.
229
230 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
231 properties for \w, \d, etc) is present in a test regex. Otherwise if the
232 test contains no characters greater than 255, Perl doesn't realise it
233 should be using Unicode semantics.
234
235 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
236 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
237 does in Unicode mode.
238
239 32. Added the "forbid" facility to pcretest so that putting tests into the
240 wrong test files can sometimes be quickly detected.
241
242 33. There is now a limit (default 250) on the depth of nesting of parentheses.
243 This limit is imposed to control the amount of system stack used at compile
244 time. It can be changed at build time by --with-parens-nest-limit=xxx or
245 the equivalent in CMake.
246
247 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
248 errors. Perl warns for these when in warning mode, but PCRE has no facility
249 for giving warnings.
250
251 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
252 However, this was not working for (?!) because it is optimized to (*FAIL),
253 for which PCRE does not allow quantifiers. The optimization is now disabled
254 when a quantifier follows (?!). I can't see any use for this, but it makes
255 things uniform.
256
257 36. Perl no longer allows group names to start with digits, so I have made this
258 change also in PCRE. It simplifies the code a bit.
259
260 37. In extended mode, Perl ignores spaces before a + that indicates a
261 possessive quantifier. PCRE allowed a space before the quantifier, but not
262 before the possessive +. It now does.
263
264 38. The use of \K (reset reported match start) within a repeated possessive
265 group such as (a\Kb)*+ was not working.
266
267 40. Document that the same character tables must be used at compile time and
268 run time, and that the facility to pass tables to pcre_exec() and
269 pcre_dfa_exec() is for use only with saved/restored patterns.
270
271 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
272 features for Builds with MSVC:
273
274 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
275 builds). The .rc files can be used to set FileDescription and many other
276 attributes.
277
278 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
279 This allows higher-level build scripts which want .pdb files to avoid
280 hard-coding the exact files needed."
281
282 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
283 mean "start of word" and "end of word", respectively, as a transition aid.
284
285 43. A minimizing repeat of a class containing codepoints greater than 255 in
286 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
287 compiled to use the heap for recursion.
288
289 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
290 is configured.
291
292
293 Version 8.33 28-May-2013
294 ------------------------
295
296 1. Added 'U' to some constants that are compared to unsigned integers, to
297 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
298 variables that are added to signed variables, to ensure the result is
299 signed and can be negated.
300
301 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
302 CMake config files.
303
304 3. Revise the creation of config.h.generic so that all boolean macros are
305 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
306 overriding via -D on the command line possible.
307
308 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
309 to unsigned int is reported to make a quite noticeable speed difference in
310 a specific Windows environment. Testing on Linux did also appear to show
311 some benefit (and it is clearly not harmful). Also fixed the definition of
312 Xop which should be unsigned.
313
314 5. Related to (4), changing the definition of the intermediate variable cc
315 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
316 improvements.
317
318 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
319 unnecessary spaces.
320
321 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
322 and later.
323
324 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
325
326 9. Optimizing fast_forward_start_bits in JIT.
327
328 10. Adding support for callouts in JIT, and fixing some issues revealed
329 during this work. Namely:
330
331 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
332
333 (b) Minimum length was not checked before the matching is started.
334
335 11. The value of capture_last that is passed to callouts was incorrect in some
336 cases when there was a capture on one path that was subsequently abandoned
337 after a backtrack. Also, the capture_last value is now reset after a
338 recursion, since all captures are also reset in this case.
339
340 12. The interpreter no longer returns the "too many substrings" error in the
341 case when an overflowing capture is in a branch that is subsequently
342 abandoned after a backtrack.
343
344 13. In the pathological case when an offset vector of size 2 is used, pcretest
345 now prints out the matched string after a yield of 0 or 1.
346
347 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
348 Only supported by the JIT compiler at the moment.
349
350 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
351
352 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
353 offset of the starting point of the matching process, provided the offsets
354 vector is large enough.
355
356 17. The \A escape now records a lookbehind value of 1, though its execution
357 does not actually inspect the previous character. This is to ensure that,
358 in partial multi-segment matching, at least one character from the old
359 segment is retained when a new segment is processed. Otherwise, if there
360 are no lookbehinds in the pattern, \A might match incorrectly at the start
361 of a new segment.
362
363 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
364
365 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
366 gives some modest performance improvement in 8-bit mode.
367
368 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
369 be expressed in certain programming languages using Universal Character
370 Names.
371
372 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
373 which points out that "non characters" are not "characters that may not
374 appear in Unicode strings" but rather "characters that are reserved for
375 internal use and have only local meaning".
376
377 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
378 there was a conditional group that depended on an assertion, if the
379 assertion was false, the callout that immediately followed the alternation
380 in the condition was skipped when pcre_exec() was used for matching.
381
382 23. Allow an explicit callout to be inserted before an assertion that is the
383 condition for a conditional group, for compatibility with automatic
384 callouts, which always insert a callout at this point.
385
386 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
387 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
388
389 25. (*PRUNE) is now supported by the JIT compiler.
390
391 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
392
393 27. Fix the case where there are two or more SKIPs with arguments that may be
394 ignored.
395
396 28. (*SKIP) is now supported by the JIT compiler.
397
398 29. (*THEN) is now supported by the JIT compiler.
399
400 30. Update RunTest with additional test selector options.
401
402 31. The way PCRE handles backtracking verbs has been changed in two ways.
403
404 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
405 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
406 backtracking. In some cases this makes it more Perl-compatible, but Perl's
407 rather obscure rules do not always do the same thing.
408
409 (2) Previously, backtracking verbs were confined within assertions. This is
410 no longer the case for positive assertions, except for (*ACCEPT). Again,
411 this sometimes improves Perl compatibility, and sometimes does not.
412
413 32. A number of tests that were in test 2 because Perl did things differently
414 have been moved to test 1, because either Perl or PCRE has changed, and
415 these tests are now compatible.
416
417 32. Backtracking control verbs are now handled in the same way in JIT and
418 interpreter.
419
420 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
421 contained a forward subroutine reference caused a compile error.
422
423 34. Auto-detect and optimize limited repetitions in JIT.
424
425 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
426 blocking (*UTF) etc.
427
428 36. In the interpreter, maximizing pattern repetitions for characters and
429 character types now use tail recursion, which reduces stack usage.
430
431 37. The value of the max lookbehind was not correctly preserved if a compiled
432 and saved regex was reloaded on a host of different endianness.
433
434 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
435 of the compiled pattern block, expand the flags field from 16 to 32 bits
436 because it was almost full.
437
438 39. Try madvise first before posix_madvise.
439
440 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
441 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
442 original change locked out all matches of empty strings. This has been
443 changed so that one match of an empty string per line is recognized.
444 Subsequent searches on the same line (for colouring or for --only-matching,
445 for example) do not recognize empty strings.
446
447 41. Applied a user patch to fix a number of spelling mistakes in comments.
448
449 42. Data lines longer than 65536 caused pcretest to crash.
450
451 43. Clarified the data type for length and startoffset arguments for pcre_exec
452 and pcre_dfa_exec in the function-specific man pages, where they were
453 explicitly stated to be in bytes, never having been updated. I also added
454 some clarification to the pcreapi man page.
455
456 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
457 a segmentation fault.
458
459
460 Version 8.32 30-November-2012
461 -----------------------------
462
463 1. Improved JIT compiler optimizations for first character search and single
464 character iterators.
465
466 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
467 Patch by Daniel Richard G.
468
469 3. Single character iterator optimizations in the JIT compiler.
470
471 4. Improved JIT compiler optimizations for character ranges.
472
473 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
474 Reported by Giuseppe D'Angelo.
475
476 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
477 of a line, was being set incorrectly in cases where .* appeared inside
478 atomic brackets at the start of a pattern, or where there was a subsequent
479 *PRUNE or *SKIP.
480
481 7. Improved instruction cache flush for POWER/PowerPC.
482 Patch by Daniel Richard G.
483
484 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
485 grep:
486
487 (a) There is now no limit to the number of patterns to be matched.
488
489 (b) An error is given if a pattern is too long.
490
491 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
492 are now supported.
493
494 (d) --exclude-from and --include-from (multiple use) have been added.
495
496 (e) Exclusions and inclusions now apply to all files and directories, not
497 just to those obtained from scanning a directory recursively.
498
499 (f) Multiple uses of -f and --file-list are now supported.
500
501 (g) In a Windows environment, the default for -d has been changed from
502 "read" (the GNU grep default) to "skip", because otherwise the presence
503 of a directory in the file list provokes an error.
504
505 (h) The documentation has been revised and clarified in places.
506
507 9. Improve the matching speed of capturing brackets.
508
509 10. Changed the meaning of \X so that it now matches a Unicode extended
510 grapheme cluster.
511
512 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
513 out POSIX threads when JIT support is configured.
514
515 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
516
517 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
518 structure could go wrong in environments where size_t is not the same size
519 as int.
520
521 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
522 set.
523
524 15. The EBCDIC support had decayed; later updates to the code had included
525 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
526 general tidy up of EBCDIC-related issues, and the documentation was also
527 not quite right. There is now a test that can be run on ASCII systems to
528 check some of the EBCDIC-related things (but is it not a full test).
529
530 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
531 in a small tidy to the code.
532
533 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
534
535 18. If the --only-matching (-o) option in pcregrep is specified multiple
536 times, each one causes appropriate output. For example, -o1 -o2 outputs the
537 substrings matched by the 1st and 2nd capturing parentheses. A separating
538 string can be specified by --om-separator (default empty).
539
540 19. Improving the first n character searches.
541
542 20. Turn case lists for horizontal and vertical white space into macros so that
543 they are defined only once.
544
545 21. This set of changes together give more compatible Unicode case-folding
546 behaviour for characters that have more than one other case when UCP
547 support is available.
548
549 (a) The Unicode property table now has offsets into a new table of sets of
550 three or more characters that are case-equivalent. The MultiStage2.py
551 script that generates these tables (the pcre_ucd.c file) now scans
552 CaseFolding.txt instead of UnicodeData.txt for character case
553 information.
554
555 (b) The code for adding characters or ranges of characters to a character
556 class has been abstracted into a generalized function that also handles
557 case-independence. In UTF-mode with UCP support, this uses the new data
558 to handle characters with more than one other case.
559
560 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
561 whose other case is greater than 256 are now correctly matched
562 caselessly. Previously, the high codepoint matched the low one, but not
563 vice versa.
564
565 (d) The processing of \h, \H, \v, and \ in character classes now makes use
566 of the new class addition function, using character lists defined as
567 macros alongside the case definitions of 20 above.
568
569 (e) Caseless back references now work with characters that have more than
570 one other case.
571
572 (f) General caseless matching of characters with more than one other case
573 is supported.
574
575 22. Unicode character properties were updated from Unicode 6.2.0
576
577 23. Improved CMake support under Windows. Patch by Daniel Richard G.
578
579 24. Add support for 32-bit character strings, and UTF-32
580
581 25. Major JIT compiler update (code refactoring and bugfixing).
582 Experimental Sparc 32 support is added.
583
584 26. Applied a modified version of Daniel Richard G's patch to create
585 pcre.h.generic and config.h.generic by "make" instead of in the
586 PrepareRelease script.
587
588 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
589 pcre_compile.c when checking for a zero character.
590
591 28. Introducing a native interface for JIT. Through this interface, the compiled
592 machine code can be directly executed. The purpose of this interface is to
593 provide fast pattern matching, so several sanity checks are not performed.
594 However, feature tests are still performed. The new interface provides
595 1.4x speedup compared to the old one.
596
597 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
598 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
599 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
600
601 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
602 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
603 the "old" RFC 2279). Instead, it ended up passing a negative length to
604 pcre_exec().
605
606 31. Add support for GCC's visibility feature to hide internal functions.
607
608 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
609 "unknown -C option" after outputting 0 or 1.
610
611 33. There is now support for generating a code coverage report for the test
612 suite in environments where gcc is the compiler and lcov is installed. This
613 is mainly for the benefit of the developers.
614
615 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
616 unaddressable using valgrind annotations, allowing valgrind to detect
617 invalid memory accesses. This is mainly for the benefit of the developers.
618
619 25. (*UTF) can now be used to start a pattern in any of the three libraries.
620
621 26. Give configure error if --enable-cpp but no C++ compiler found.
622
623
624 Version 8.31 06-July-2012
625 -------------------------
626
627 1. Fixing a wrong JIT test case and some compiler warnings.
628
629 2. Removed a bashism from the RunTest script.
630
631 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
632 to unsigned type, result still unsigned" that was given by an MS compiler
633 on encountering the code "-sizeof(xxx)".
634
635 4. Partial matching support is added to the JIT compiler.
636
637 5. Fixed several bugs concerned with partial matching of items that consist
638 of more than one character:
639
640 (a) /^(..)\1/ did not partially match "aba" because checking references was
641 done on an "all or nothing" basis. This also applied to repeated
642 references.
643
644 (b) \R did not give a hard partial match if \r was found at the end of the
645 subject.
646
647 (c) \X did not give a hard partial match after matching one or more
648 characters at the end of the subject.
649
650 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
651 a partial match for the string "\r".
652
653 (e) When newline was set to CRLF, the metacharacter "." did not recognize
654 a partial match for a CR character at the end of the subject string.
655
656 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
657 running pcretest, the text "(JIT)" added to the output whenever JIT is
658 actually used to run the match.
659
660 7. Individual JIT compile options can be set in pcretest by following -s+[+]
661 or /S+[+] with a digit between 1 and 7.
662
663 8. OP_NOT now supports any UTF character not just single-byte ones.
664
665 9. (*MARK) control verb is now supported by the JIT compiler.
666
667 10. The command "./RunTest list" lists the available tests without actually
668 running any of them. (Because I keep forgetting what they all are.)
669
670 11. Add PCRE_INFO_MAXLOOKBEHIND.
671
672 12. Applied a (slightly modified) user-supplied patch that improves performance
673 when the heap is used for recursion (compiled with --disable-stack-for-
674 recursion). Instead of malloc and free for each heap frame each time a
675 logical recursion happens, frames are retained on a chain and re-used where
676 possible. This sometimes gives as much as 30% improvement.
677
678 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
679 call.
680
681 14. As documented, (*COMMIT) is now confined to within a positive assertion.
682
683 15. It is now possible to link pcretest with libedit as an alternative to
684 libreadline.
685
686 16. (*COMMIT) control verb is now supported by the JIT compiler.
687
688 17. The Unicode data tables have been updated to Unicode 6.1.0.
689
690 18. Added --file-list option to pcregrep.
691
692 19. Added binary file support to pcregrep, including the -a, --binary-files,
693 -I, and --text options.
694
695 20. The madvise function is renamed for posix_madvise for QNX compatibility
696 reasons. Fixed by Giuseppe D'Angelo.
697
698 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
699
700 22. Changed the option for creating symbolic links for 16-bit man pages from
701 -s to -sf so that re-installing does not cause issues.
702
703 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
704
705 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
706 matches in certain environments (the workspace was not being correctly
707 retained). Also added to pcre_dfa_exec() a simple plausibility check on
708 some of the workspace data at the beginning of a restart.
709
710 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
711 was not doing so when it should - probably a typo introduced by SVN 528
712 (change 8.10/14).
713
714 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
715 \w+ when the character tables indicated that \x{c4} was a word character.
716 There were several related cases, all because the tests for doing a table
717 lookup were testing for characters less than 127 instead of 255.
718
719 27. If a pattern contains capturing parentheses that are not used in a match,
720 their slots in the ovector are set to -1. For those that are higher than
721 any matched groups, this happens at the end of processing. In the case when
722 there were back references that the ovector was too small to contain
723 (causing temporary malloc'd memory to be used during matching), and the
724 highest capturing number was not used, memory off the end of the ovector
725 was incorrectly being set to -1. (It was using the size of the temporary
726 memory instead of the true size.)
727
728 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
729 ovector size, it uses memory at the end of the block that it has got.
730
731 29. Check for an overlong MARK name and give an error at compile time. The
732 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
733
734 30. JIT compiler update.
735
736 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
737 Rill for the patch.
738
739 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
740
741 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
742
743 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
744 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
745 was enabled.
746
747 35. Improve JIT code generation for greedy plus quantifier.
748
749 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
750 1 to "aa" instead of to an empty string. The bug affected repeated groups
751 that could potentially match an empty string.
752
753 37. Optimizing single character iterators in JIT.
754
755 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
756 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
757 codepoints that are too big for the mode are faulted, and in a UTF mode,
758 disallowed codepoints are also faulted.
759
760 39. If PCRE was compiled with UTF support, in three places in the DFA
761 matcher there was code that should only have been obeyed in UTF mode, but
762 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
763 processing when bytes with values greater than 127 were present. In 16-bit
764 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
765 both cases the values are those that cannot be the first data item in a UTF
766 character. The three items that might have provoked this were recursions,
767 possessively repeated groups, and atomic groups.
768
769 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
770 and pcregrep, because some OS require shared objects to be explicitly
771 passed to ld, causing the link step to fail if they are not.
772
773 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
774 mode, patterns that started with \h* or \R* might be incorrectly matched.
775
776
777 Version 8.30 04-February-2012
778 -----------------------------
779
780 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
781 name is defined in ctype.h.
782
783 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
784 only in quite long subpatterns.
785
786 3. Removed the function pcre_info(), which has been obsolete and deprecated
787 since it was replaced by pcre_fullinfo() in February 2000.
788
789 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
790 match a (*MARK), and the match failed at the start of the subject, a
791 reference to memory before the start of the subject could occur. This bug
792 was introduced by fix 17 of release 8.21.
793
794 5. A reference to an unset group with zero minimum repetition was giving
795 totally wrong answers (in non-JavaScript-compatibility mode). For example,
796 /(another)?(\1?)test/ matched against "hello world test". This bug was
797 introduced in release 8.13.
798
799 6. Add support for 16-bit character strings (a large amount of work involving
800 many changes and refactorings).
801
802 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
803 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
804 from a file.
805
806 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
807 rounding is not applied in this particular case).
808
809 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
810 if they appear, or are escaped, in patterns.
811
812 10. Get rid of a number of -Wunused-but-set-variable warnings.
813
814 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
815 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
816 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
817 also returns the mark "x". This bug applied to capturing parentheses,
818 non-capturing parentheses, and atomic parentheses. It also applied to some
819 assertions.
820
821 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
822 information out of configure.ac instead of relying on pcre.h.generic, which
823 is not stored in the repository.
824
825 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
826 -lreadline.
827
828 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
829
830 15. Applied Graycode's patch to put the top-level frame on the stack rather
831 than the heap when not using the stack for recursion. This gives a
832 performance improvement in many cases when recursion is not deep.
833
834 16. Experimental code added to "pcretest -C" to output the stack frame size.
835
836
837 Version 8.21 12-Dec-2011
838 ------------------------
839
840 1. Updating the JIT compiler.
841
842 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
843 are added as well.
844
845 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
846 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
847 calling _pcre_jit_exec. Some extra comments are added.
848
849 4. (*MARK) settings inside atomic groups that do not contain any capturing
850 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
851 was introduced by change 18 for 8.20.
852
853 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
854 ECMA-262 standard.
855
856 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
857 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
858 This bug was probably introduced by change 9 of 8.13.
859
860 7. While fixing 6 above, I noticed that a number of other items were being
861 incorrectly rejected as "not fixed length". This arose partly because newer
862 opcodes had not been added to the fixed-length checking code. I have (a)
863 corrected the bug and added tests for these items, and (b) arranged for an
864 error to occur if an unknown opcode is encountered while checking for fixed
865 length instead of just assuming "not fixed length". The items that were
866 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
867 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
868 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
869
870 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
871 being incorrectly compiled and would have given unpredicatble results.
872
873 9. A possessively repeated subpattern with minimum repeat count greater than
874 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
875 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
876 the first (A) could occur when it should not.
877
878 10. Add a cast and remove a redundant test from the code.
879
880 11. JIT should use pcre_malloc/pcre_free for allocation.
881
882 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
883 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
884 is anything other than /usr, -L is still shown).
885
886 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
887
888 14. Perl does not support \N without a following name in a [] class; PCRE now
889 also gives an error.
890
891 15. If a forward reference was repeated with an upper limit of around 2000,
892 it caused the error "internal error: overran compiling workspace". The
893 maximum number of forward references (including repeats) was limited by the
894 internal workspace, and dependent on the LINK_SIZE. The code has been
895 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
896 the default depends on LINK_SIZE. There is a new upper limit (for safety)
897 of around 200,000 forward references. While doing this, I also speeded up
898 the filling in of repeated forward references.
899
900 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
901 incorrectly expecting the subject to contain another "a" after the start.
902
903 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
904 in the match, the SKIP should be ignored. This was not happening; instead
905 the SKIP was being treated as NOMATCH. For patterns such as
906 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
907 tested.
908
909 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
910 now much more compatible with Perl, in particular in cases where the result
911 is a non-match for a non-anchored pattern. For example, if
912 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
913 "m", where previously it did not return a name. A side effect of this
914 change is that for partial matches, the last encountered mark name is
915 returned, as for non matches. A number of tests that were previously not
916 Perl-compatible have been moved into the Perl-compatible test files. The
917 refactoring has had the pleasing side effect of removing one argument from
918 the match() function, thus reducing its stack requirements.
919
920 19. If the /S+ option was used in pcretest to study a pattern using JIT,
921 subsequent uses of /S (without +) incorrectly behaved like /S+.
922
923 21. Retrieve executable code size support for the JIT compiler and fixing
924 some warnings.
925
926 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
927 not work when the shorter character appeared right at the end of the
928 subject string.
929
930 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
931 systems.
932
933 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
934 output it when the /M option is used in pcretest.
935
936 25. The CheckMan script was not being included in the distribution. Also, added
937 an explicit "perl" to run Perl scripts from the PrepareRelease script
938 because this is reportedly needed in Windows.
939
940 26. If study data was being save in a file and studying had not found a set of
941 "starts with" bytes for the pattern, the data written to the file (though
942 never used) was taken from uninitialized memory and so caused valgrind to
943 complain.
944
945 27. Updated RunTest.bat as provided by Sheri Pierce.
946
947 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
948
949 29. Computation of memory usage for the table of capturing group names was
950 giving an unnecessarily large value.
951
952
953 Version 8.20 21-Oct-2011
954 ------------------------
955
956 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
957 a POSIX class. After further experiments with Perl, which convinced me that
958 Perl has bugs and confusions, a closing square bracket is no longer allowed
959 in a POSIX name. This bug also affected patterns with classes that started
960 with full stops.
961
962 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
963 captured substring, but while checking the failing first alternative,
964 substring 1 is temporarily captured. If the output vector supplied to
965 pcre_exec() was not big enough for this capture, the yield of the function
966 was still zero ("insufficient space for captured substrings"). This cannot
967 be totally fixed without adding another stack variable, which seems a lot
968 of expense for a edge case. However, I have improved the situation in cases
969 such as /(a)(b)x|abc/ matched against "abc", where the return code
970 indicates that fewer than the maximum number of slots in the ovector have
971 been set.
972
973 3. Related to (2) above: when there are more back references in a pattern than
974 slots in the output vector, pcre_exec() uses temporary memory during
975 matching, and copies in the captures as far as possible afterwards. It was
976 using the entire output vector, but this conflicts with the specification
977 that only 2/3 is used for passing back captured substrings. Now it uses
978 only the first 2/3, for compatibility. This is, of course, another edge
979 case.
980
981 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
982 main code base, and can be used by building with --enable-jit. When this is
983 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
984 runtime --no-jit option is given.
985
986 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
987 ovector, the return from the function was zero, implying that there were
988 other matches that did not fit. The correct "exactly full" value is now
989 returned.
990
991 6. If a subpattern that was called recursively or as a subroutine contained
992 (*PRUNE) or any other control that caused it to give a non-standard return,
993 invalid errors such as "Error -26 (nested recursion at the same subject
994 position)" or even infinite loops could occur.
995
996 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
997 computing the minimum length on reaching *ACCEPT, and so ended up with the
998 wrong value of 1 rather than 0. Further investigation indicates that
999 computing a minimum subject length in the presence of *ACCEPT is difficult
1000 (think back references, subroutine calls), and so I have changed the code
1001 so that no minimum is registered for a pattern that contains *ACCEPT.
1002
1003 8. If (*THEN) was present in the first (true) branch of a conditional group,
1004 it was not handled as intended. [But see 16 below.]
1005
1006 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1007 Sheri Pierce.
1008
1009 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1010 the first byte in a match must be "a".
1011
1012 11. Change 17 for 8.13 increased the recursion depth for patterns like
1013 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1014 pattern contains any instances of (*THEN). If it does not, the old
1015 optimizations are restored. It would be nice to do this on a per-group
1016 basis, but at the moment that is not feasible.
1017
1018 12. In some environments, the output of pcretest -C is CRLF terminated. This
1019 broke RunTest's code that checks for the link size. A single white space
1020 character after the value is now allowed for.
1021
1022 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1023 For "fr", it uses the Windows-specific input and output files.
1024
1025 14. If (*THEN) appeared in a group that was called recursively or as a
1026 subroutine, it did not work as intended. [But see next item.]
1027
1028 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1029 pattern fragments (but not containing any | characters). If A and B are
1030 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1031 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1032 D. In other words, Perl considers parentheses that do not contain any |
1033 characters to be part of a surrounding alternative, whereas PCRE was
1034 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1035 differently. PCRE now behaves in the same way as Perl, except in the case
1036 of subroutine/recursion calls such as (?1) which have in any case always
1037 been different (but PCRE had them first :-).
1038
1039 16. Related to 15 above: Perl does not treat the | in a conditional group as
1040 creating alternatives. Such a group is treated in the same way as an
1041 ordinary group without any | characters when processing (*THEN). PCRE has
1042 been changed to match Perl's behaviour.
1043
1044 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1045 RunGrepTest script failed.
1046
1047 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1048 inevitable for groups that contain captures, but it can lead to a lot of
1049 stack use in large patterns. The old behaviour has been restored for atomic
1050 groups that do not contain any capturing parentheses.
1051
1052 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1053 suppress the check for a minimum subject length at run time. (If it was
1054 given to pcre_exec() or pcre_dfa_exec() it did work.)
1055
1056 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1057 fail to work when decoding hex characters in data strings in EBCDIC
1058 environments.
1059
1060 21. It appears that in at least one Mac OS environment, the isxdigit() function
1061 is implemented as a macro that evaluates to its argument more than once,
1062 contravening the C 90 Standard (I haven't checked a later standard). There
1063 was an instance in pcretest which caused it to go wrong when processing
1064 \x{...} escapes in subject strings. The has been rewritten to avoid using
1065 things like p++ in the argument of isxdigit().
1066
1067
1068 Version 8.13 16-Aug-2011
1069 ------------------------
1070
1071 1. The Unicode data tables have been updated to Unicode 6.0.0.
1072
1073 2. Two minor typos in pcre_internal.h have been fixed.
1074
1075 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1076 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1077 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1078
1079 4. There were a number of related bugs in the code for matching backrefences
1080 caselessly in UTF-8 mode when codes for the characters concerned were
1081 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1082 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1083 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1084 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1085 2-byte code at the end of the subject (it thought there wasn't enough data
1086 left).
1087
1088 5. Comprehensive information about what went wrong is now returned by
1089 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1090 as the output vector has at least 2 elements. The offset of the start of
1091 the failing character and a reason code are placed in the vector.
1092
1093 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1094 now returned is for the first byte of the failing character, instead of the
1095 last byte inspected. This is an incompatible change, but I hope it is small
1096 enough not to be a problem. It makes the returned offset consistent with
1097 pcre_exec() and pcre_dfa_exec().
1098
1099 7. pcretest now gives a text phrase as well as the error number when
1100 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1101 failure, the offset and reason code are output.
1102
1103 8. When \R was used with a maximizing quantifier it failed to skip backwards
1104 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1105 back over a single character (\n). This seems wrong (because it treated the
1106 two characters as a single entity when going forwards), conflicts with the
1107 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1108 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1109 has been changed.
1110
1111 9. Some internal refactoring has changed the processing so that the handling
1112 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1113 time (the PCRE_DOTALL option was changed this way some time ago: version
1114 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1115 which was always a bit of a fudge. It also means that there is one less
1116 argument for the match() function, which reduces its stack requirements
1117 slightly. This change also fixes an incompatibility with Perl: the pattern
1118 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1119
1120 10. More internal refactoring has drastically reduced the number of recursive
1121 calls to match() for possessively repeated groups such as (abc)++ when
1122 using pcre_exec().
1123
1124 11. While implementing 10, a number of bugs in the handling of groups were
1125 discovered and fixed:
1126
1127 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1128 (a|)*(?1) gave a compile-time internal error.
1129 ((a|)+)+ did not notice that the outer group could match an empty string.
1130 (^a|^)+ was not marked as anchored.
1131 (.*a|.*)+ was not marked as matching at start or after a newline.
1132
1133 12. Yet more internal refactoring has removed another argument from the match()
1134 function. Special calls to this function are now indicated by setting a
1135 value in a variable in the "match data" data block.
1136
1137 13. Be more explicit in pcre_study() instead of relying on "default" for
1138 opcodes that mean there is no starting character; this means that when new
1139 ones are added and accidentally left out of pcre_study(), testing should
1140 pick them up.
1141
1142 14. The -s option of pcretest has been documented for ages as being an old
1143 synonym of -m (show memory usage). I have changed it to mean "force study
1144 for every regex", that is, assume /S for every regex. This is similar to -i
1145 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1146 using it. It makes it easier to run collections of tests with and without
1147 study enabled, and thereby test pcre_study() more easily. All the standard
1148 tests are now run with and without -s (but some patterns can be marked as
1149 "never study" - see 20 below).
1150
1151 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1152 restoration of the capturing data to the outer values was not happening
1153 correctly.
1154
1155 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1156 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1157 pattern had matched an empty string, and so incorrectly returned a no
1158 match.
1159
1160 17. There was optimizing code for the last branch of non-capturing parentheses,
1161 and also for the obeyed branch of a conditional subexpression, which used
1162 tail recursion to cut down on stack usage. Unfortunately, now that there is
1163 the possibility of (*THEN) occurring in these branches, tail recursion is
1164 no longer possible because the return has to be checked for (*THEN). These
1165 two optimizations have therefore been removed. [But see 8.20/11 above.]
1166
1167 18. If a pattern containing \R was studied, it was assumed that \R always
1168 matched two bytes, thus causing the minimum subject length to be
1169 incorrectly computed because \R can also match just one byte.
1170
1171 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1172 was incorrectly computed.
1173
1174 20. If /S is present twice on a test pattern in pcretest input, it now
1175 *disables* studying, thereby overriding the use of -s on the command line
1176 (see 14 above). This is necessary for one or two tests to keep the output
1177 identical in both cases.
1178
1179 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1180 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1181
1182 22. When an atomic group that contained a capturing parenthesis was
1183 successfully matched, but the branch in which it appeared failed, the
1184 capturing was not being forgotten if a higher numbered group was later
1185 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1186 group 1 to "a", when in fact it should be unset. This applied to multi-
1187 branched capturing and non-capturing groups, repeated or not, and also to
1188 positive assertions (capturing in negative assertions does not happen
1189 in PCRE) and also to nested atomic groups.
1190
1191 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1192 subject after a captured substring, to make it easier to tell which of a
1193 number of identical substrings has been captured.
1194
1195 24. The way atomic groups are processed by pcre_exec() has been changed so that
1196 if they are repeated, backtracking one repetition now resets captured
1197 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1198 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1199 "aaa". Previously, it would have been "a". As part of this code
1200 refactoring, the way recursive calls are handled has also been changed.
1201
1202 25. If an assertion condition captured any substrings, they were not passed
1203 back unless some other capturing happened later. For example, if
1204 (?(?=(a))a) was matched against "a", no capturing was returned.
1205
1206 26. When studying a pattern that contained subroutine calls or assertions,
1207 the code for finding the minimum length of a possible match was handling
1208 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1209 group 1 called group 2 while simultaneously a separate group 2 called group
1210 1). A stack overflow occurred in this case. I have fixed this by limiting
1211 the recursion depth to 10.
1212
1213 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1214 Fortmann. This supports explicit test numbers on the command line, and has
1215 argument validation and error reporting.
1216
1217 28. An instance of \X with an unlimited repeat could fail if at any point the
1218 first character it looked at was a mark character.
1219
1220 29. Some minor code refactoring concerning Unicode properties and scripts
1221 should reduce the stack requirement of match() slightly.
1222
1223 30. Added the '=' option to pcretest to check the setting of unused capturing
1224 slots at the end of the pattern, which are documented as being -1, but are
1225 not included in the return count.
1226
1227 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1228 compiled something random. Now it gives a compile-time error (as does
1229 Perl).
1230
1231 32. A *MARK encountered during the processing of a positive assertion is now
1232 recorded and passed back (compatible with Perl).
1233
1234 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1235 had alternative anchored branches, the search for a second match in a line
1236 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1237 matched the line "0102" twice. The same bug affected patterns that started
1238 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1239 twice.
1240
1241 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1242 does, and because of capturing effects, quantifying parenthesized
1243 assertions may at times be useful. Quantifiers are now allowed for
1244 parenthesized assertions.
1245
1246 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1247
1248 36. \g was being checked for fancy things in a character class, when it should
1249 just be a literal "g".
1250
1251 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1252 appearance of a nested POSIX class supersedes an apparent external class.
1253 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1254 unescaped square brackets may also appear as part of class names. For
1255 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1256 more like Perl. (But see 8.20/1 above.)
1257
1258 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1259 was because it thought it was \N{name}, which is not supported).
1260
1261 39. Add minix to OS list not supporting the -S option in pcretest.
1262
1263 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1264 cannot analyze patterns in sufficient detail to catch mutual recursions
1265 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1266 subgroup is called recursively as a subpattern for a second time at the
1267 same position in the subject string. In previous releases this might have
1268 been caught by the recursion limit, or it might have run out of stack.
1269
1270 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1271 happen only once. PCRE was, however incorrectly giving a compile time error
1272 "recursive call could loop indefinitely" because it cannot analyze the
1273 pattern in sufficient detail. The compile time test no longer happens when
1274 PCRE is compiling a conditional subpattern, but actual runaway loops are
1275 now caught at runtime (see 40 above).
1276
1277 42. It seems that Perl allows any characters other than a closing parenthesis
1278 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1279 has been changed to be the same.
1280
1281 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1282 as not to get warnings when autogen.sh is called. Also changed
1283 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1284
1285 44. To help people who use pcregrep to scan files containing exceedingly long
1286 lines, the following changes have been made:
1287
1288 (a) The default value of the buffer size parameter has been increased from
1289 8K to 20K. (The actual buffer used is three times this size.)
1290
1291 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1292 PCRE is built.
1293
1294 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1295 to be set at run time.
1296
1297 (d) Numerical values in pcregrep options can be followed by K or M, for
1298 example --buffer-size=50K.
1299
1300 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1301 given and the return code is set to 2.
1302
1303 45. Add a pointer to the latest mark to the callout data block.
1304
1305 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1306 partial match of an empty string instead of no match. This was specific to
1307 the use of ".".
1308
1309 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1310 complete match instead of a partial match. This bug was dependent on both
1311 the PCRE_UTF8 and PCRE_DOTALL options being set.
1312
1313 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1314 starting byte set, because \b was not being ignored.
1315
1316
1317 Version 8.12 15-Jan-2011
1318 ------------------------
1319
1320 1. Fixed some typos in the markup of the man pages, and wrote a script that
1321 checks for such things as part of the documentation building process.
1322
1323 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1324 --match-limit and --recursion-limit options (added for 8.11). In
1325 particular, this made one of the standard tests fail. (The integer value
1326 went into the wrong half of a long int.)
1327
1328 3. If the --colour option was given to pcregrep with -v (invert match), it
1329 did strange things, either producing crazy output, or crashing. It should,
1330 of course, ignore a request for colour when reporting lines that do not
1331 match.
1332
1333 4. Another pcregrep bug caused similar problems if --colour was specified with
1334 -M (multiline) and the pattern match finished with a line ending.
1335
1336 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1337 matched in multiline mode, the following line was shown as part of the
1338 match. This seems wrong, so I have changed it.
1339
1340 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1341 the check for further matches in the same line (so they could be coloured)
1342 to overrun the end of the current line. If another match was found, it was
1343 incorrectly shown (and then shown again when found in the next line).
1344
1345 7. If pcregrep was compiled under Windows, there was a reference to the
1346 function pcregrep_exit() before it was defined. I am assuming this was
1347 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1348 reported by a user. I've moved the definition above the reference.
1349
1350
1351 Version 8.11 10-Dec-2010
1352 ------------------------
1353
1354 1. (*THEN) was not working properly if there were untried alternatives prior
1355 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1356 backtracked to try for "b" instead of moving to the next alternative branch
1357 at the same level (in this case, to look for "c"). The Perl documentation
1358 is clear that when (*THEN) is backtracked onto, it goes to the "next
1359 alternative in the innermost enclosing group".
1360
1361 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1362 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1363 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1364 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1365 (*THEN).
1366
1367 3. If \s appeared in a character class, it removed the VT character from
1368 the class, even if it had been included by some previous item, for example
1369 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1370 of \s, but is part of the POSIX "space" class.)
1371
1372 4. A partial match never returns an empty string (because you can always
1373 match an empty string at the end of the subject); however the checking for
1374 an empty string was starting at the "start of match" point. This has been
1375 changed to the "earliest inspected character" point, because the returned
1376 data for a partial match starts at this character. This means that, for
1377 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1378 (previously it gave "no match").
1379
1380 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1381 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1382 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1383 has an implication that the given string is incomplete (because a partial
1384 match is preferred over a full match). For this reason, these items now
1385 give a partial match in this situation. [Aside: previously, the one case
1386 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1387 match rather than a full match, which was wrong by the old rules, but is
1388 now correct.]
1389
1390 6. There was a bug in the handling of #-introduced comments, recognized when
1391 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1392 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1393 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1394 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1395 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1396 places in pcre_compile().
1397
1398 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1399 comments when looking ahead for named forward references to subpatterns,
1400 the only newline sequence it recognized was NL. It now handles newlines
1401 according to the set newline convention.
1402
1403 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1404 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1405 cater for a lack of strerror(). These oversights have been fixed.
1406
1407 9. Added --match-limit and --recursion-limit to pcregrep.
1408
1409 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1410
1411 11. When the -o option was used, pcregrep was setting a return code of 1, even
1412 when matches were found, and --line-buffered was not being honoured.
1413
1414 12. Added an optional parentheses number to the -o and --only-matching options
1415 of pcregrep.
1416
1417 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1418 can match an empty string. The code to do it in pcretest and pcredemo
1419 needed fixing:
1420
1421 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1422 only one byte after an empty string match just before CRLF (this case
1423 just got forgotten; "any" and "anycrlf" were OK).
1424
1425 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1426 mode when an empty string match preceded an ASCII character followed by
1427 a non-ASCII character. (The code for advancing by one character rather
1428 than one byte was nonsense.)
1429
1430 (c) The pcredemo.c sample program did not have any code at all to handle
1431 the cases when CRLF is a valid newline sequence.
1432
1433 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1434 as a starting offset was within the subject string. There is now a new
1435 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1436 negative or greater than the length of the string. In order to test this,
1437 pcretest is extended to allow the setting of negative starting offsets.
1438
1439 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1440 starting offset points to the beginning of a UTF-8 character was
1441 unnecessarily clumsy. I tidied it up.
1442
1443 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1444 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1445
1446 17. Nobody had reported that the --include_dir option, which was added in
1447 release 7.7 should have been called --include-dir (hyphen, not underscore)
1448 for compatibility with GNU grep. I have changed it to --include-dir, but
1449 left --include_dir as an undocumented synonym, and the same for
1450 --exclude-dir, though that is not available in GNU grep, at least as of
1451 release 2.5.4.
1452
1453 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1454 characters from a string of bytes) have been redefined so as not to use
1455 loops, in order to improve performance in some environments. At the same
1456 time, I abstracted some of the common code into auxiliary macros to save
1457 repetition (this should not affect the compiled code).
1458
1459 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1460 compile-time error is now given if \c is not followed by an ASCII
1461 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1462 different, and any byte value is allowed.)
1463
1464 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1465 START_OPTIMIZE option, which is now allowed at compile time - but just
1466 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1467 to pcregrep and other applications that have no direct access to PCRE
1468 options. The new /Y option in pcretest sets this option when calling
1469 pcre_compile().
1470
1471 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1472 back references. Groups containing recursive back references were forced to
1473 be atomic by that change, but in the case of named groups, the amount of
1474 memory required was incorrectly computed, leading to "Failed: internal
1475 error: code overflow". This has been fixed.
1476
1477 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1478 pcretest.c, to avoid build problems in some Borland environments.
1479
1480
1481 Version 8.10 25-Jun-2010
1482 ------------------------
1483
1484 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1485 THEN.
1486
1487 2. (*ACCEPT) was not working when inside an atomic group.
1488
1489 3. Inside a character class, \B is treated as a literal by default, but
1490 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1491 causes the error). The code is unchanged, but I tidied the documentation.
1492
1493 4. Inside a character class, PCRE always treated \R and \X as literals,
1494 whereas Perl faults them if its -w option is set. I have changed PCRE so
1495 that it faults them when PCRE_EXTRA is set.
1496
1497 5. Added support for \N, which always matches any character other than
1498 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1499
1500 6. When compiling pcregrep with newer versions of gcc which may have
1501 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1502 declared with attribute warn_unused_result" were given. Just casting the
1503 result to (void) does not stop the warnings; a more elaborate fudge is
1504 needed. I've used a macro to implement this.
1505
1506 7. Minor change to pcretest.c to avoid a compiler warning.
1507
1508 8. Added four artifical Unicode properties to help with an option to make
1509 \s etc use properties (see next item). The new properties are: Xan
1510 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1511
1512 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1513 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1514 this option. Modified pcretest to add /W to test this facility. Added
1515 REG_UCP to make it available via the POSIX interface.
1516
1517 10. Added --line-buffered to pcregrep.
1518
1519 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1520 studied, and the match started with a letter with a code point greater than
1521 127 whose first byte was different to the first byte of the other case of
1522 the letter, the other case of this starting letter was not recognized
1523 (#976).
1524
1525 12. If a pattern that was studied started with a repeated Unicode property
1526 test, for example, \p{Nd}+, there was the theoretical possibility of
1527 setting up an incorrect bitmap of starting bytes, but fortunately it could
1528 not have actually happened in practice until change 8 above was made (it
1529 added property types that matched character-matching opcodes).
1530
1531 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1532 possible starting bytes for non-anchored patterns.
1533
1534 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1535 \R, and also a number of cases that involve Unicode properties, both
1536 explicit and implicit when PCRE_UCP is set.
1537
1538 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1539 input, it could crash or give wrong results if characters with values
1540 greater than 0xc0 were present in the subject string. (Detail: it assumed
1541 UTF-8 input when processing these items.)
1542
1543 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1544 size_t is 64-bit (#991).
1545
1546 17. Added a check for running out of memory when PCRE is compiled with
1547 --disable-stack-for-recursion (#990).
1548
1549 18. If the last data line in a file for pcretest does not have a newline on
1550 the end, a newline was missing in the output.
1551
1552 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1553 less than 128) in its various bitmaps. However, there is a facility for
1554 generating tables according to the current locale when PCRE is compiled. It
1555 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1556 characters, are recognized by isspace() and therefore were getting set in
1557 these tables, and indeed these tables seem to approximate to ISO 8859. This
1558 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1559 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1560 which of course cannot start UTF-8 characters. I have changed the code so
1561 that only real ASCII characters (less than 128) and the correct starting
1562 bytes for UTF-8 encodings are set for characters greater than 127 when in
1563 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1564 altogether.)
1565
1566 20. Added the /T option to pcretest so as to be able to run tests with non-
1567 standard character tables, thus making it possible to include the tests
1568 used for 19 above in the standard set of tests.
1569
1570 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1571 reference to a subpattern the other side of a comment that contains an
1572 opening parenthesis caused either an internal compiling error, or a
1573 reference to the wrong subpattern.
1574
1575
1576 Version 8.02 19-Mar-2010
1577 ------------------------
1578
1579 1. The Unicode data tables have been updated to Unicode 5.2.0.
1580
1581 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1582 configured.
1583
1584 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1585 original author of that file, following a query about its status.
1586
1587 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1588 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1589
1590 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1591 quantifier applied to a forward-referencing subroutine call, could compile
1592 incorrect code or give the error "internal error: previously-checked
1593 referenced subpattern not found".
1594
1595 6. Both MS Visual Studio and Symbian OS have problems with initializing
1596 variables to point to external functions. For these systems, therefore,
1597 pcre_malloc etc. are now initialized to local functions that call the
1598 relevant global functions.
1599
1600 7. There were two entries missing in the vectors called coptable and poptable
1601 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1602 I've fixed the data, and added a kludgy way of testing at compile time that
1603 the lengths are correct (equal to the number of opcodes).
1604
1605 8. Following on from 7, I added a similar kludge to check the length of the
1606 eint vector in pcreposix.c.
1607
1608 9. Error texts for pcre_compile() are held as one long string to avoid too
1609 much relocation at load time. To find a text, the string is searched,
1610 counting zeros. There was no check for running off the end of the string,
1611 which could happen if a new error number was added without updating the
1612 string.
1613
1614 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1615
1616 11. \K was not working if it appeared in an atomic group or in a group that
1617 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1618 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1619 the assertion is positive, but not if it is negative.
1620
1621 12. Change 11 fortuitously reduced the size of the stack frame used in the
1622 "match()" function of pcre_exec.c by one pointer. Forthcoming
1623 implementation of support for (*MARK) will need an extra pointer on the
1624 stack; I have reserved it now, so that the stack frame size does not
1625 decrease.
1626
1627 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1628 item in branch that calls a recursion is a subroutine call - as in the
1629 second branch in the above example - was incorrectly given the compile-
1630 time error "recursive call could loop indefinitely" because pcre_compile()
1631 was not correctly checking the subroutine for matching a non-empty string.
1632
1633 14. The checks for overrunning compiling workspace could trigger after an
1634 overrun had occurred. This is a "should never occur" error, but it can be
1635 triggered by pathological patterns such as hundreds of nested parentheses.
1636 The checks now trigger 100 bytes before the end of the workspace.
1637
1638 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1639
1640
1641 Version 8.01 19-Jan-2010
1642 ------------------------
1643
1644 1. If a pattern contained a conditional subpattern with only one branch (in
1645 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1646 computed the wrong minimum data length (which is of course zero for such
1647 subpatterns). This could cause incorrect "no match" results.
1648
1649 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1650 the pattern is reset in the first branch, pcre_compile() failed with
1651 "internal error: code overflow at offset...". This happened only when
1652 the reset was to the original external option setting. (An optimization
1653 abstracts leading options settings into an external setting, which was the
1654 cause of this.)
1655
1656 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1657 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1658 assertion pattern did not match (meaning that the assertion was true), it
1659 was incorrectly treated as false if the SKIP had been reached during the
1660 matching. This also applied to assertions used as conditions.
1661
1662 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1663 assertion subpattern, including such a pattern used as a condition,
1664 unpredictable results occurred, instead of the error return
1665 PCRE_ERROR_DFA_UITEM.
1666
1667 5. The C++ GlobalReplace function was not working like Perl for the special
1668 situation when an empty string is matched. It now does the fancy magic
1669 stuff that is necessary.
1670
1671 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1672 removed. (These were left over from very, very early versions of PCRE.)
1673
1674 7. Some cosmetic changes to the code to make life easier when compiling it
1675 as part of something else:
1676
1677 (a) Change DEBUG to PCRE_DEBUG.
1678
1679 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1680 called "current" as "current_branch", to prevent a collision with the
1681 Linux macro when compiled as a kernel module.
1682
1683 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1684 prevent a collision with the Linux macro when compiled as a kernel
1685 module.
1686
1687 8. In pcre_compile() there are some checks for integer overflows that used to
1688 cast potentially large values to (double). This has been changed to that
1689 when building, a check for int64_t is made, and if it is found, it is used
1690 instead, thus avoiding the use of floating point arithmetic. (There is no
1691 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1692 double.
1693
1694 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1695 2005 (difference between two addresses compared to an unsigned value).
1696
1697 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1698 custom one, because of the following reported problem in Windows:
1699
1700 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1701 under Win32.
1702 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1703 therefore missing the function definition.
1704 - The compiler thus generates a "C" signature for the test function.
1705 - The linker fails to find the "C" function.
1706 - PCRE fails to configure if asked to do so against libbz2.
1707
1708 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1709 messages were output:
1710
1711 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1712 rerunning libtoolize, to keep the correct libtool macros in-tree.
1713 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1714
1715 I have done both of these things.
1716
1717 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1718 most of the time, it *can* run out if it is given a pattern that contains a
1719 runaway infinite recursion. I updated the discussion in the pcrestack man
1720 page.
1721
1722 13. Now that we have gone to the x.xx style of version numbers, the minor
1723 version may start with zero. Using 08 or 09 is a bad idea because users
1724 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1725 interpreted as invalid octal numbers. I've updated the previous comment in
1726 configure.ac, and also added a check that gives an error if 08 or 09 are
1727 used.
1728
1729 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1730 causing partial matching to fail when the end of the subject matched \W
1731 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1732
1733 15. There were some discrepancies between the declarations in pcre_internal.h
1734 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1735 their definitions. The declarations used "const uschar *" and the
1736 definitions used USPTR. Even though USPTR is normally defined as "const
1737 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1738 reported that: "This difference in casting confuses some C++ compilers, for
1739 example, SunCC recognizes above declarations as different functions and
1740 generates broken code for hbpcre." I have changed the declarations to use
1741 USPTR.
1742
1743 16. GNU libtool is named differently on some systems. The autogen.sh script now
1744 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1745 (FreeBSD).
1746
1747 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1748 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1749 comment: "Figure out how to create a longlong from a string: strtoll and
1750 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1751 instance, but it only takes 2 args instead of 3!"
1752
1753 18. A subtle bug concerned with back references has been fixed by a change of
1754 specification, with a corresponding code fix. A pattern such as
1755 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1756 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1757 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1758 same bug. Such groups have to be quantified to be useful, or contained
1759 inside another quantified group. (If there's no repetition, the reference
1760 can never match.) The problem arises because, having left the group and
1761 moved on to the rest of the pattern, a later failure that backtracks into
1762 the group uses the captured value from the final iteration of the group
1763 rather than the correct earlier one. I have fixed this in PCRE by forcing
1764 any group that contains a reference to itself to be an atomic group; that
1765 is, there cannot be any backtracking into it once it has completed. This is
1766 similar to recursive and subroutine calls.
1767
1768
1769 Version 8.00 19-Oct-09
1770 ----------------------
1771
1772 1. The table for translating pcre_compile() error codes into POSIX error codes
1773 was out-of-date, and there was no check on the pcre_compile() error code
1774 being within the table. This could lead to an OK return being given in
1775 error.
1776
1777 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1778 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1779 in a Windows environment.
1780
1781 3. The pcregrep --count option prints the count for each file even when it is
1782 zero, as does GNU grep. However, pcregrep was also printing all files when
1783 --files-with-matches was added. Now, when both options are given, it prints
1784 counts only for those files that have at least one match. (GNU grep just
1785 prints the file name in this circumstance, but including the count seems
1786 more useful - otherwise, why use --count?) Also ensured that the
1787 combination -clh just lists non-zero counts, with no names.
1788
1789 4. The long form of the pcregrep -F option was incorrectly implemented as
1790 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1791 but it seems right to fix it, and I didn't think it was worth preserving
1792 the old behaviour.
1793
1794 5. The command line items --regex=pattern and --regexp=pattern were not
1795 recognized by pcregrep, which required --regex pattern or --regexp pattern
1796 (with a space rather than an '='). The man page documented the '=' forms,
1797 which are compatible with GNU grep; these now work.
1798
1799 6. No libpcreposix.pc file was created for pkg-config; there was just
1800 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1801
1802 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1803 when UCP support is not needed, by modifying the Python script that
1804 generates it from Unicode data files. This should not matter if the module
1805 is correctly used as a library, but I received one complaint about 50K of
1806 unwanted data. My guess is that the person linked everything into his
1807 program rather than using a library. Anyway, it does no harm.
1808
1809 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1810 was a minimum greater than 1 for a wide character in a possessive
1811 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1812 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1813 character. Chaos in the form of incorrect output or a compiling loop could
1814 result.
1815
1816 9. The restrictions on what a pattern can contain when partial matching is
1817 requested for pcre_exec() have been removed. All patterns can now be
1818 partially matched by this function. In addition, if there are at least two
1819 slots in the offset vector, the offset of the earliest inspected character
1820 for the match and the offset of the end of the subject are set in them when
1821 PCRE_ERROR_PARTIAL is returned.
1822
1823 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1824 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1825 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1826 and may be more useful for multi-segment matching.
1827
1828 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1829 used to be given if ever the end of the subject was reached; now it is
1830 given only if matching could not proceed because another character was
1831 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1832 string "Z", which now yields "no match" instead of "partial match". In the
1833 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1834 final character ended with (*FAIL).
1835
1836 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1837 if the pattern had a "must contain" character that was already found in the
1838 earlier partial match, unless partial matching was again requested. For
1839 example, with the pattern /dog.(body)?/, the "must contain" character is
1840 "g". If the first part-match was for the string "dog", restarting with
1841 "sbody" failed. This bug has been fixed.
1842
1843 13. The string returned by pcre_dfa_exec() after a partial match has been
1844 changed so that it starts at the first inspected character rather than the
1845 first character of the match. This makes a difference only if the pattern
1846 starts with a lookbehind assertion or \b or \B (\K is not supported by
1847 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1848 matching functions compatible, and I think it's the right thing to do.
1849
1850 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1851 so that the demonstration program is easily available in environments where
1852 PCRE has not been installed from source.
1853
1854 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1855 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1856 library.
1857
1858 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1859 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1860 is not the first non-POSIX option to be added. Clearly some people find
1861 these options useful.
1862
1863 17. If a caller to the POSIX matching function regexec() passes a non-zero
1864 value for nmatch with a NULL value for pmatch, the value of
1865 nmatch is forced to zero.
1866
1867 18. RunGrepTest did not have a test for the availability of the -u option of
1868 the diff command, as RunTest does. It now checks in the same way as
1869 RunTest, and also checks for the -b option.
1870
1871 19. If an odd number of negated classes containing just a single character
1872 interposed, within parentheses, between a forward reference to a named
1873 subpattern and the definition of the subpattern, compilation crashed with
1874 an internal error, complaining that it could not find the referenced
1875 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1876 [The bug was that it was starting one character too far in when skipping
1877 over the character class, thus treating the ] as data rather than
1878 terminating the class. This meant it could skip too much.]
1879
1880 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1881 /g option in pcretest when the pattern contains \K, which makes it possible
1882 to have an empty string match not at the start, even when the pattern is
1883 anchored. Updated pcretest and pcredemo to use this option.
1884
1885 21. If the maximum number of capturing subpatterns in a recursion was greater
1886 than the maximum at the outer level, the higher number was returned, but
1887 with unset values at the outer level. The correct (outer level) value is
1888 now given.
1889
1890 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1891 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1892 make it do so. The string so far is captured, making this feature
1893 compatible with Perl.
1894
1895 23. The tests have been re-organized, adding tests 11 and 12, to make it
1896 possible to check the Perl 5.10 features against Perl 5.10.
1897
1898 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1899 pattern matches a fixed length string. PCRE did not allow this; now it
1900 does. Neither allows recursion.
1901
1902 25. I finally figured out how to implement a request to provide the minimum
1903 length of subject string that was needed in order to match a given pattern.
1904 (It was back references and recursion that I had previously got hung up
1905 on.) This code has now been added to pcre_study(); it finds a lower bound
1906 to the length of subject needed. It is not necessarily the greatest lower
1907 bound, but using it to avoid searching strings that are too short does give
1908 some useful speed-ups. The value is available to calling programs via
1909 pcre_fullinfo().
1910
1911 26. While implementing 25, I discovered to my embarrassment that pcretest had
1912 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1913 study optimizations had never been tested with that matching function.
1914 Oops. What is worse, even when it was passed study data, there was a bug in
1915 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1916 were also very few tests of studied patterns with pcre_dfa_exec().
1917
1918 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1919 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1920 on the other side of the coin, they are no longer allowed to have different
1921 names, because these cannot be distinguished in PCRE, and this has caused
1922 confusion. (This is a difference from Perl.)
1923
1924 28. When duplicate subpattern names are present (necessarily with different
1925 numbers, as required by 27 above), and a test is made by name in a
1926 conditional pattern, either for a subpattern having been matched, or for
1927 recursion in such a pattern, all the associated numbered subpatterns are
1928 tested, and the overall condition is true if the condition is true for any
1929 one of them. This is the way Perl works, and is also more like the way
1930 testing by number works.
1931
1932
1933 Version 7.9 11-Apr-09
1934 ---------------------
1935
1936 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1937 (pcretest), all targets were linked against these libraries. This included
1938 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1939 libraries. This caused unwanted dependencies to be created. This problem
1940 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1941 pcretest is linked with readline.
1942
1943 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1944 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1945 moved outside it again, because FALSE and TRUE are already defined in AIX,
1946 but BOOL is not.
1947
1948 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1949 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1950
1951 4. The pcregrep documentation said spaces were inserted as well as colons (or
1952 hyphens) following file names and line numbers when outputting matching
1953 lines. This is not true; no spaces are inserted. I have also clarified the
1954 wording for the --colour (or --color) option.
1955
1956 5. In pcregrep, when --colour was used with -o, the list of matching strings
1957 was not coloured; this is different to GNU grep, so I have changed it to be
1958 the same.
1959
1960 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1961 each matching line was coloured. Now it goes on to look for further matches
1962 of any of the test patterns, which is the same behaviour as GNU grep.
1963
1964 7. A pattern that could match an empty string could cause pcregrep to loop; it
1965 doesn't make sense to accept an empty string match in pcregrep, so I have
1966 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1967 seems to be how GNU grep behaves. [But see later change 40 for release
1968 8.33.]
1969
1970 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1971 start or after a newline", because the conditional assertion was not being
1972 correctly handled. The rule now is that both the assertion and what follows
1973 in the first alternative must satisfy the test.
1974
1975 9. If auto-callout was enabled in a pattern with a conditional group whose
1976 condition was an assertion, PCRE could crash during matching, both with
1977 pcre_exec() and pcre_dfa_exec().
1978
1979 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1980 used for matching.
1981
1982 11. Unicode property support in character classes was not working for
1983 characters (bytes) greater than 127 when not in UTF-8 mode.
1984
1985 12. Added the -M command line option to pcretest.
1986
1987 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1988
1989 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1990
1991 16. Added comments and documentation about mis-use of no_arg in the C++
1992 wrapper.
1993
1994 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1995 from Martin Jerabek that uses macro names for all relevant character and
1996 string constants.
1997
1998 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1999 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2000 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2001 these, but not everybody uses configure.
2002
2003 19. A conditional group that had only one branch was not being correctly
2004 recognized as an item that could match an empty string. This meant that an
2005 enclosing group might also not be so recognized, causing infinite looping
2006 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2007 with the subject "ab", where knowledge that the repeated group can match
2008 nothing is needed in order to break the loop.
2009
2010 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2011 exec(), but without supplying a callout function, matching went wrong.
2012
2013 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2014 leak if the size of the offset vector was greater than 30. When the vector
2015 is smaller, the saved offsets during recursion go onto a local stack
2016 vector, but for larger vectors malloc() is used. It was failing to free
2017 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2018 error, in fact).
2019
2020 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2021 heapframe that is used only when UTF-8 support is enabled. This caused no
2022 problem, but was untidy.
2023
2024 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2025 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2026 included within another project.
2027
2028 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2029 slightly modified by me:
2030
2031 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2032 not building pcregrep.
2033
2034 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2035 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2036
2037 25. Forward references, both numeric and by name, in patterns that made use of
2038 duplicate group numbers, could behave incorrectly or give incorrect errors,
2039 because when scanning forward to find the reference group, PCRE was not
2040 taking into account the duplicate group numbers. A pattern such as
2041 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2042
2043 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2044 the feature of a custom pointer more persuasive (as requested by a user).
2045
2046 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2047 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2048 pre-defined.
2049
2050 28. Added support for (*UTF8) at the start of a pattern.
2051
2052 29. Arrange for flags added by the "release type" setting in CMake to be shown
2053 in the configuration summary.
2054
2055
2056 Version 7.8 05-Sep-08
2057 ---------------------
2058
2059 1. Replaced UCP searching code with optimized version as implemented for Ad
2060 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2061 stage table and inline lookup instead of a function, giving speed ups of 2
2062 to 5 times on some simple patterns that I tested. Permission was given to
2063 distribute the MultiStage2.py script that generates the tables (it's not in
2064 the tarball, but is in the Subversion repository).
2065
2066 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2067 scripts.
2068
2069 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2070 a group with a zero qualifier. The result of the study could be incorrect,
2071 or the function might crash, depending on the pattern.
2072
2073 4. Caseless matching was not working for non-ASCII characters in back
2074 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2075 It now works when Unicode Property Support is available.
2076
2077 5. In pcretest, an escape such as \x{de} in the data was always generating
2078 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2079 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2080 truncation.
2081
2082 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2083
2084 7. Added two (int) casts to pcregrep when printing the difference of two
2085 pointers, in case they are 64-bit values.
2086
2087 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2088 test 2 if it fails.
2089
2090 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2091 and a #define of that name to empty if it is not externally set. This is to
2092 allow users of MSVC to set it if necessary.
2093
2094 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2095 the convenience functions in the pcre_get.c source file.
2096
2097 11. An option change at the start of a pattern that had top-level alternatives
2098 could cause overwriting and/or a crash. This command provoked a crash in
2099 some environments:
2100
2101 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2102
2103 This potential security problem was recorded as CVE-2008-2371.
2104
2105 12. For a pattern where the match had to start at the beginning or immediately
2106 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2107 pcre_dfa_exec() could read past the end of the passed subject if there was
2108 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2109 pcretest so that it places the subject at the end of its malloc-ed buffer.
2110
2111 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2112 exec() might read past the end of the data buffer in UTF-8 mode.
2113
2114 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2115 the data contained the byte 0x85 as part of a UTF-8 character within its
2116 first line. This applied both to normal and DFA matching.
2117
2118 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2119 /^[^d]*?$/8 failed to match "abc".
2120
2121 16. Added a missing copyright notice to pcrecpp_internal.h.
2122
2123 17. Make it more clear in the documentation that values returned from
2124 pcre_exec() in ovector are byte offsets, not character counts.
2125
2126 18. Tidied a few places to stop certain compilers from issuing warnings.
2127
2128 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2129 supplied by Stefan Weber. I made a further small update for 7.8 because
2130 there is a change of source arrangements: the pcre_searchfuncs.c module is
2131 replaced by pcre_ucd.c.
2132
2133
2134 Version 7.7 07-May-08
2135 ---------------------
2136
2137 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2138 a string to a long long, pretend we don't even have a long long." This is
2139 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2140
2141 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2142 pre-7.6 versions, which defined a global no_arg variable instead of putting
2143 it in the RE class. (See also #8 below.)
2144
2145 3. Remove a line of dead code, identified by coverity and reported by Nuno
2146 Lopes.
2147
2148 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2149
2150 (1) The include/exclude patterns were being applied to the whole pathnames
2151 of files, instead of just to the final components.
2152
2153 (2) If there was more than one level of directory, the subdirectories were
2154 skipped unless they satisfied the include/exclude conditions. This is
2155 inconsistent with GNU grep (and could even be seen as contrary to the
2156 pcregrep specification - which I improved to make it absolutely clear).
2157 The action now is always to scan all levels of directory, and just
2158 apply the include/exclude patterns to regular files.
2159
2160 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2161 --exclude_dir in the tests to avoid scanning .svn directories.
2162
2163 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2164 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2165 doesn't support NULs in patterns.
2166
2167 7. Added some missing "const"s to declarations of static tables in
2168 pcre_compile.c and pcre_dfa_exec.c.
2169
2170 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2171 caused by fix #2 above. (Subsequently also a second patch to fix the
2172 first patch. And a third patch - this was a messy problem.)
2173
2174 9. Applied Craig's patch to remove the use of push_back().
2175
2176 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2177 matching function regexec().
2178
2179 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2180 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2181 references. PCRE supports relative numbers with this syntax (I don't think
2182 Oniguruma does).
2183
2184 12. Previously, a group with a zero repeat such as (...){0} was completely
2185 omitted from the compiled regex. However, this means that if the group
2186 was called as a subroutine from elsewhere in the pattern, things went wrong
2187 (an internal error was given). Such groups are now left in the compiled
2188 pattern, with a new opcode that causes them to be skipped at execution
2189 time.
2190
2191 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2192 to the way PCRE behaves:
2193
2194 (a) A lone ] character is dis-allowed (Perl treats it as data).
2195
2196 (b) A back reference to an unmatched subpattern matches an empty string
2197 (Perl fails the current match path).
2198
2199 (c) A data ] in a character class must be notated as \] because if the
2200 first data character in a class is ], it defines an empty class. (In
2201 Perl it is not possible to have an empty class.) The empty class []
2202 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2203 The negative empty class [^] matches any one character, independently
2204 of the DOTALL setting.
2205
2206 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2207 non-existent subpattern following a character class starting with ']' and
2208 containing () gave an internal compiling error instead of "reference to
2209 non-existent subpattern". Fortunately, when the pattern did exist, the
2210 compiled code was correct. (When scanning forwards to check for the
2211 existence of the subpattern, it was treating the data ']' as terminating
2212 the class, so got the count wrong. When actually compiling, the reference
2213 was subsequently set up correctly.)
2214
2215 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2216 it was being rejected as not supported by pcre_dfa_exec(), even though
2217 other assertions are supported. I have made pcre_dfa_exec() support
2218 (*FAIL).
2219
2220 16. The implementation of 13c above involved the invention of a new opcode,
2221 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2222 cannot be changed at match time, I realized I could make a small
2223 improvement to matching performance by compiling OP_ALLANY instead of
2224 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2225 on the OP_ANY path.
2226
2227 17. Compiling pcretest on Windows with readline support failed without the
2228 following two fixes: (1) Make the unistd.h include conditional on
2229 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2230
2231 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2232 ncurses library to be included for pcretest when ReadLine support is
2233 requested, but also to allow for it to be overridden. This patch came from
2234 Daniel Bergström.
2235
2236 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2237 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2238 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2239 spotting this.
2240
2241
2242 Version 7.6 28-Jan-08
2243 ---------------------
2244
2245 1. A character class containing a very large number of characters with
2246 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2247 overflow.
2248
2249 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2250 HAVE_LONG_LONG is not defined.
2251
2252 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2253 bring them up to date and include new features. This patch includes:
2254
2255 - Fixed PH's badly added libz and libbz2 support.
2256 - Fixed a problem with static linking.
2257 - Added pcredemo. [But later removed - see 7 below.]
2258 - Fixed dftables problem and added an option.
2259 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2260 HAVE_LONG_LONG.
2261 - Added readline support for pcretest.
2262 - Added an listing of the option settings after cmake has run.
2263
2264 4. A user submitted a patch to Makefile that makes it easy to create
2265 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2266 Makefile.am that cause it to include this special target, without
2267 affecting anything else. Note that the same mingw target plus all
2268 the other distribution libraries and programs are now supported
2269 when configuring with CMake (see 6 below) instead of with
2270 Configure/Make.
2271
2272 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2273 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2274 exported in the Windows port". It has not yet been confirmed that the patch
2275 solves the problem, but it does no harm.
2276
2277 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2278 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2279 with CMake, and also correct the comment about stack recursion.
2280
2281 7. Remove the automatic building of pcredemo from the ./configure system and
2282 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2283 of a program that users should build themselves after PCRE is installed, so
2284 building it automatically is not really right. What is more, it gave
2285 trouble in some build environments.
2286
2287 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2288
2289
2290 Version 7.5 10-Jan-08
2291 ---------------------
2292
2293 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2294 values in parens when parsing an RE using the C++ wrapper."
2295
2296 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2297 Characters greater than 255 were excluded from the class instead of being
2298 included.
2299
2300 3. The same bug as (2) above applied to negated POSIX classes such as
2301 [:^space:].
2302
2303 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2304 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2305 I have changed it.
2306
2307 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2308 first named subpattern) and a construct such as (?&a) would reference the
2309 first named subpattern whose name started with "a" (in other words, the
2310 length check was missing). Both these problems are fixed. "Subpattern name
2311 expected" is now given for (?&) (a zero-length name), and this patch also
2312 makes it give the same error for \k'' (previously it complained that that
2313 was a reference to a non-existent subpattern).
2314
2315 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2316 this is right because (?- can be followed by option settings as well as by
2317 digits. I have, however, made the messages clearer.
2318
2319 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2320 than the number used in the conditional) now cause a compile-time error.
2321 This is actually not compatible with Perl, which accepts such patterns, but
2322 treats the conditional as always being FALSE (as PCRE used to), but it
2323 seems to me that giving a diagnostic is better.
2324
2325 8. Change "alphameric" to the more common word "alphanumeric" in comments
2326 and messages.
2327
2328 9. Fix two occurrences of "backslash" in comments that should have been
2329 "backspace".
2330
2331 10. Remove two redundant lines of code that can never be obeyed (their function
2332 was moved elsewhere).
2333
2334 11. The program that makes PCRE's Unicode character property table had a bug
2335 which caused it to generate incorrect table entries for sequences of
2336 characters that have the same character type, but are in different scripts.
2337 It amalgamated them into a single range, with the script of the first of
2338 them. In other words, some characters were in the wrong script. There were
2339 thirteen such cases, affecting characters in the following ranges:
2340
2341 U+002b0 - U+002c1
2342 U+0060c - U+0060d
2343 U+0061e - U+00612
2344 U+0064b - U+0065e
2345 U+0074d - U+0076d
2346 U+01800 - U+01805
2347 U+01d00 - U+01d77
2348 U+01d9b - U+01dbf
2349 U+0200b - U+0200f
2350 U+030fc - U+030fe
2351 U+03260 - U+0327f
2352 U+0fb46 - U+0fbb1
2353 U+10450 - U+1049d
2354
2355 12. The -o option (show only the matching part of a line) for pcregrep was not
2356 compatible with GNU grep in that, if there was more than one match in a
2357 line, it showed only the first of them. It now behaves in the same way as
2358 GNU grep.
2359
2360 13. If the -o and -v options were combined for pcregrep, it printed a blank
2361 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2362 does the same. The return code can be used to tell if there were any
2363 non-matching lines.
2364
2365 14. Added --file-offsets and --line-offsets to pcregrep.
2366
2367 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2368 infinitely looping recursion. The bug was that positive lookaheads were not
2369 being skipped when checking for a possible empty match (negative lookaheads
2370 and both kinds of lookbehind were skipped).
2371
2372 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2373 inclusion of <windows.h> to before rather than after the definition of
2374 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2375
2376 17. Specifying a possessive quantifier with a specific limit for a Unicode
2377 character property caused pcre_compile() to compile bad code, which led at
2378 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2379 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2380 caused the error; without that there was no problem.
2381
2382 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2383
2384 19. Added --enable-pcretest-libreadline.
2385
2386 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2387 RE::GlobalReplace(). As a result, the number of replacements returned was
2388 double what it should be. I removed one of the increments, but Craig sent a
2389 later patch that removed the other one (the right fix) and added unit tests
2390 that check the return values (which was not done before).
2391
2392 21. Several CMake things:
2393
2394 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2395 the names libpcre and libpcreposix, not just pcre and pcreposix.
2396
2397 (2) The above change means that pcretest and pcregrep are now correctly
2398 linked with the newly-built libraries, not previously installed ones.
2399
2400 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2401
2402 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2403 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2404 UTF-8 newline character). The key issue is that the pattern starts .*;
2405 this means that the match must be either at the beginning, or after a
2406 newline. The bug was in the code for advancing after a failed match and
2407 checking that the new position followed a newline. It was not taking
2408 account of UTF-8 characters correctly.
2409
2410 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2411 character classes. PCRE was not treating the sequence [:...:] as a
2412 character class unless the ... were all letters. Perl, however, seems to
2413 allow any characters between [: and :], though of course it rejects as
2414 unknown any "names" that contain non-letters, because all the known class
2415 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2416 for example, whereas PCRE did not - it did not recognize a POSIX character
2417 class. This seemed a bit dangerous, so the code has been changed to be
2418 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2419 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2420 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2421 Perl does, and where it didn't before.
2422
2423 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2424 Windows environments %n is disabled by default.
2425
2426
2427 Version 7.4 21-Sep-07
2428 ---------------------
2429
2430 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2431 means that a class such as [\s] counted as "explicit reference to CR or
2432 LF". That isn't really right - the whole point of the change was to try to
2433 help when there was an actual mention of one of the two characters. So now
2434 the change happens only if \r or \n (or a literal CR or LF) character is
2435 encountered.
2436
2437 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2438 of both had grown to the point where there were only 3 bits left.
2439 Fortunately, there was spare space in the data structure, and so I have
2440 moved the internal flags into a new 16-bit field to free up more option
2441 bits.
2442
2443 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2444 but did not set the internal JCHANGED flag - either of these is enough to
2445 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2446 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2447 start sets both bits.
2448
2449 4. Added options (at build time, compile time, exec time) to change \R from
2450 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2451
2452 5. doc/pcresyntax.html was missing from the distribution.
2453
2454 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2455 compatibility, even though it is no longer used.
2456
2457 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2458 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2459 windows.h file is present (where different names are used). [This was
2460 reversed later after testing - see 16 below.]
2461
2462 8. Changed all #include <config.h> to #include "config.h". There were also
2463 some further <pcre.h> cases that I changed to "pcre.h".
2464
2465 9. When pcregrep was used with the --colour option, it missed the line ending
2466 sequence off the lines that it output.
2467
2468 10. It was pointed out to me that arrays of string pointers cause lots of
2469 relocations when a shared library is dynamically loaded. A technique of
2470 using a single long string with a table of offsets can drastically reduce
2471 these. I have refactored PCRE in four places to do this. The result is
2472 dramatic:
2473
2474 Originally: 290
2475 After changing UCP table: 187
2476 After changing error message table: 43
2477 After changing table of "verbs" 36
2478 After changing table of Posix names 22
2479
2480 Thanks to the folks working on Gregex for glib for this insight.
2481
2482 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2483 unicode-properties was also set.
2484
2485 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2486
2487 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2488 checked only for CRLF.
2489
2490 14. Added casts to pcretest.c to avoid compiler warnings.
2491
2492 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2493
2494 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2495 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2496 entirely. This removes changes made in 7 above.
2497
2498 17. The CMake files have been updated, and there is now more information about
2499 building with CMake in the NON-UNIX-USE document.
2500
2501
2502 Version 7.3 28-Aug-07
2503 ---------------------
2504
2505 1. In the rejigging of the build system that eventually resulted in 7.1, the
2506 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2507 brackets there is not right, since it causes compilers to look for an
2508 installed pcre.h, not the version that is in the source that is being
2509 compiled (which of course may be different). I have changed it back to:
2510
2511 #include "pcre.h"
2512
2513 I have a vague recollection that the change was concerned with compiling in
2514 different directories, but in the new build system, that is taken care of
2515 by the VPATH setting the Makefile.
2516
2517 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2518 when the subject happened to end in the byte 0x85 (e.g. if the last
2519 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2520 characters but of course it shouldn't be taken as a newline when it is part
2521 of another character. The bug was that, for an unlimited repeat of . in
2522 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2523 characters when looking for a newline.
2524
2525 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2526
2527 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2528 in debug output.
2529
2530 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2531 long printing in the pcrecpp unittest when running under MinGW.
2532
2533 6. ESC_K was left out of the EBCDIC table.
2534
2535 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2536 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2537 limit also applies to "virtual nesting" when a pattern is recursive, and in
2538 this case 1000 isn't so big. I have been able to remove this limit at the
2539 expense of backing off one optimization in certain circumstances. Normally,
2540 when pcre_exec() would call its internal match() function recursively and
2541 immediately return the result unconditionally, it uses a "tail recursion"
2542 feature to save stack. However, when a subpattern that can match an empty
2543 string has an unlimited repetition quantifier, it no longer makes this
2544 optimization. That gives it a stack frame in which to save the data for
2545 checking that an empty string has been matched. Previously this was taken
2546 from the 1000-entry workspace that had been reserved. So now there is no
2547 explicit limit, but more stack is used.
2548
2549 8. Applied Daniel's patches to solve problems with the import/export magic
2550 syntax that is required for Windows, and which was going wrong for the
2551 pcreposix and pcrecpp parts of the library. These were overlooked when this
2552 problem was solved for the main library.
2553
2554 9. There were some crude static tests to avoid integer overflow when computing
2555 the size of patterns that contain repeated groups with explicit upper
2556 limits. As the maximum quantifier is 65535, the maximum group length was
2557 set at 30,000 so that the product of these two numbers did not overflow a
2558 32-bit integer. However, it turns out that people want to use groups that
2559 are longer than 30,000 bytes (though not repeat them that many times).
2560 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2561 made it possible to implement the integer overflow checks in a much more
2562 dynamic way, which I have now done. The artificial limitation on group
2563 length has been removed - we now have only the limit on the total length of
2564 the compiled pattern, which depends on the LINK_SIZE setting.
2565
2566 10. Fixed a bug in the documentation for get/copy named substring when
2567 duplicate names are permitted. If none of the named substrings are set, the
2568 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2569 empty string.
2570
2571 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2572 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2573 because the ] is interpreted as the first data character and the
2574 terminating ] is not found. PCRE has been made compatible with Perl in this
2575 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2576 cause memory overwriting.
2577
2578 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2579 string has been matched (to stop an infinite loop). It was not recognizing
2580 a conditional subpattern that could match an empty string if that
2581 subpattern was within another subpattern. For example, it looped when
2582 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2583 condition was not nested. This bug has been fixed.
2584
2585 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2586 past the start of the subject in the presence of bytes with the top bit
2587 set, for example "\x8aBCD".
2588
2589 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2590 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2591
2592 14. Optimized (?!) to (*FAIL).
2593
2594 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2595 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2596 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2597 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2598 does: it's just the validity check that is more restrictive.
2599
2600 16. Inserted checks for integer overflows during escape sequence (backslash)
2601 processing, and also fixed erroneous offset values for syntax errors during
2602 backslash processing.
2603
2604 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2605 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2606
2607 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2608 caused an overrun.
2609
2610 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2611 something other than just ASCII characters) inside a group that had an
2612 unlimited repeat caused a loop at compile time (while checking to see
2613 whether the group could match an empty string).
2614
2615 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2616 [\P{Any}] did so. (Error in the code for printing property names.)
2617
2618 21. An orphan \E inside a character class could cause a crash.
2619
2620 22. A repeated capturing bracket such as (A)? could cause a wild memory
2621 reference during compilation.
2622
2623 23. There are several functions in pcre_compile() that scan along a compiled
2624 expression for various reasons (e.g. to see if it's fixed length for look
2625 behind). There were bugs in these functions when a repeated \p or \P was
2626 present in the pattern. These operators have additional parameters compared
2627 with \d, etc, and these were not being taken into account when moving along
2628 the compiled data. Specifically:
2629
2630 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2631 length.
2632
2633 (b) An item such as \pL+ within a repeated group could cause crashes or
2634 loops.
2635
2636 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2637 "reference to non-existent subpattern" error.
2638
2639 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2640
2641 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2642 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2643
2644 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2645
2646 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2647 character were causing crashes (broken optimization).
2648
2649 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2650 \p or \P) caused a compile-time loop.
2651
2652 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2653 break. For example, the unstudied pattern [\r\n]A does not match the string
2654 "\r\nA" because change 7.0/46 below moves the current point on by two
2655 characters after failing to match at the start. However, the pattern \nA
2656 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2657 the same is true. There doesn't seem any very clean way out of this, but
2658 what I have chosen to do makes the common cases work: PCRE now takes note
2659 of whether there can be an explicit match for \r or \n anywhere in the
2660 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2661 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2662 pattern has explicit CR or LF references.
2663
2664 29. Added (*CR) etc for changing newline setting at start of pattern.
2665
2666
2667 Version 7.2 19-Jun-07
2668 ---------------------
2669
2670 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2671 which is apparently normally available under Windows.
2672
2673 2. Re-jig the pcregrep tests with different newline settings in an attempt
2674 to make them independent of the local environment's newline setting.
2675
2676 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2677
2678 4. Some of the "internals" tests were previously cut out when the link size
2679 was not 2, because the output contained actual offsets. The recent new
2680 "Z" feature of pcretest means that these can be cut out, making the tests
2681 usable with all link sizes.
2682
2683 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2684 stack recursion. This gives a massive performance boost under BSD, but just
2685 a small improvement under Linux. However, it saves one field in the frame
2686 in all cases.
2687
2688 6. Added more features from the forthcoming Perl 5.10:
2689
2690 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2691 recursion call. It refers to the nth most recently opened parentheses.
2692
2693 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2694 to be opened parentheses.
2695
2696 (c) Conditions that refer to capturing parentheses can be specified
2697 relatively, for example, (?(-2)... or (?(+3)...
2698
2699 (d) \K resets the start of the current match so that everything before
2700 is not part of it.
2701
2702 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2703
2704 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2705 reference syntax.
2706
2707 (g) (?| introduces a group in which the numbering of parentheses in each
2708 alternative starts with the same number.
2709
2710 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2711
2712 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2713 PCRE_INFO_JCHANGED.
2714
2715 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2716 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2717 for detecting groups that can match an empty string.
2718
2719 9. A pattern with a very large number of alternatives (more than several
2720 hundred) was running out of internal workspace during the pre-compile
2721 phase, where pcre_compile() figures out how much memory will be needed. A
2722 bit of new cunning has reduced the workspace needed for groups with
2723 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2724 workspace instead of running out of the 4096 that are available.
2725
2726 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2727
2728 11. Applied patch from Google to remove an optimization that didn't quite work.
2729 The report of the bug said:
2730
2731 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2732 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2733 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2734
2735 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2736 it matched the wrong number of bytes.
2737
2738
2739 Version 7.1 24-Apr-07
2740 ---------------------
2741
2742 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2743 that is more "standard", making use of automake and other Autotools. There
2744 is some re-arrangement of the files and adjustment of comments consequent
2745 on this.
2746
2747 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2748 for recursive directory scanning broke on some systems because the files
2749 are not scanned in any specific order and on different systems the order
2750 was different. A call to "sort" has been inserted into RunGrepTest for the
2751 approprate test as a short-term fix. In the longer term there may be an
2752 alternative.
2753
2754 3. I had an email from Eric Raymond about problems translating some of PCRE's
2755 man pages to HTML (despite the fact that I distribute HTML pages, some
2756 people do their own conversions for various reasons). The problems
2757 concerned the use of low-level troff macros .br and .in. I have therefore
2758 removed all such uses from the man pages (some were redundant, some could
2759 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2760 HTML has been updated to handle .nf/.fi and to complain if it encounters
2761 .br or .in.
2762
2763 4. Updated comments in configure.ac that get placed in config.h.in and also
2764 arranged for config.h to be included in the distribution, with the name
2765 config.h.generic, for the benefit of those who have to compile without
2766 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2767
2768 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2769 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2770 makevp.bat for the current PCRE, using the additional files
2771 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2772
2773 6. A Windows user reported a minor discrepancy with test 2, which turned out
2774 to be caused by a trailing space on an input line that had got lost in his
2775 copy. The trailing space was an accident, so I've just removed it.
2776
2777 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2778 that is needed.
2779
2780 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2781 as "const" (a) because they are and (b) because it helps the PHP
2782 maintainers who have recently made a script to detect big data structures
2783 in the php code that should be moved to the .rodata section. I remembered
2784 to update Builducptable as well, so it won't revert if ucptable.h is ever
2785 re-created.
2786
2787 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2788 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2789 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2790 support is not required. This saves 1.5-2K of code, which is important in
2791 some applications.
2792
2793 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2794 so as not to refer to the tables, even though these functions will never be
2795 called when UTF-8 support is disabled. Otherwise there are problems with a
2796 shared library.
2797
2798 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2799
2800 (a) It was defining its arguments as char * instead of void *.
2801
2802 (b) It was assuming that all moves were upwards in memory; this was true
2803 a long time ago when I wrote it, but is no longer the case.
2804
2805 The emulated memove() is provided for those environments that have neither
2806 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2807 is clearly not the case, as these two bugs were recently reported.
2808
2809 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2810 and Detrail to create the HTML documentation, the .txt form of the man
2811 pages, and it removes trailing spaces from listed files. It also creates
2812 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2813 case, it wraps all the #defines with #ifndefs. This script should be run
2814 before "make dist".
2815
2816 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2817 with Unicode property support.
2818
2819 (a) For a maximizing quantifier, if the two different cases of the
2820 character were of different lengths in their UTF-8 codings (there are
2821 some cases like this - I found 11), and the matching function had to
2822 back up over a mixture of the two cases, it incorrectly assumed they
2823 were both the same length.
2824
2825 (b) When PCRE was configured to use the heap rather than the stack for
2826 recursion during matching, it was not correctly preserving the data for
2827 the other case of a UTF-8 character when checking ahead for a match
2828 while processing a minimizing repeat. If the check also involved
2829 matching a wide character, but failed, corruption could cause an
2830 erroneous result when trying to check for a repeat of the original
2831 character.
2832
2833 13. Some tidying changes to the testing mechanism:
2834
2835 (a) The RunTest script now detects the internal link size and whether there
2836 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2837 values substituted by "configure". (The RunGrepTest script already did
2838 this for UTF-8.) The configure.ac script no longer substitutes the
2839 relevant variables.
2840
2841 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2842 with length and offset values. This means that the output is different
2843 for different internal link sizes. Test 2 is skipped for link sizes
2844 other than 2 because of this, bypassing the problem. Unfortunately,
2845 there was also a test in test 3 (the locale tests) that used /B and
2846 failed for link sizes other than 2. Rather than cut the whole test out,
2847 I have added a new /Z option to pcretest that replaces the length and
2848 offset values with spaces. This is now used to make test 3 independent
2849 of link size. (Test 2 will be tidied up later.)
2850
2851 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2852 segmentation fault instead of returning the appropriate error message.
2853
2854 15. In multiline mode when the newline sequence was set to "any", the pattern
2855 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2856 This doesn't seem right; it now treats the CRLF combination as the line
2857 ending, and so does not match in that case. It's only a pattern such as ^$
2858 that would hit this one: something like ^ABC$ would have failed after \r
2859 and then tried again after \r\n.
2860
2861 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2862 in an attempt to make files that differ only in their line terminators
2863 compare equal. This works on Linux.
2864
2865 17. Under certain error circumstances pcregrep might try to free random memory
2866 as it exited. This is now fixed, thanks to valgrind.
2867
2868 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2869 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2870 was because its rules for how to advance for /g after matching an empty
2871 string at the end of a line did not allow for this case. They now check for
2872 it specially.
2873
2874 20. pcretest is supposed to handle patterns and data of any length, by
2875 extending its buffers when necessary. It was getting this wrong when the
2876 buffer for a data line had to be extended.
2877
2878 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2879 CRLF as a newline sequence.
2880
2881 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2882 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2883 I have nevertheless tidied it up.
2884
2885 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2886
2887 24. Added a man page for pcre-config.
2888
2889
2890 Version 7.0 19-Dec-06
2891 ---------------------
2892
2893 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2894 moving to gcc 4.1.1.
2895
2896 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2897 sys/time.h, which is documented as needed for this function. It doesn't
2898 seem to matter on Linux, but it showed up on some releases of OS X.
2899
2900 3. It seems that there are systems where bytes whose values are greater than
2901 127 match isprint() in the "C" locale. The "C" locale should be the
2902 default when a C program starts up. In most systems, only ASCII printing
2903 characters match isprint(). This difference caused the output from pcretest
2904 to vary, making some of the tests fail. I have changed pcretest so that:
2905
2906 (a) When it is outputting text in the compiled version of a pattern, bytes
2907 other than 32-126 are always shown as hex escapes.
2908
2909 (b) When it is outputting text that is a matched part of a subject string,
2910 it does the same, unless a different locale has been set for the match
2911 (using the /L modifier). In this case, it uses isprint() to decide.
2912
2913 4. Fixed a major bug that caused incorrect computation of the amount of memory
2914 required for a compiled pattern when options that changed within the
2915 pattern affected the logic of the preliminary scan that determines the
2916 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2917 that the computed length was too small. The symptoms of this bug were
2918 either the PCRE error "internal error: code overflow" from pcre_compile(),
2919 or a glibc crash with a message such as "pcretest: free(): invalid next
2920 size (fast)". Examples of patterns that provoked this bug (shown in
2921 pcretest format) are:
2922
2923 /(?-x: )/x
2924 /(?x)(?-x: \s*#\s*)/
2925 /((?i)[\x{c0}])/8
2926 /(?i:[\x{c0}])/8
2927
2928 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2929 is now done differently.
2930
2931 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2932 wrapper classes; (b) implement a new function in the C++ scanner that is
2933 more efficient than the old way of doing things because it avoids levels of
2934 recursion in the regex matching; (c) add a paragraph to the documentation
2935 for the FullMatch() function.
2936
2937 6. The escape sequence \n was being treated as whatever was defined as
2938 "newline". Not only was this contrary to the documentation, which states
2939 that \n is character 10 (hex 0A), but it also went horribly wrong when
2940 "newline" was defined as CRLF. This has been fixed.
2941
2942 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2943 was being set to -1 for the "end of line" case (supposedly a value that no
2944 character can have). Though this value is never used (the check for end of
2945 line is "zero bytes in current character"), it caused compiler complaints.
2946 I've changed it to 0xffffffff.
2947
2948 8. In pcre_version.c, the version string was being built by a sequence of
2949 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2950 string (as it is for production releases) called a macro with an empty
2951 argument. The C standard says the result of this is undefined. The gcc
2952 compiler treats it as an empty string (which was what was wanted) but it is
2953 reported that Visual C gives an error. The source has been hacked around to
2954 avoid this problem.
2955
2956 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2957 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2958 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2959 of them did).
2960
2961 10. Originally, pcretest opened its input and output without "b"; then I was
2962 told that "b" was needed in some environments, so it was added for release
2963 5.0 to both the input and output. (It makes no difference on Unix-like
2964 systems.) Later I was told that it is wrong for the input on Windows. I've
2965 now abstracted the modes into two macros, to make it easier to fiddle with
2966 them, and removed "b" from the input mode under Windows.
2967
2968 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2969
2970 12. Added -help and --help to pcretest as an official way of being reminded
2971 of the options.
2972
2973 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2974 and pcrecpp.cc because they annoy compilers at high warning levels.
2975
2976 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2977
2978 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2979 scripts are not C programs :-) and which was not noticed because it works
2980 on Linux.
2981
2982 16. pcretest is supposed to handle any length of pattern and data line (as one
2983 line or as a continued sequence of lines) by extending its input buffer if
2984 necessary. This feature was broken for very long pattern lines, leading to
2985 a string of junk being passed to pcre_compile() if the pattern was longer
2986 than about 50K.
2987
2988 17. I have done a major re-factoring of the way pcre_compile() computes the
2989 amount of memory needed for a compiled pattern. Previously, there was code
2990 that made a preliminary scan of the pattern in order to do this. That was
2991 OK when PCRE was new, but as the facilities have expanded, it has become
2992 harder and harder to keep it in step with the real compile phase, and there
2993 have been a number of bugs (see for example, 4 above). I have now found a
2994 cunning way of running the real compile function in a "fake" mode that
2995 enables it to compute how much memory it would need, while actually only
2996 ever using a few hundred bytes of working memory and without too many
2997 tests of the mode. This should make future maintenance and development
2998 easier. A side effect of this work is that the limit of 200 on the nesting
2999 depth of parentheses has been removed (though this was never a serious
3000 limitation, I suspect). However, there is a downside: pcre_compile() now
3001 runs more slowly than before (30% or more, depending on the pattern). I
3002 hope this isn't a big issue. There is no effect on runtime performance.
3003
3004 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3005 newline (only possible for the last line of a file) and it was a
3006 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3007
3008 19. Added additional timing features to pcretest. (1) The -tm option now times
3009 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3010 separate command line item, by a number that specifies the number of
3011 repeats to use when timing. The default is 50000; this gives better
3012 precision, but takes uncomfortably long for very large patterns.
3013
3014 20. Extended pcre_study() to be more clever in cases where a branch of a
3015 subpattern has no definite first character. For example, (a*|b*)[cd] would
3016 previously give no result from pcre_study(). Now it recognizes that the
3017 first character must be a, b, c, or d.
3018
3019 21. There was an incorrect error "recursive call could loop indefinitely" if
3020 a subpattern (or the entire pattern) that was being tested for matching an
3021 empty string contained only one non-empty item after a nested subpattern.
3022 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3023 incorrectly, because the \d was being skipped in the check.
3024
3025 22. The pcretest program now has a new pattern option /B and a command line
3026 option -b, which is equivalent to adding /B to every pattern. This causes
3027 it to show the compiled bytecode, without the additional information that
3028 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3029 is the same as /B/I).
3030
3031 23. A new optimization is now able automatically to treat some sequences such
3032 as a*b as a*+b. More specifically, if something simple (such as a character
3033 or a simple class like \d) has an unlimited quantifier, and is followed by
3034 something that cannot possibly match the quantified thing, the quantifier
3035 is automatically "possessified".
3036
3037 24. A recursive reference to a subpattern whose number was greater than 39
3038 went wrong under certain circumstances in UTF-8 mode. This bug could also
3039 have affected the operation of pcre_study().
3040
3041 25. Realized that a little bit of performance could be had by replacing
3042 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3043
3044 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3045
3046 27. Possessive quantifiers such as a++ were previously implemented by turning
3047 them into atomic groups such as ($>a+). Now they have their own opcodes,
3048 which improves performance. This includes the automatically created ones
3049 from 23 above.
3050
3051 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3052 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3053 the first matched character to be a colon. This applied both to named and
3054 numbered groups.
3055
3056 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3057
3058 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3059 building PCRE on Windows easier, so I have included it in the distribution.
3060
3061 31. There is now a check in pcretest against a ridiculously large number being
3062 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3063 loop, the loop is abandoned.
3064
3065 32. Forward references to subpatterns in conditions such as (?(2)...) where
3066 subpattern 2 is defined later cause pcre_compile() to search forwards in
3067 the pattern for the relevant set of parentheses. This search went wrong
3068 when there were unescaped parentheses in a character class, parentheses
3069 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3070
3071 33. "Subroutine" calls and backreferences were previously restricted to
3072 referencing subpatterns earlier in the regex. This restriction has now
3073 been removed.
3074
3075 34. Added a number of extra features that are going to be in Perl 5.10. On the
3076 whole, these are just syntactic alternatives for features that PCRE had
3077 previously implemented using the Python syntax or my own invention. The
3078 other formats are all retained for compatibility.
3079
3080 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3081 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3082 also .NET compatible.
3083
3084 (b) A recursion or subroutine call to a named group can now be defined as
3085 (?&name) as well as (?P>name).
3086
3087 (c) A backreference to a named group can now be defined as \k<name> or
3088 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3089 5.10, are also .NET compatible.
3090
3091 (d) A conditional reference to a named group can now use the syntax
3092 (?(<name>) or (?('name') as well as (?(name).
3093
3094 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3095 groups (named and numbered) that are never evaluated inline, but can be
3096 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3097 is always false. There may be only one alternative in such a group.
3098
3099 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3100 as the simple (?(R). The condition is true only if the most recent
3101 recursion is that of the given number or name. It does not search out
3102 through the entire recursion stack.
3103
3104 (g) The escape \gN or \g{N} has been added, where N is a positive or
3105 negative number, specifying an absolute or relative reference.
3106
3107 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3108 some "unreachable code" warnings.
3109
3110 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3111 things, this adds five new scripts.
3112
3113 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3114 There were also incompatibilities regarding the handling of \Q..\E inside
3115 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3116 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3117
3118 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3119 matches an empty string, and forcibly breaks the loop. There were bugs in
3120 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3121 against aaaa the result was just "a" rather than "aaaa", for example. Two
3122 separate and independent bugs (that affected different cases) have been
3123 fixed.
3124
3125 39. Refactored the code to abolish the use of different opcodes for small
3126 capturing bracket numbers. This is a tidy that I avoided doing when I
3127 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3128 The new approach is not only tidier, it makes it possible to reduce the
3129 memory needed to fix the previous bug (38).
3130
3131 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3132 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3133 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3134 mode.
3135
3136 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3137 report.
3138
3139 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3140 copy construction and assignment in the C++ wrapper.
3141
3142 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3143 couple of bugs that could have given wrong results in the "--newline=crlf"
3144 case.
3145
3146 44. Added a number of casts and did some reorganization of signed/unsigned int
3147 variables following suggestions from Dair Grant. Also renamed the variable
3148 "this" as "item" because it is a C++ keyword.
3149
3150 45. Arranged for dftables to add
3151
3152 #include "pcre_internal.h"
3153
3154 to pcre_chartables.c because without it, gcc 4.x may remove the array
3155 definition from the final binary if PCRE is built into a static library and
3156 dead code stripping is activated.
3157
3158 46. For an unanchored pattern, if a match attempt fails at the start of a
3159 newline sequence, and the newline setting is CRLF or ANY, and the next two
3160 characters are CRLF, advance by two characters instead of one.
3161
3162
3163 Version 6.7 04-Jul-06
3164 ---------------------
3165
3166 1. In order to handle tests when input lines are enormously long, pcretest has
3167 been re-factored so that it automatically extends its buffers when
3168 necessary. The code is crude, but this _is_ just a test program. The
3169 default size has been increased from 32K to 50K.
3170
3171 2. The code in pcre_study() was using the value of the re argument before
3172 testing it for NULL. (Of course, in any sensible call of the function, it
3173 won't be NULL.)
3174
3175 3. The memmove() emulation function in pcre_internal.h, which is used on
3176 systems that lack both memmove() and bcopy() - that is, hardly ever -
3177 was missing a "static" storage class specifier.
3178
3179 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3180 containing an extended class (one that cannot be represented by a bitmap
3181 because it contains high-valued characters or Unicode property items, e.g.
3182 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3183 pattern, but PCRE should not loop if you do not (it no longer does).
3184 [Detail: two cases were found: (a) a repeated subpattern containing an
3185 extended class; (b) a recursive reference to a subpattern that followed a
3186 previous extended class. It wasn't skipping over the extended class
3187 correctly when UTF-8 mode was not set.]
3188
3189 5. A negated single-character class was not being recognized as fixed-length
3190 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3191 compile error "lookbehind assertion is not fixed length".
3192
3193 6. The RunPerlTest auxiliary script was showing an unexpected difference
3194 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3195 write a Perl script that can interpret lines of an input file either as
3196 byte characters or as UTF-8, which is what "perltest" was being required to
3197 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3198 can't do is switch easily at run time between having the "use utf8;" pragma
3199 or not. In the end, I fudged it by using the RunPerlTest script to insert
3200 "use utf8;" explicitly for the UTF-8 tests.
3201
3202 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3203 the end of the subject string, contrary to the documentation and to what
3204 Perl does. This was true of both matching functions. Now it matches only at
3205 the start of the subject and immediately after *internal* newlines.
3206
3207 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3208 a pointer to an int instead of a pointer to an unsigned long int. This
3209 caused problems on 64-bit systems.
3210
3211 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3212 instance of the 'standard' template library not being so standard".
3213
3214 10. There was no check on the number of named subpatterns nor the maximum
3215 length of a subpattern name. The product of these values is used to compute
3216 the size of the memory block for a compiled pattern. By supplying a very
3217 long subpattern name and a large number of named subpatterns, the size
3218 computation could be caused to overflow. This is now prevented by limiting
3219 the length of names to 32 characters, and the number of named subpatterns
3220 to 10,000.
3221
3222 11. Subpatterns that are repeated with specific counts have to be replicated in
3223 the compiled pattern. The size of memory for this was computed from the
3224 length of the subpattern and the repeat count. The latter is limited to
3225 65535, but there was no limit on the former, meaning that integer overflow
3226 could in principle occur. The compiled length of a repeated subpattern is
3227 now limited to 30,000 bytes in order to prevent this.
3228
3229 12. Added the optional facility to have named substrings with the same name.
3230
3231 13. Added the ability to use a named substring as a condition, using the
3232 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3233 are numbers (not recommended). Forward references are permitted.
3234
3235 14. Added forward references in named backreferences (if you see what I mean).
3236
3237 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3238 pattern could run off the end of the subject. For example, the pattern
3239 "(?s)(.{1,5})"8 did this with the subject "ab".
3240
3241 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3242 PCRE_CASELESS was set when matching characters that were quantified with ?
3243 or *.
3244
3245 17. A character class other than a single negated character that had a minimum
3246 but no maximum quantifier - for example [ab]{6,} - was not handled
3247 correctly by pce_dfa_exec(). It would match only one character.
3248
3249 18. A valid (though odd) pattern that looked like a POSIX character
3250 class but used an invalid character after [ (for example [[,abc,]]) caused
3251 pcre_compile() to give the error "Failed: internal error: code overflow" or
3252 in some cases to crash with a glibc free() error. This could even happen if
3253 the pattern terminated after [[ but there just happened to be a sequence of
3254 letters, a binary zero, and a closing ] in the memory that followed.
3255
3256 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3257 over the years. Originally (before any Unicode support), just the bottom 8
3258 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3259 output from "man perlunicode" includes this:
3260
3261 The regular expression compiler produces polymorphic opcodes. That
3262 is, the pattern adapts to the data and automatically switches to
3263 the Unicode character scheme when presented with Unicode data--or
3264 instead uses a traditional byte scheme when presented with byte
3265 data.
3266
3267 Sadly, a wide octal escape does not cause a switch, and in a string with
3268 no other multibyte characters, these octal escapes are treated as before.
3269 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3270 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3271 Unicode string.
3272
3273 I have not perpetrated such confusion in PCRE. Up till now, it took just
3274 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3275 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3276 translate to the appropriate multibyte character.
3277
3278 29. Applied some refactoring to reduce the number of warnings from Microsoft
3279 and Borland compilers. This has included removing the fudge introduced
3280 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3281 a warning about an unused variable.
3282
3283 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3284 characters since release 4.0, because Perl (from release 5.004) does not.
3285 [Or at least, is documented not to: some releases seem to be in conflict
3286 with the documentation.] However, when a pattern was studied with
3287 pcre_study() and all its branches started with \s, PCRE still included VT
3288 as a possible starting character. Of course, this did no harm; it just
3289 caused an unnecessary match attempt.
3290
3291 22. Removed a now-redundant internal flag bit that recorded the fact that case
3292 dependency changed within the pattern. This was once needed for "required
3293 byte" processing, but is no longer used. This recovers a now-scarce options
3294 bit. Also moved the least significant internal flag bit to the most-
3295 significant bit of the word, which was not previously used (hangover from
3296 the days when it was an int rather than a uint) to free up another bit for
3297 the future.
3298
3299 23. Added support for CRLF line endings as well as CR and LF. As well as the
3300 default being selectable at build time, it can now be changed at runtime
3301 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3302 specify that it is scanning data with non-default line endings.
3303
3304 24. Changed the definition of CXXLINK to make it agree with the definition of
3305 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3306
3307 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3308 recursions. This makes a big different to stack usage for some patterns.
3309
3310 26. If a subpattern containing a named recursion or subroutine reference such
3311 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3312 the space required for the compiled pattern went wrong and gave too small a
3313 value. Depending on the environment, this could lead to "Failed: internal
3314 error: code overflow at offset 49" or "glibc detected double free or
3315 corruption" errors.
3316
3317 27. Applied patches from Google (a) to support the new newline modes and (b) to
3318 advance over multibyte UTF-8 characters in GlobalReplace.
3319
3320 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3321 difference for some implementation of PCRE in some Windows version.
3322
3323 29. Added some extra testing facilities to pcretest:
3324
3325 \q<number> in a data line sets the "match limit" value
3326 \Q<number> in a data line sets the "match recursion limt" value
3327 -S <number> sets the stack size, where <number> is in megabytes
3328
3329 The -S option isn't available for Windows.
3330
3331
3332 Version 6.6 06-Feb-06
3333 ---------------------
3334
3335 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3336 in pcreposix.h. I have copied the definition from pcre.h.
3337
3338 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3339 because pcre.h is no longer a built file.
3340
3341 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3342 not normally included in the compiled code.
3343
3344
3345 Version 6.5 01-Feb-06
3346 ---------------------
3347
3348 1. When using the partial match feature with pcre_dfa_exec(), it was not
3349 anchoring the second and subsequent partial matches at the new starting
3350 point. This could lead to incorrect results. For example, with the pattern
3351 /1234/, partially matching against "123" and then "a4" gave a match.
3352
3353 2. Changes to pcregrep:
3354
3355 (a) All non-match returns from pcre_exec() were being treated as failures
3356 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3357 error message is output. Some extra information is given for the
3358 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3359 probably the only errors that are likely to be caused by users (by
3360 specifying a regex that has nested indefinite repeats, for instance).
3361 If there are more than 20 of these errors, pcregrep is abandoned.
3362
3363 (b) A binary zero was treated as data while matching, but terminated the
3364 output line if it was written out. This has been fixed: binary zeroes
3365 are now no different to any other data bytes.
3366
3367 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3368 used to set a locale for matching. The --locale=xxxx long option has
3369 been added (no short equivalent) to specify a locale explicitly on the
3370 pcregrep command, overriding the environment variables.
3371
3372 (d) When -B was used with -n, some line numbers in the output were one less
3373 than they should have been.
3374
3375 (e) Added the -o (--only-matching) option.
3376
3377 (f) If -A or -C was used with -c (count only), some lines of context were
3378 accidentally printed for the final match.
3379
3380 (g) Added the -H (--with-filename) option.
3381
3382 (h) The combination of options -rh failed to suppress file names for files
3383 that were found from directory arguments.
3384
3385 (i) Added the -D (--devices) and -d (--directories) options.
3386
3387 (j) Added the -F (--fixed-strings) option.
3388
3389 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3390
3391 (l) Added the --colo(u)r option.
3392
3393 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3394 is not present by default.
3395
3396 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3397 items such as (?R) or (?1), when the recursion could match a number of
3398 alternatives. If it matched one of the alternatives, but subsequently,
3399 outside the recursion, there was a failure, the code tried to back up into
3400 the recursion. However, because of the way PCRE is implemented, this is not
3401 possible, and the result was an incorrect result from the match.
3402
3403 In order to prevent this happening, the specification of recursion has
3404 been changed so that all such subpatterns are automatically treated as
3405 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3406
3407 4. I had overlooked the fact that, in some locales, there are characters for
3408 which isalpha() is true but neither isupper() nor islower() are true. In
3409 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3410 and ordfeminine) are like this. This affected the treatment of \w and \W
3411 when they appeared in character classes, but not when they appeared outside
3412 a character class. The bit map for "word" characters is now created
3413 separately from the results of isalnum() instead of just taking it from the
3414 upper, lower, and digit maps. (Plus the underscore character, of course.)
3415
3416 5. The above bug also affected the handling of POSIX character classes such as
3417 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3418 permanent tables. Instead, the bit maps for such a class were previously
3419 created as the appropriate unions of the upper, lower, and digit bitmaps.
3420 Now they are created by subtraction from the [[:word:]] class, which has
3421 its own bitmap.
3422
3423 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3424 It is created by subtracting the vertical space characters (\x09, \x0a,
3425 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3426 subtraction was done in the overall bitmap for a character class, meaning
3427 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3428 be recognized. This bug has been fixed.
3429
3430 7. Patches from the folks at Google:
3431
3432 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3433 real life, but is still worth protecting against".
3434
3435 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3436 regular expressions".
3437
3438 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3439 have it.
3440
3441 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3442 "configure" and the latter not, in order to fix a problem somebody had
3443 with compiling the Arg class on HP-UX.
3444
3445 (e) Improve the error-handling of the C++ wrapper a little bit.
3446
3447 (f) New tests for checking recursion limiting.
3448
3449 8. The pcre_memmove() function, which is used only if the environment does not
3450 have a standard memmove() function (and is therefore rarely compiled),
3451 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3452 returning a result (though PCRE never actually uses the result).
3453
3454 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3455 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3456 returned instead of calling malloc() with an overflowing number that would
3457 most likely cause subsequent chaos.
3458
3459 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3460
3461 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3462 with this option is matched, the nmatch and pmatch options of regexec() are
3463 ignored.
3464
3465 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3466 provided in case anyone wants to the the POSIX interface with UTF-8
3467 strings.
3468
3469 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3470 C++ linking (needed for some HP-UX environments).
3471
3472 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3473 (unused parameter) and in the pcre_printint() function (omitted "default"
3474 switch label when the default is to do nothing).
3475
3476 15. Added some code to make it possible, when PCRE is compiled as a C++
3477 library, to replace subject pointers for pcre_exec() with a smart pointer
3478 class, thus making it possible to process discontinuous strings.
3479
3480 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3481 much the same function. They were added by different people who were trying
3482 to make PCRE easy to compile on non-Unix systems. It has been suggested
3483 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3484 for compiling on Windows systems. I have therefore replaced it with
3485 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3486 defaults to "extern" for C or "extern C" for C++, which works fine on
3487 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3488 SCOPE with something explicit in config.h. In addition:
3489
3490 (a) pcreposix.h still had just "extern" instead of either of these macros;
3491 I have replaced it with PCRE_DATA_SCOPE.
3492
3493 (b) Functions such as _pcre_xclass(), which are internal to the library,
3494 but external in the C sense, all had PCRE_EXPORT in their definitions.
3495 This is apparently wrong for the Windows case, so I have removed it.
3496 (It makes no difference on Unix-like systems.)
3497
3498 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3499 of recursive calls to match(). This is different to MATCH_LIMIT because
3500 that limits the total number of calls to match(), not all of which increase
3501 the depth of recursion. Limiting the recursion depth limits the amount of
3502 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3503 when PCRE is compiled, and changed at run time. A patch from Google adds
3504 this functionality to the C++ interface.
3505
3506 18. Changes to the handling of Unicode character properties:
3507
3508 (a) Updated the table to Unicode 4.1.0.
3509
3510 (b) Recognize characters that are not in the table as "Cn" (undefined).
3511
3512 (c) I revised the way the table is implemented to a much improved format
3513 which includes recognition of ranges. It now supports the ranges that
3514 are defined in UnicodeData.txt, and it also amalgamates other
3515 characters into ranges. This has reduced the number of entries in the
3516 table from around 16,000 to around 3,000, thus reducing its size
3517 considerably. I realized I did not need to use a tree structure after
3518 all - a binary chop search is just as efficient. Having reduced the
3519 number of entries, I extended their size from 6 bytes to 8 bytes to
3520 allow for more data.
3521
3522 (d) Added support for Unicode script names via properties such as \p{Han}.
3523
3524 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3525 matching that character.
3526
3527 20. When matching a repeated Unicode property with a minimum greater than zero,
3528 (for example \pL{2,}), PCRE could look past the end of the subject if it
3529 reached it while seeking the minimum number of characters. This could
3530 happen only if some of the characters were more than one byte long, because
3531 there is a check for at least the minimum number of bytes.
3532
3533 21. Refactored the implementation of \p and \P so as to be more general, to
3534 allow for more different types of property in future. This has changed the
3535 compiled form incompatibly. Anybody with saved compiled patterns that use
3536 \p or \P will have to recompile them.
3537
3538 22. Added "Any" and "L&" to the supported property types.
3539
3540 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3541 but give a compile time error if the value is greater than 0xff.
3542
3543 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3544 accidentally not being installed or uninstalled.
3545
3546 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3547 made were to insert the current release number. This seemed silly, because
3548 it made things harder for people building PCRE on systems that don't run
3549 "configure". I have turned pcre.h into a distributed file, no longer built
3550 by "configure", with the version identification directly included. There is
3551 no longer a pcre.h.in file.
3552
3553 However, this change necessitated a change to the pcre-config script as
3554 well. It is built from pcre-config.in, and one of the substitutions was the
3555 release number. I have updated configure.ac so that ./configure now finds
3556 the release number by grepping pcre.h.
3557
3558 26. Added the ability to run the tests under valgrind.
3559
3560
3561 Version 6.4 05-Sep-05
3562 ---------------------
3563
3564 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3565 "--" to be printed when multiple files were scanned, even when none of the
3566 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3567 consider it to be a bug, and have restored the previous behaviour.
3568
3569 2. A couple of code tidies to get rid of compiler warnings.
3570
3571 3. The pcretest program used to cheat by referring to symbols in the library
3572 whose names begin with _pcre_. These are internal symbols that are not
3573 really supposed to be visible externally, and in some environments it is
3574 possible to suppress them. The cheating is now confined to including
3575 certain files from the library's source, which is a bit cleaner.
3576
3577 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3578 file's purpose clearer.
3579
3580 5. Reorganized pcre_ucp_findchar().
3581
3582
3583 Version 6.3 15-Aug-05
3584 ---------------------
3585
3586 1. The file libpcre.pc.in did not have general read permission in the tarball.
3587
3588 2. There were some problems when building without C++ support:
3589
3590 (a) If C++ support was not built, "make install" and "make test" still
3591 tried to test it.
3592
3593 (b) There were problems when the value of CXX was explicitly set. Some
3594 changes have been made to try to fix these, and ...
3595
3596 (c) --disable-cpp can now be used to explicitly disable C++ support.
3597
3598 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3599 backslash in a target when C++ was disabled. This confuses some
3600 versions of "make", apparently. Using an intermediate variable solves
3601 this. (Same for CPP_LOBJ.)
3602
3603 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3604 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3605 necessary on certain architectures.
3606
3607 4. Added a setting of -export-symbols-regex to the link command to remove
3608 those symbols that are exported in the C sense, but actually are local
3609 within the library, and not documented. Their names all begin with
3610 "_pcre_". This is not a perfect job, because (a) we have to except some
3611 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3612 available (and never for static libraries). I have made a note to try to
3613 find a way round (a) in the future.
3614
3615
3616 Version 6.2 01-Aug-05
3617 ---------------------
3618
3619 1. There was no test for integer overflow of quantifier values. A construction
3620 such as {1111111111111111} would give undefined results. What is worse, if
3621 a minimum quantifier for a parenthesized subpattern overflowed and became
3622 negative, the calculation of the memory size went wrong. This could have
3623 led to memory overwriting.
3624
3625 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3626
3627 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3628 operating environments where this matters.
3629
3630 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3631 PCRE options from within the C++ wrapper.
3632
3633 5. Named capturing subpatterns were not being correctly counted when a pattern
3634 was compiled. This caused two problems: (a) If there were more than 100
3635 such subpatterns, the calculation of the memory needed for the whole
3636 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3637 back references of the form \12, where the number was greater than 9, were
3638 not recognized as back references, even though there were sufficient
3639 previous subpatterns.
3640
3641 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3642 versions of gcc, e.g. 2.95.4.
3643
3644
3645 Version 6.1 21-Jun-05
3646 ---------------------
3647
3648 1. There was one reference to the variable "posix" in pcretest.c that was not
3649 surrounded by "#if !defined NOPOSIX".
3650
3651 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3652 the cross-check on the old pcre_info() function, for the benefit of the
3653 cut-down version of PCRE that is currently imported into Exim.
3654
3655 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3656 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3657 patterns that start (?i) but I don't think that matters. The use of (?i) is
3658 just an example; this all applies to the other options as well.
3659
3660 4. Since libtool seems to echo the compile commands it is issuing, the output
3661 from "make" can be reduced a bit by putting "@" in front of each libtool
3662 compile command.
3663
3664 5. Patch from the folks at Google for configure.in to be a bit more thorough
3665 in checking for a suitable C++ installation before trying to compile the
3666 C++ stuff. This should fix a reported problem when a compiler was present,
3667 but no suitable headers.
3668
3669 6. The man pages all had just "PCRE" as their title. I have changed them to
3670 be the relevant file name. I have also arranged that these names are
3671 retained in the file doc/pcre.txt, which is a concatenation in text format
3672 of all the man pages except the little individual ones for each function.
3673
3674 7. The NON-UNIX-USE file had not been updated for the different set of source
3675 files that come with release 6. I also added a few comments about the C++
3676 wrapper.
3677
3678
3679 Version 6.0 07-Jun-05
3680 ---------------------
3681
3682 1. Some minor internal re-organization to help with my DFA experiments.
3683
3684 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3685 didn't matter for the library itself when fully configured, but did matter
3686 when compiling without UCP support, or within Exim, where the ucp files are
3687 not imported.
3688
3689 3. Refactoring of the library code to split up the various functions into
3690 different source modules. The addition of the new DFA matching code (see
3691 below) to a single monolithic source would have made it really too
3692 unwieldy, quite apart from causing all the code to be include in a
3693 statically linked application, when only some functions are used. This is
3694 relevant even without the DFA addition now that patterns can be compiled in
3695 one application and matched in another.
3696
3697 The downside of splitting up is that there have to be some external
3698 functions and data tables that are used internally in different modules of
3699 the library but which are not part of the API. These have all had their
3700 names changed to start with "_pcre_" so that they are unlikely to clash
3701 with other external names.
3702
3703 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3704 a different (DFA) algorithm. Although it is slower than the original
3705 function, it does have some advantages for certain types of matching
3706 problem.
3707
3708 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3709 including restarting after a partial match.
3710
3711 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3712 defined when compiling for Windows was sent to me. I have put it into the
3713 code, though I have no means of testing or verifying it.
3714
3715 7. Added the pcre_refcount() auxiliary function.
3716
3717 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3718 match before or at the first newline in the subject string. In pcretest,
3719 the /f option on a pattern can be used to set this.
3720
3721 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3722 would behave wrongly. This has been present in PCRE since release 4.0.
3723
3724 10. A number of changes to the pcregrep command:
3725
3726 (a) Refactored how -x works; insert ^(...)$ instead of setting
3727 PCRE_ANCHORED and checking the length, in preparation for adding
3728 something similar for -w.
3729
3730 (b) Added the -w (match as a word) option.
3731
3732 (c) Refactored the way lines are read and buffered so as to have more
3733 than one at a time available.
3734
3735 (d) Implemented a pcregrep test script.
3736
3737 (e) Added the -M (multiline match) option. This allows patterns to match
3738 over several lines of the subject. The buffering ensures that at least
3739 8K, or the rest of the document (whichever is the shorter) is available
3740 for matching (and similarly the previous 8K for lookbehind assertions).
3741
3742 (f) Changed the --help output so that it now says
3743
3744 -w, --word-regex(p)
3745
3746 instead of two lines, one with "regex" and the other with "regexp"
3747 because that confused at least one person since the short forms are the
3748 same. (This required a bit of code, as the output is generated
3749 automatically from a table. It wasn't just a text change.)
3750
3751 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3752 option but starts with a hyphen. Could be a pattern or a path name
3753 starting with a hyphen, for instance.
3754
3755 (h) "-" can be given as a file name to represent stdin.
3756
3757 (i) When file names are being printed, "(standard input)" is used for
3758 the standard input, for compatibility with GNU grep. Previously
3759 "<stdin>" was used.
3760
3761 (j) The option --label=xxx can be used to supply a name to be used for
3762 stdin when file names are being printed. There is no short form.
3763
3764 (k) Re-factored the options decoding logic because we are going to add
3765 two more options that take data. Such options can now be given in four
3766 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3767
3768 (l) Added the -A, -B, and -C options for requesting that lines of context
3769 around matches be printed.
3770
3771 (m) Added the -L option to print the names of files that do not contain
3772 any matching lines, that is, the complement of -l.
3773
3774 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3775 continue to scan other files.
3776
3777 (o) The -s option was incorrectly implemented. For compatibility with other
3778 greps, it now suppresses the error message for a non-existent or non-
3779 accessible file (but not the return code). There is a new option called
3780 -q that suppresses the output of matching lines, which was what -s was
3781 previously doing.
3782
3783 (p) Added --include and --exclude options to specify files for inclusion
3784 and exclusion when recursing.
3785
3786 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3787 Hopefully, it now does.
3788
3789 12. Missing cast in pcre_study().
3790
3791 13. Added an "uninstall" target to the makefile.
3792
3793 14. Replaced "extern" in the function prototypes in Makefile.in with
3794 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3795 world, but is set differently for Windows.
3796
3797 15. Added a second compiling function called pcre_compile2(). The only
3798 difference is that it has an extra argument, which is a pointer to an
3799 integer error code. When there is a compile-time failure, this is set
3800 non-zero, in addition to the error test pointer being set to point to an
3801 error message. The new argument may be NULL if no error number is required
3802 (but then you may as well call pcre_compile(), which is now just a
3803 wrapper). This facility is provided because some applications need a
3804 numeric error indication, but it has also enabled me to tidy up the way
3805 compile-time errors are handled in the POSIX wrapper.
3806
3807 16. Added VPATH=.libs to the makefile; this should help when building with one
3808 prefix path and installing with another. (Or so I'm told by someone who
3809 knows more about this stuff than I do.)
3810
3811 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3812 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3813 match everything, including newlines. This is not POSIX-compatible, but
3814 somebody wanted the feature. From pcretest it can be activated by using
3815 both the P and the s flags.
3816
3817 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3818
3819 19. libpcre.pc was being incorrectly installed as executable.
3820
3821 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3822 it now also looks for '\r' so that it will work unmodified on Windows.
3823
3824 21. Added Google's contributed C++ wrapper to the distribution.
3825
3826 22. Added some untidy missing memory free() calls in pcretest, to keep
3827 Electric Fence happy when testing.
3828
3829
3830
3831 Version 5.0 13-Sep-04
3832 ---------------------
3833
3834 1. Internal change: literal characters are no longer packed up into items
3835 containing multiple characters in a single byte-string. Each character
3836 is now matched using a separate opcode. However, there may be more than one
3837 byte in the character in UTF-8 mode.
3838
3839 2. The pcre_callout_block structure has two new fields: pattern_position and
3840 next_item_length. These contain the offset in the pattern to the next match
3841 item, and its length, respectively.
3842
3843 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3844 insertion of callouts before each pattern item. Added the /C option to
3845 pcretest to make use of this.
3846
3847 4. On the advice of a Windows user, the lines
3848
3849 #if defined(_WIN32) || defined(WIN32)
3850 _setmode( _fileno( stdout ), 0x8000 );
3851 #endif /* defined(_WIN32) || defined(WIN32) */
3852
3853 have been added to the source of pcretest. This apparently does useful
3854 magic in relation to line terminators.
3855
3856 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3857 for the benefit of those environments where the "b" makes a difference.
3858
3859 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3860 to know about it. I have put a hack into configure.in that adds in code
3861 to set GCC=yes if CC=icc. This seems to end up at a point in the
3862 generated configure script that is early enough to affect the setting of
3863 compiler options, which is what is needed, but I have no means of testing
3864 whether it really works. (The user who reported this had patched the
3865 generated configure script, which of course I cannot do.)
3866
3867 LATER: After change 22 below (new libtool files), the configure script
3868 seems to know about icc (and also ecc). Therefore, I have commented out
3869 this hack in configure.in.
3870
3871 7. Added support for pkg-config (2 patches were sent in).
3872
3873 8. Negated POSIX character classes that used a combination of internal tables
3874 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3875 [[:^ascii]]. Typically, they would match almost any characters. The other
3876 POSIX classes were not broken in this way.
3877
3878 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3879 to find the match, as PCRE was deluded into thinking that the match had to
3880 start at the start point or following a newline. The same bug applied to
3881 patterns with negative forward assertions or any backward assertions
3882 preceding ".*" at the start, unless the pattern required a fixed first
3883 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3884
3885 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3886 starting at the last subject character, bytes beyond the end of the subject
3887 string were read.
3888
3889 11. Renamed the variable "class" as "classbits" to make life easier for C++
3890 users. (Previously there was a macro definition, but it apparently wasn't
3891 enough.)
3892
3893 12. Added the new field "tables" to the extra data so that tables can be passed
3894 in at exec time, or the internal tables can be re-selected. This allows
3895 a compiled regex to be saved and re-used at a later time by a different
3896 program that might have everything at different addresses.
3897
3898 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3899 -R library as well as a -L library.
3900
3901 14. The debugging options of pcretest (-d on the command line or D on a
3902 pattern) showed incorrect output for anything following an extended class
3903 that contained multibyte characters and which was followed by a quantifier.
3904
3905 15. Added optional support for general category Unicode character properties
3906 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3907 support. It adds about 90K to the size of the library. The meanings of the
3908 inbuilt class escapes such as \d and \s have NOT been changed.
3909
3910 16. Updated pcredemo.c to include calls to free() to release the memory for the
3911 compiled pattern.
3912
3913 17. The generated file chartables.c was being created in the source directory
3914 instead of in the building directory. This caused the build to fail if the
3915 source directory was different from the building directory, and was
3916 read-only.
3917
3918 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3919 file. No doubt somebody will tell me if they don't make sense... Also added
3920 Dan Mooney's comments about building on OpenVMS.
3921
3922 19. Added support for partial matching via the PCRE_PARTIAL option for
3923 pcre_exec() and the \P data escape in pcretest.
3924
3925 20. Extended pcretest with 3 new pattern features:
3926
3927 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3928 write the compiled pattern to the file whose name is "rest-of-line".
3929 This is a straight binary dump of the data, with the saved pointer to
3930 the character tables forced to be NULL. The study data, if any, is
3931 written too. After writing, pcretest reads a new pattern.
3932
3933 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3934 compiled pattern from the given file. There must not be any
3935 occurrences of "<" in the file name (pretty unlikely); if there are,
3936 pcretest will instead treat the initial "<" as a pattern delimiter.
3937 After reading in the pattern, pcretest goes on to read data lines as
3938 usual.
3939
3940 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3941 and 16-bit fields in a compiled pattern, to simulate a pattern that
3942 was compiled on a host of opposite endianness.
3943
3944 21. The pcre-exec() function can now cope with patterns that were compiled on
3945 hosts of opposite endianness, with this restriction:
3946
3947 As for any compiled expression that is saved and used later, the tables
3948 pointer field cannot be preserved; the extra_data field in the arguments
3949 to pcre_exec() should be used to pass in a tables address if a value
3950 other than the default internal tables were used at compile time.
3951
3952 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3953 now diagnosed as an error. Previously, most of the time, a negative number
3954 would have been treated as zero, but if in addition "ovector" was passed as
3955 NULL, a crash could occur.
3956
3957 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3958 new versions from the libtool 1.5 distribution (the last one is a copy of
3959 a file called libtool.m4). This seems to have fixed the need to patch
3960 "configure" to support Darwin 1.3 (which I used to do). However, I still
3961 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3962 workstation).
3963
3964 24. Changed the PCRE licence to be the more standard "BSD" licence.
3965
3966
3967 Version 4.5 01-Dec-03
3968 ---------------------
3969
3970 1. There has been some re-arrangement of the code for the match() function so
3971 that it can be compiled in a version that does not call itself recursively.
3972 Instead, it keeps those local variables that need separate instances for
3973 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3974 needs to "recurse". Keeping track of where control must go is done by means
3975 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3976 hide most of the details from the main code, and operates only if
3977 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3978 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3979 operating.
3980
3981 To make it easier for callers to provide specially tailored get/free
3982 functions for this usage, two new functions, pcre_stack_malloc, and
3983 pcre_stack_free, are used. They are always called in strict stacking order,
3984 and the size of block requested is always the same.
3985
3986 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3987 PCRE has been compiled to use the stack or the heap for recursion. The
3988 -C option of pcretest uses this to show which version is compiled.
3989
3990 A new data escape \S, is added to pcretest; it causes the amounts of store
3991 obtained and freed by both kinds of malloc/free at match time to be added
3992 to the output.
3993
3994 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3995 what's available on my current Linux desktop machine.
3996
3997 3. When matching a UTF-8 string, the test for a valid string at the start has
3998 been extended. If start_offset is not zero, PCRE now checks that it points
3999 to a byte that is the start of a UTF-8 character. If not, it returns
4000 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4001 this is necessary because there may be backward assertions in the pattern.
4002 When matching the same subject several times, it may save resources to use
4003 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4004
4005 4. The code for checking the validity of UTF-8 strings has been tightened so
4006 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4007 containing "overlong sequences".
4008
4009 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4010 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4011 should have been "&", but it just so happened that all the cases this let
4012 through by mistake were picked up later in the function.
4013
4014 6. I had used a variable called "isblank" - this is a C99 function, causing
4015 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4016
4017 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4018 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4019 so the version is shown; (c) stop "make test" echoing "./RunTest".
4020
4021 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4022
4023 9. The prototype for memmove() for systems that don't have it was using
4024 size_t, but the inclusion of the header that defines size_t was later. I've
4025 moved the #includes for the C headers earlier to avoid this.
4026
4027 10. Added some adjustments to the code to make it easier to compiler on certain
4028 special systems:
4029
4030 (a) Some "const" qualifiers were missing.
4031 (b) Added the macro EXPORT before all exported functions; by default this
4032 is defined to be empty.
4033 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4034 that it reads its output file name as an argument instead of writing
4035 to the standard output and assuming this can be redirected.
4036
4037 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4038 class containing characters with values greater than 255, PCRE compilation
4039 went into a loop.
4040
4041 12. A recursive reference to a subpattern that was within another subpattern
4042 that had a minimum quantifier of zero caused PCRE to crash. For example,
4043 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4044 recursion. If the recursively-called subpattern itself had a zero repeat,
4045 that was OK.
4046
4047 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4048 buffer into which it was copied (for escape processing) was still set at
4049 1024, so long lines caused crashes.
4050
4051 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4052 "internal error: code overflow...". This applied to any character class
4053 that was followed by a possessive quantifier.
4054
4055 15. Modified the Makefile to add libpcre.la as a prerequisite for
4056 libpcreposix.la because I was told this is needed for a parallel build to
4057 work.
4058
4059 16. If a pattern that contained .* following optional items at the start was
4060 studied, the wrong optimizing data was generated, leading to matching
4061 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4062 matching string must start with a or b or c. The correct conclusion for
4063 this pattern is that a match can start with any character.
4064
4065
4066 Version 4.4 13-Aug-03
4067 ---------------------
4068
4069 1. In UTF-8 mode, a character class containing characters with values between
4070 127 and 255 was not handled correctly if the compiled pattern was studied.
4071 In fixing this, I have also improved the studying algorithm for such
4072 classes (slightly).
4073
4074 2. Three internal functions had redundant arguments passed to them. Removal
4075 might give a very teeny performance improvement.
4076
4077 3. Documentation bug: the value of the capture_top field in a callout is *one
4078 more than* the number of the hightest numbered captured substring.
4079
4080 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4081 in incorrectly linking with a previously installed version. They now link
4082 explicitly with libpcre.la.
4083
4084 5. configure.in no longer needs to recognize Cygwin specially.
4085
4086 6. A problem in pcre.in for Windows platforms is fixed.
4087
4088 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4089 pcretest, it used to include the size of the study block as part of its
4090 output. Unfortunately, the structure contains a field that has a different
4091 size on different hardware architectures. This meant that the tests that
4092 showed this size failed. As the block is currently always of a fixed size,
4093 this information isn't actually particularly useful in pcretest output, so
4094 I have just removed it.
4095
4096 8. Three pre-processor statements accidentally did not start in column 1.
4097 Sadly, there are *still* compilers around that complain, even though
4098 standard C has not required this for well over a decade. Sigh.
4099
4100 9. In pcretest, the code for checking callouts passed small integers in the
4101 callout_data field, which is a void * field. However, some picky compilers
4102 complained about the casts involved for this on 64-bit systems. Now
4103 pcretest passes the address of the small integer instead, which should get
4104 rid of the warnings.
4105
4106 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4107 both compile and run time, and gives an error if an invalid UTF-8 sequence
4108 is found. There is a option for disabling this check in cases where the
4109 string is known to be correct and/or the maximum performance is wanted.
4110
4111 11. In response to a bug report, I changed one line in Makefile.in from
4112
4113 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4114 to
4115 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4116
4117 to look similar to other lines, but I have no way of telling whether this
4118 is the right thing to do, as I do not use Windows. No doubt I'll get told
4119 if it's wrong...
4120
4121
4122 Version 4.3 21-May-03
4123 ---------------------
4124
4125 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4126 Makefile.
4127
4128 2. Some refactoring to improve the quality of the code:
4129
4130 (i) The utf8_table... variables are now declared "const".
4131
4132 (ii) The code for \cx, which used the "case flipping" table to upper case
4133 lower case letters, now just substracts 32. This is ASCII-specific,
4134 but the whole concept of \cx is ASCII-specific, so it seems
4135 reasonable.
4136
4137 (iii) PCRE was using its character types table to recognize decimal and
4138 hexadecimal digits in the pattern. This is silly, because it handles
4139 only 0-9, a-f, and A-F, but the character types table is locale-
4140 specific, which means strange things might happen. A private
4141 table is now used for this - though it costs 256 bytes, a table is
4142 much faster than multiple explicit tests. Of course, the standard
4143 character types table is still used for matching digits in subject
4144 strings against \d.
4145
4146 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4147 ending in _t are). So I've renamed it as ESC_tee.
4148
4149 3. The first argument for regexec() in the POSIX wrapper should have been
4150 defined as "const".
4151
4152 4. Changed pcretest to use malloc() for its buffers so that they can be
4153 Electric Fenced for debugging.
4154
4155 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4156 to read one or more bytes before the start of the subject string. Often this
4157 had no effect on PCRE's behaviour, but in some circumstances it could
4158 provoke a segmentation fault.
4159
4160 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4161 to try to read one or more bytes before the start of the subject string.
4162
4163 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4164 UTF-8 support could misbehave in various ways if the subject string
4165 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4166 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4167 back over UTF-8 characters.)
4168
4169
4170 Version 4.2 14-Apr-03
4171 ---------------------
4172
4173 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4174
4175 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4176 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4177 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4178 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4179 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4180 and BUILD_EXEEXT
4181 Note: automatic setting of the BUILD variables is not yet working
4182 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4183 compile-time but not at link-time
4184 [LINK]: use for linking executables only
4185 make different versions for Windows and non-Windows
4186 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4187 libraries
4188 [LINK_FOR_BUILD]: new variable
4189 [OBJEXT]: use throughout
4190 [EXEEXT]: use throughout
4191 <winshared>: new target
4192 <wininstall>: new target
4193 <dftables.o>: use native compiler
4194 <dftables>: use native linker
4195 <install>: handle Windows platform correctly
4196 <clean>: ditto
4197 <check>: ditto
4198 copy DLL to top builddir before testing
4199
4200 As part of these changes, -no-undefined was removed again. This was reported
4201 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4202 in any case.
4203
4204 3. Some tidies to get rid of compiler warnings:
4205
4206 . In the match_data structure, match_limit was an unsigned long int, whereas
4207 match_call_count was an int. I've made them both unsigned long ints.
4208
4209 . In pcretest the fact that a const uschar * doesn't automatically cast to
4210 a void * provoked a warning.
4211
4212 . Turning on some more compiler warnings threw up some "shadow" variables
4213 and a few more missing casts.
4214
4215 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4216 option, a class that contained a single character with a value between 128
4217 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4218
4219 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4220 option, a class that contained several characters, but with at least one
4221 whose value was between 128 and 255 caused PCRE to crash.
4222
4223
4224 Version 4.1 12-Mar-03
4225 ---------------------
4226
4227 1. Compiling with gcc -pedantic found a couple of places where casts were
4228 needed, and a string in dftables.c that was longer than standard compilers are
4229 required to support.
4230
4231 2. Compiling with Sun's compiler found a few more places where the code could
4232 be tidied up in order to avoid warnings.
4233
4234 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4235 first of these names is deprecated in the latest Autoconf in favour of the name
4236 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4237 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4238 analogy I have changed it to CFLAGS_FOR_BUILD.
4239
4240 4. Added -no-undefined to the linking command in the Makefile, because this is
4241 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4242 linking step for the pcreposix library.
4243
4244 5. PCRE was failing to diagnose the case of two named groups with the same
4245 name.
4246
4247 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4248 literal character that is needed in the subject for a match, and scans along to
4249 ensure that it is present before embarking on the full matching process. This
4250 saves time in cases of nested unlimited repeats that are never going to match.
4251 Problem: the scan can take a lot of time if the subject is very long (e.g.
4252 megabytes), thus penalizing straightforward matches. It is now done only if the
4253 amount of subject to be scanned is less than 1000 bytes.
4254
4255 7. A lesser problem with the same optimization is that it was recording the
4256 first character of an anchored pattern as "needed", thus provoking a search
4257 right along the subject, even when the first match of the pattern was going to
4258 fail. The "needed" character is now not set for anchored patterns, unless it
4259 follows something in the pattern that is of non-fixed length. Thus, it still
4260 fulfils its original purpose of finding quick non-matches in cases of nested
4261 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4262
4263
4264 Version 4.0 17-Feb-03
4265 ---------------------
4266
4267 1. If a comment in an extended regex that started immediately after a meta-item
4268 extended to the end of string, PCRE compiled incorrect data. This could lead to
4269 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4270
4271 2. Moved to autoconf 2.53 and libtool 1.4.2.
4272
4273 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4274 the special perltest8 script is no longer needed - all the tests can be run
4275 from a single perltest script.
4276
4277 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4278 by \s. It has now been removed in PCRE. This means it isn't recognized as
4279 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4280 class [:space:] *does* include VT, thereby creating a mess.
4281
4282 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4283 space and tab.
4284
4285 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4286 its new features into the main test script, reducing the number of scripts.
4287
4288 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4289 were backward compatible, and made the (?i) apply to the whole pattern, as if
4290 /i were given. Now it behaves more logically, and applies the option setting
4291 only to what follows. PCRE has been changed to follow suit. However, if it
4292 finds options settings right at the start of the pattern, it extracts them into
4293 the global options, as before. Thus, they show up in the info data.
4294
4295 8. Added support for the \Q...\E escape sequence. Characters in between are
4296 treated as literals. This is slightly different from Perl in that $ and @ are
4297 also handled as literals inside the quotes. In Perl, they will cause variable
4298 interpolation. Note the following examples:
4299
4300 Pattern PCRE matches Perl matches
4301
4302 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4303 \Qabc\$xyz\E abc\$xyz abc\$xyz
4304 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4305
4306 For compatibility with Perl, \Q...\E sequences are recognized inside character
4307 classes as well as outside them.
4308
4309 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4310 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4311 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4312 signed/unsigned warnings.
4313
4314 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4315 option for pcretest, so I've replaced it by a simple function that does just
4316 that job.
4317
4318 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4319 "pcregrep -".
4320
4321 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4322 Java package. This provides some syntactic sugar for simple cases of what my
4323 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4324 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4325 item, you can use this simplified notation. Note that only makes sense with
4326 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4327 greediness, whatever the setting of the PCRE_UNGREEDY option.
4328
4329 13. A change of greediness default within a pattern was not taking effect at
4330 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4331 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4332 was abstracted outside.
4333
4334 14. PCRE now supports the \G assertion. It is true when the current matching
4335 position is at the start point of the match. This differs from \A when the
4336 starting offset is non-zero. Used with the /g option of pcretest (or similar
4337 code), it works in the same way as it does for Perl's /g option. If all
4338 alternatives of a regex begin with \G, the expression is anchored to the start
4339 match position, and the "anchored" flag is set in the compiled expression.
4340
4341 15. Some bugs concerning the handling of certain option changes within patterns
4342 have been fixed. These applied to options other than (?ims). For example,
4343 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4344 been the other way round. Some of this was related to change 7 above.
4345
4346 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4347 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4348 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4349 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4350
4351 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4352 mode. Unlike ".", it always matches newline, whatever the setting of
4353 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4354 assertions. Perl allows it, but it doesn't (in general) work because it can't
4355 calculate the length of the lookbehind. At least, that's the case for Perl
4356 5.8.0 - I've been told they are going to document that it doesn't work in
4357 future.
4358
4359 18. Added an error diagnosis for escapes that PCRE does not support: these are
4360 \L, \l, \N, \P, \p, \U, \u, and \X.
4361
4362 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4363 reading past the end of the pattern in cases such as /[abcd/.
4364
4365 20. PCRE was getting more memory than necessary for patterns with classes that
4366 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4367
4368 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4369 compiling PCRE for use with Virtual Pascal.
4370
4371 22. Small fix to the Makefile to make it work properly if the build is done
4372 outside the source tree.
4373
4374 23. Added a new extension: a condition to go with recursion. If a conditional
4375 subpattern starts with (?(R) the "true" branch is used if recursion has
4376 happened, whereas the "false" branch is used only at the top level.
4377
4378 24. When there was a very long string of literal characters (over 255 bytes
4379 without UTF support, over 250 bytes with UTF support), the computation of how
4380 much memory was required could be incorrect, leading to segfaults or other
4381 strange effects.
4382
4383 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4384 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4385 there was a subsequent back reference to those brackets. This meant that, for
4386 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4387 possible to check for precisely this case. All we can do is abandon the
4388 optimization if .* occurs inside capturing brackets when there are any back
4389 references whatsoever. (See below for a better fix that came later.)
4390
4391 26. The handling of the optimization for finding the first character of a
4392 non-anchored pattern, and for finding a character that is required later in the
4393 match were failing in some cases. This didn't break the matching; it just
4394 failed to optimize when it could. The way this is done has been re-implemented.
4395
4396 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4397
4398 28. Added a new feature that provides some of the functionality that Perl
4399 provides with (?{...}). The facility is termed a "callout". The way it is done
4400 in PCRE is for the caller to provide an optional function, by setting
4401 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4402 global variable. By default it is unset, which disables all calling out. To get
4403 the function called, the regex must include (?C) at appropriate points. This
4404 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4405 This provides a means of identifying different callout points. When PCRE
4406 reaches such a point in the regex, if pcre_callout has been set, the external
4407 function is called. It is provided with data in a structure called
4408 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4409 matching continues; if it returns a non-zero value, the match at the current
4410 point fails. However, backtracking will occur if possible. [This was changed
4411 later and other features added - see item 49 below.]
4412
4413 29. pcretest is upgraded to test the callout functionality. It provides a
4414 callout function that displays information. By default, it shows the start of
4415 the match and the current position in the text. There are some new data escapes
4416 to vary what happens:
4417
4418 \C+ in addition, show current contents of captured substrings
4419 \C- do not supply a callout function
4420 \C!n return 1 when callout number n is reached
4421 \C!n!m return 1 when callout number n is reached for the mth time
4422
4423 30. If pcregrep was called with the -l option and just a single file name, it
4424 output "<stdin>" if a match was found, instead of the file name.
4425
4426 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4427 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4428 pcre_exec(). This saves a malloc/free per call. The default value of
4429 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4430 when configuring.
4431
4432 32. The default maximum size of a compiled pattern is 64K. There have been a
4433 few cases of people hitting this limit. The code now uses macros to handle the
4434 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4435 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4436 configuring. Tests 2 and 5 work only with 2-byte links because they output
4437 debugging information about compiled patterns.
4438
4439 33. Internal code re-arrangements:
4440
4441 (a) Moved the debugging function for printing out a compiled regex into
4442 its own source file (printint.c) and used #include to pull it into
4443 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4444 separate copies.
4445
4446 (b) Defined the list of op-code names for debugging as a macro in
4447 internal.h so that it is next to the definition of the opcodes.
4448
4449 (c) Defined a table of op-code lengths for simpler skipping along compiled
4450 code. This is again a macro in internal.h so that it is next to the
4451 definition of the opcodes.
4452
4453 34. Added support for recursive calls to individual subpatterns, along the
4454 lines of Robin Houston's patch (but implemented somewhat differently).
4455
4456 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4457 allow it to read and process whole directories in Win32. This code was
4458 contributed by Lionel Fourquaux; it has not been tested by me.
4459
4460 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4461 used to name a group. Names consist of alphanumerics and underscores, and must
4462 be unique. Back references use the syntax (?P=name) and recursive calls use
4463 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4464 numbers. The function pcre_fullinfo() can be used after compilation to extract
4465 a name/number map. There are three relevant calls:
4466
4467 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4468 PCRE_INFO_NAMECOUNT yields the number of entries
4469 PCRE_INFO_NAMETABLE yields a pointer to the map.
4470
4471 The map is a vector of fixed-size entries. The size of each entry depends on
4472 the length of the longest name used. The first two bytes of each entry are the
4473 group number, most significant byte first. There follows the corresponding
4474 name, zero terminated. The names are in alphabetical order.
4475
4476 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4477 case instead of 255. Making it the same both with and without UTF-8 support
4478 means that the same test output works with both.
4479
4480 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4481 calling malloc() with a zero argument.
4482
4483 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4484 optimization. I've improved things by keeping a bitmap of backreferences with
4485 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4486 fact referenced, the optimization can be applied. It is unlikely that a
4487 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4488 the match to follow \n) will appear inside brackets with a number greater than
4489 31, but if it does, any back reference > 31 suppresses the optimization.
4490
4491 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4492 of disabling numbered capturing parentheses. Any opening parenthesis that is
4493 not followed by ? behaves as if it were followed by ?: but named parentheses
4494 can still be used for capturing (and they will acquire numbers in the usual
4495 way).
4496
4497 41. Redesigned the return codes from the match() function into yes/no/error so
4498 that errors can be passed back from deep inside the nested calls. A malloc
4499 failure while inside a recursive subpattern call now causes the
4500 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4501
4502 42. It is now possible to set a limit on the number of times the match()
4503 function is called in a call to pcre_exec(). This facility makes it possible to
4504 limit the amount of recursion and backtracking, though not in a directly
4505 obvious way, because the match() function is used in a number of different
4506 circumstances. The count starts from zero for each position in the subject
4507 string (for non-anchored patterns). The default limit is, for compatibility, a
4508 large number, namely 10 000 000. You can change this in two ways:
4509
4510 (a) When configuring PCRE before making, you can use --with-match-limit=n
4511 to set a default value for the compiled library.
4512
4513 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4514 a different value is set. See 45 below.
4515
4516 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4517
4518 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4519 of things that can be changed at compile time. The first argument specifies
4520 what is wanted and the second points to where the information is to be placed.
4521 The current list of available information is:
4522
4523 PCRE_CONFIG_UTF8
4524
4525 The output is an integer that is set to one if UTF-8 support is available;
4526 otherwise it is set to zero.
4527
4528 PCRE_CONFIG_NEWLINE
4529
4530 The output is an integer that it set to the value of the code that is used for
4531 newline. It is either LF (10) or CR (13).
4532
4533 PCRE_CONFIG_LINK_SIZE
4534
4535 The output is an integer that contains the number of bytes used for internal
4536 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4537
4538 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4539
4540 The output is an integer that contains the threshold above which the POSIX
4541 interface uses malloc() for output vectors. See item 31 above.
4542
4543 PCRE_CONFIG_MATCH_LIMIT
4544
4545 The output is an unsigned integer that contains the default limit of the number
4546 of match() calls in a pcre_exec() execution. See 42 above.
4547
4548 44. pcretest has been upgraded by the addition of the -C option. This causes it
4549 to extract all the available output from the new pcre_config() function, and to
4550 output it. The program then exits immediately.
4551
4552 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4553 order to support additional features. One way would have been to define
4554 pcre_exec2() (for example) with extra arguments, but this would not have been
4555 extensible, and would also have required all calls to the original function to
4556 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4557 is used for passing in "extra" data from pcre_study().
4558
4559 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4560 contains the following fields:
4561
4562 flags a bitmap indicating which of the following fields are set
4563 study_data opaque data from pcre_study()
4564 match_limit a way of specifying a limit on match() calls for a specific
4565 call to pcre_exec()
4566 callout_data data for callouts (see 49 below)
4567
4568 The flag bits are also defined in pcre.h, and are
4569
4570 PCRE_EXTRA_STUDY_DATA
4571 PCRE_EXTRA_MATCH_LIMIT
4572 PCRE_EXTRA_CALLOUT_DATA
4573
4574 The pcre_study() function now returns one of these new pcre_extra blocks, with
4575 the actual study data pointed to by the study_data field, and the
4576 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4577 before. That is, this change is entirely upwards-compatible and requires no
4578 change to existing code.
4579
4580 If you want to pass in additional data to pcre_exec(), you can either place it
4581 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4582 block.
4583
4584 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4585 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4586 times with different match limits, until it finds the minimum value needed for
4587 pcre_exec() to complete. The value is then output. This can be instructive; for
4588 most simple matches the number is quite small, but for pathological cases it
4589 gets very large very quickly.
4590
4591 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4592 returns the size of the data block pointed to by the study_data field in a
4593 pcre_extra block, that is, the value that was passed as the argument to
4594 pcre_malloc() when PCRE was getting memory in which to place the information
4595 created by pcre_study(). The fourth argument should point to a size_t variable.
4596 pcretest has been extended so that this information is shown after a successful
4597 pcre_study() call when information about the compiled regex is being displayed.
4598
4599 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4600 because what follows is always an absolute path. (Later: it turns out that this
4601 is more than cosmetic for MinGW, because it doesn't like empty path
4602 components.)
4603
4604 49. Some changes have been made to the callout feature (see 28 above):
4605
4606 (i) A callout function now has three choices for what it returns:
4607
4608 0 => success, carry on matching
4609 > 0 => failure at this point, but backtrack if possible
4610 < 0 => serious error, return this value from pcre_exec()
4611
4612 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4613 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4614 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4615 use by callout functions. It will never be used by PCRE itself.
4616
4617 (ii) The pcre_extra structure (see 45 above) has a void * field called
4618 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4619 pcre_callout_block structure has a field of the same name. The contents of
4620 the field passed in the pcre_extra structure are passed to the callout
4621 function in the corresponding field in the callout block. This makes it
4622 easier to use the same callout-containing regex from multiple threads. For
4623 testing, the pcretest program has a new data escape
4624
4625 \C*n pass the number n (may be negative) as callout_data
4626
4627 If the callout function in pcretest receives a non-zero value as
4628 callout_data, it returns that value.
4629
4630 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4631 there were some redundant $(CFLAGS) in commands that are now specified as
4632 $(LINK), which already includes $(CFLAGS).
4633
4634 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4635 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4636 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4637 one-byte characters throughout. Note that case-insensitive matching applies
4638 only to characters whose values are less than 256. PCRE doesn't support the
4639 notion of cases for higher-valued characters.
4640
4641 (i) A character class whose characters are all within 0-255 is handled as
4642 a bit map, and the map is inverted for negative classes. Previously, a
4643 character > 255 always failed to match such a class; however it should
4644 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4645
4646 (ii) A negated character class with a single character < 255 is coded as
4647 "not this character" (OP_NOT). This wasn't working properly when the test
4648 character was multibyte, either singly or repeated.
4649
4650 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4651 mode, for example: \x{100}{2,3}.
4652
4653 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4654 singly or repeated) now correctly test multibyte characters. However,
4655 PCRE doesn't recognize any characters with values greater than 255 as
4656 digits, spaces, or word characters. Such characters always match \D, \S,
4657 and \W, and never match \d, \s, or \w.
4658
4659 (v) Classes may now contain characters and character ranges with values
4660 greater than 255. For example: [ab\x{100}-\x{400}].
4661
4662 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4663 PCRE in UTF-8 mode.
4664
4665 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4666 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4667 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4668 value.)
4669
4670 53. The single man page has become too large. I have therefore split it up into
4671 a number of separate man pages. These also give rise to individual HTML pages;
4672 these are now put in a separate directory, and there is an index.html page that
4673 lists them all. Some hyperlinking between the pages has been installed.
4674
4675 54. Added convenience functions for handling named capturing parentheses.
4676
4677 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4678 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4679 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4680 are faulted.
4681
4682 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4683 calling configure. These values are used when compiling the dftables.c program
4684 which is run to generate the source of the default character tables. They
4685 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4686 you will need to set these values.
4687
4688 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4689
4690
4691 Version 3.9 02-Jan-02
4692 ---------------------
4693
4694 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4695
4696 2. If --disable-static was given, the building process failed when trying to
4697 build pcretest and pcregrep. (For some reason it was using libtool to compile
4698 them, which is not right, as they aren't part of the library.)
4699
4700
4701 Version 3.8 18-Dec-01
4702 ---------------------
4703
4704 1. The experimental UTF-8 code was completely screwed up. It was packing the
4705 bytes in the wrong order. How dumb can you get?
4706
4707
4708 Version 3.7 29-Oct-01
4709 ---------------------
4710
4711 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4712 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4713 this didn't happen under Solaris 8, where I normally test things.
4714
4715 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4716 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4717 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4718 this fix an hour or so after the initial 3.7 release.)
4719
4720
4721 Version 3.6 23-Oct-01
4722 ---------------------
4723
4724 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4725 offsets passed as NULL with zero offset count.
4726
4727 2. The config.guess and config.sub files had not been updated when I moved to
4728 the latest autoconf.
4729
4730
4731 Version 3.5 15-Aug-01
4732 ---------------------
4733
4734 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4735 had been forgotten.
4736
4737 2. By using declared but undefined structures, we can avoid using "void"
4738 definitions in pcre.h while keeping the internal definitions of the structures
4739 private.
4740
4741 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4742 user point of view, this means that both static and shared libraries are built
4743 by default, but this can be individually controlled. More of the work of
4744 handling this static/shared cases is now inside libtool instead of PCRE's make
4745 file.
4746
4747 4. The pcretest utility is now installed along with pcregrep because it is
4748 useful for users (to test regexs) and by doing this, it automatically gets
4749 relinked by libtool. The documentation has been turned into a man page, so
4750 there are now .1, .txt, and .html versions in /doc.
4751
4752 5. Upgrades to pcregrep:
4753 (i) Added long-form option names like gnu grep.
4754 (ii) Added --help to list all options with an explanatory phrase.
4755 (iii) Added -r, --recursive to recurse into sub-directories.
4756 (iv) Added -f, --file to read patterns from a file.
4757
4758 6. pcre_exec() was referring to its "code" argument before testing that
4759 argument for NULL (and giving an error if it was NULL).
4760
4761 7. Upgraded Makefile.in to allow for compiling in a different directory from
4762 the source directory.
4763
4764 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4765 options bits, the pointer it was passed was to an int instead of to an unsigned
4766 long int. This mattered only on 64-bit systems.
4767
4768 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4769 generated) instead of pcre.in, which it its source. Also made the same change
4770 in several of the .c files.
4771
4772 10. A new release of gcc defines printf() as a macro, which broke pcretest
4773 because it had an ifdef in the middle of a string argument for printf(). Fixed
4774 by using separate calls to printf().
4775
4776 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4777 script, to force use of CR or LF instead of \n in the source. On non-Unix
4778 systems, the value can be set in config.h.
4779
4780 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4781 absolute limit. Changed the text of the error message to make this clear, and
4782 likewise updated the man page.
4783
4784 13. The limit of 99 on the number of capturing subpatterns has been removed.
4785 The new limit is 65535, which I hope will not be a "real" limit.
4786
4787
4788 Version 3.4 22-Aug-00
4789 ---------------------
4790
4791 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4792
4793 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4794
4795
4796 Version 3.3 01-Aug-00
4797 ---------------------
4798
4799 1. If an octal character was given, but the value was greater than \377, it
4800 was not getting masked to the least significant bits, as documented. This could
4801 lead to crashes in some systems.
4802
4803 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4804 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4805
4806 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4807 These just pass their arguments on to (pcre_free)(), but they are provided
4808 because some uses of PCRE bind it to non-C systems that can call its functions,
4809 but cannot call free() or pcre_free() directly.
4810
4811 4. Add "make test" as a synonym for "make check". Corrected some comments in
4812 the Makefile.
4813
4814 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4815 Makefile.
4816
4817 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4818 command called pgrep for grepping around the active processes.
4819
4820 7. Added the beginnings of support for UTF-8 character strings.
4821
4822 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4823 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4824 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4825 out for the ar command.)
4826
4827
4828 Version 3.2 12-May-00
4829 ---------------------
4830
4831 This is purely a bug fixing release.
4832
4833 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4834 of ZA. This was just one example of several cases that could provoke this bug,
4835 which was introduced by change 9 of version 2.00. The code for breaking
4836 infinite loops after an iteration that matches an empty string was't working
4837 correctly.
4838
4839 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4840 when matched against abbab (for example). After matching an empty string, it
4841 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4842 caused it to match further down the string than it should.
4843
4844 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4845 was there because it doesn't seem to be needed, and it causes trouble on some
4846 systems, as it is not a Standard C header. It has been removed.
4847
4848 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4849 were reported on the Macintosh. The changes were from
4850
4851 while ((c = *(++ptr)) != 0 && c != '\n');
4852 to
4853 while ((c = *(++ptr)) != 0 && c != '\n') ;
4854
4855 Totally extraordinary, but if that's what it takes...
4856
4857 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4858 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4859 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4860 assumes the way PCRE uses memmove() (always moving upwards).
4861
4862 6. PCRE is being used in one environment where strchr() is not available. There
4863 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4864 faster code anyway.
4865
4866
4867 Version 3.1 09-Feb-00
4868 ---------------------
4869
4870 The only change in this release is the fixing of some bugs in Makefile.in for
4871 the "install" target:
4872
4873 (1) It was failing to install pcreposix.h.
4874
4875 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4876
4877
4878 Version 3.0 01-Feb-00
4879 ---------------------
4880
4881 1. Add support for the /+ modifier to perltest (to output $` like it does in
4882 pcretest).
4883
4884 2. Add support for the /g modifier to perltest.
4885
4886 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4887 matches null strings.
4888
4889 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4890 pattern. Perl treats empty patterns specially - it reuses the most recent
4891 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4892 effect.
4893
4894 5. The POSIX interface was broken in that it was just handing over the POSIX
4895 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4896 required a bigger vector, with some working space on the end. This means that
4897 the POSIX wrapper now has to get and free some memory, and copy the results.
4898
4899 6. Added some simple autoconf support, placing the test data and the
4900 documentation in separate directories, re-organizing some of the
4901 information files, and making it build pcre-config (a GNU standard). Also added
4902 libtool support for building PCRE as a shared library, which is now the
4903 default.
4904
4905 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4906 09 are not valid octal constants. Single digits will be used for minor values
4907 less than 10.
4908
4909 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4910 existing programs that set these in the POSIX interface can use PCRE without
4911 modification.
4912
4913 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4914 return all that pcre_info() returns, plus additional data. The pcre_info()
4915 function is retained for compatibility, but is considered to be obsolete.
4916
4917 10. Added experimental recursion feature (?R) to handle one common case that
4918 Perl 5.6 will be able to do with (?p{...}).
4919
4920 11. Added support for POSIX character classes like [:alpha:], which Perl is
4921 adopting.
4922
4923
4924 Version 2.08 31-Aug-99
4925 ----------------------
4926
4927 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4928 trying to match at the startoffset position, but instead was moving forward to
4929 the next newline as if a previous match had failed.
4930
4931 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4932 and could get into a loop if a null string was matched other than at the start
4933 of the subject.
4934
4935 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4936 be distinguished at compile time, and for completeness also added PCRE_DATE.
4937
4938 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4939 in GnuWin32 environments.
4940
4941
4942 Version 2.07 29-Jul-99
4943 ----------------------
4944
4945 1. The documentation is now supplied in plain text form and HTML as well as in
4946 the form of man page sources.
4947
4948 2. C++ compilers don't like assigning (void *) values to other pointer types.
4949 In particular this affects malloc(). Although there is no problem in Standard
4950 C, I've put in casts to keep C++ compilers happy.
4951
4952 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4953 should be (const char *).
4954
4955 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4956 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4957 However, I haven't made this a standard facility. The documentation doesn't
4958 mention it, and the Makefile doesn't support it.
4959
4960 5. The Makefile now contains an "install" target, with editable destinations at
4961 the top of the file. The pcretest program is not installed.
4962
4963 6. pgrep -V now gives the PCRE version number and date.
4964
4965 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4966 causing the entire string to be ignored, instead of just the last character.
4967
4968 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4969 non-matching string, it can take a very, very long time, even for strings of
4970 quite modest length, because of the nested recursion. PCRE now does better in
4971 some of these cases. It does this by remembering the last required literal
4972 character in the pattern, and pre-searching the subject to ensure it is present
4973 before running the real match. In other words, it applies a heuristic to detect
4974 some types of certain failure quickly, and in the above example, if presented
4975 with a string that has no trailing " it gives "no match" very quickly.
4976
4977 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4978 other alternatives are tried instead.
4979
4980
4981 Version 2.06 09-Jun-99
4982 ----------------------
4983
4984 1. Change pcretest's output for amount of store used to show just the code
4985 space, because the remainder (the data block) varies in size between 32-bit and
4986 64-bit systems.
4987
4988 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4989 start matching at. This allows lookbehinds to work when searching for multiple
4990 occurrences in a string.
4991
4992 3. Added additional options to pcretest for testing multiple occurrences:
4993
4994 /+ outputs the rest of the string that follows a match
4995 /g loops for multiple occurrences, using the new startoffset argument
4996 /G loops for multiple occurrences by passing an incremented pointer
4997
4998 4. PCRE wasn't doing the "first character" optimization for patterns starting
4999 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5000 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5001 the letter 'x'. On long subject strings, this gives a significant speed-up.
5002
5003
5004 Version 2.05 21-Apr-99
5005 ----------------------
5006
5007 1. Changed the type of magic_number from int to long int so that it works
5008 properly on 16-bit systems.
5009
5010 2. Fixed a bug which caused patterns starting with .* not to work correctly
5011 when the subject string contained newline characters. PCRE was assuming
5012 anchoring for such patterns in all cases, which is not correct because .* will
5013 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5014 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5015 must be retried after every newline in the subject.
5016
5017
5018 Version 2.04 18-Feb-99
5019 ----------------------
5020
5021 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5022 computation of the store needed to hold the pattern was incorrect (too large).
5023 If such patterns were nested a few deep, this could multiply and become a real
5024 problem.
5025
5026 2. Added /M option to pcretest to show the memory requirement of a specific
5027 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5028
5029 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5030 compiled in such a way that the backtracking after subsequent failure was
5031 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5032 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5033
5034
5035 Version 2.03 02-Feb-99
5036 ----------------------
5037
5038 1. Fixed typo and small mistake in man page.
5039
5040 2. Added 4th condition (GPL supersedes if conflict) and created separate
5041 LICENCE file containing the conditions.
5042
5043 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5044 Perl, that is the internal \ allows the delimiter to be included in the
5045 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5046 the final delimiter, add \ to the end of the pattern (to test the error).
5047
5048 4. Added the convenience functions for extracting substrings after a successful
5049 match. Updated pcretest to make it able to test these functions.
5050
5051
5052 Version 2.02 14-Jan-99
5053 ----------------------
5054
5055 1. Initialized the working variables associated with each extraction so that
5056 their saving and restoring doesn't refer to uninitialized store.
5057
5058 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5059 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5060 fix the problem.
5061
5062 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5063 calls, and wasn't printing the correct value for compiling calls. Increased the
5064 default value of LOOPREPEAT, and the number of significant figures in the
5065 times.
5066
5067 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5068
5069 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5070 a building problem on Windows NT with a FAT file system.
5071
5072
5073 Version 2.01 21-Oct-98
5074 ----------------------
5075
5076 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5077 to character tables built by pcre_maketables() in the current locale. If NULL
5078 is passed, the default tables are used.
5079
5080
5081 Version 2.00 24-Sep-98
5082 ----------------------
5083
5084 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5085 it any more.
5086
5087 2. Allow quantification of (?>) groups, and make it work correctly.
5088
5089 3. The first character computation wasn't working for (?>) groups.
5090
5091 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5092 end of the subject) and add 5.005's \z, which really does match only at the
5093 very end of the subject.
5094
5095 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5096
5097 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5098 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5099 localized options. All options to pcre_study() were also removed.
5100
5101 7. Add other new features from 5.005:
5102
5103 $(?<= positive lookbehind
5104 $(?<! negative lookbehind
5105 (?imsx-imsx) added the unsetting capability
5106 such a setting is global if at outer level; local otherwise
5107 (?imsx-imsx:) non-capturing groups with option setting
5108 (?(cond)re|re) conditional pattern matching
5109
5110 A backreference to itself in a repeated group matches the previous
5111 captured string.
5112
5113 8. General tidying up of studying (both automatic and via "study")
5114 consequential on the addition of new assertions.
5115
5116 9. As in 5.005, unlimited repeated groups that could match an empty substring
5117 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5118 runtime if any iteration does actually match an empty substring.
5119
5120 10. Include the RunTest script in the distribution.
5121
5122 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5123 discrepancies, some of which were old and were also with respect to 5.004. They
5124 have now been fixed.
5125
5126
5127 Version 1.09 28-Apr-98
5128 ----------------------
5129
5130 1. A negated single character class followed by a quantifier with a minimum
5131 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5132 program crashes, or just wrong answers. This did not apply to negated classes
5133 containing more than one character, or to minima other than one.
5134
5135
5136 Version 1.08 27-Mar-98
5137 ----------------------
5138
5139 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5140
5141 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5142 latter must appear before anything that relies on it in the pattern.
5143
5144
5145 Version 1.07 16-Feb-98
5146 ----------------------
5147
5148 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5149 repeat of a potentially empty string).
5150
5151
5152 Version 1.06 23-Jan-98
5153 ----------------------
5154
5155 1. Added Markus Oberhumer's little patches for C++.
5156
5157 2. Literal strings longer than 255 characters were broken.
5158
5159
5160 Version 1.05 23-Dec-97
5161 ----------------------
5162
5163 1. Negated character classes containing more than one character were failing if
5164 PCRE_CASELESS was set at run time.
5165
5166
5167 Version 1.04 19-Dec-97
5168 ----------------------
5169
5170 1. Corrected the man page, where some "const" qualifiers had been omitted.
5171
5172 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5173 input syntax.
5174
5175 3. Fixed memory leak which occurred when a regex with back references was
5176 matched with an offsets vector that wasn't big enough. The temporary memory
5177 that is used in this case wasn't being freed if the match failed.
5178
5179 4. Tidied pcretest to ensure it frees memory that it gets.
5180
5181 5. Temporary memory was being obtained in the case where the passed offsets
5182 vector was exactly big enough.
5183
5184 6. Corrected definition of offsetof() from change 5 below.
5185
5186 7. I had screwed up change 6 below and broken the rules for the use of
5187 setjmp(). Now fixed.
5188
5189
5190 Version 1.03 18-Dec-97
5191 ----------------------
5192
5193 1. A erroneous regex with a missing opening parenthesis was correctly
5194 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5195 on some systems.
5196
5197 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5198 it was reported that one broken compiler failed on the former because "code" is
5199 also an independent variable.
5200
5201 3. The erroneous regex a[]b caused an array overrun reference.
5202
5203 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5204 fail on data ending with that character. (It was going on too far, and checking
5205 the next character, typically a binary zero.) This was specific to the
5206 optimized code for single-character negative classes.
5207
5208 5. Added a contributed patch from the TIN world which does the following:
5209
5210 + Add an undef for memmove, in case the the system defines a macro for it.
5211
5212 + Add a definition of offsetof(), in case there isn't one. (I don't know
5213 the reason behind this - offsetof() is part of the ANSI standard - but
5214 it does no harm).
5215
5216 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5217 most of the places where whitespace preceded '#'. I have given up and
5218 allowed the remaining 2 cases to be at the margin.
5219
5220 + Rename some variables in pcre to eliminate shadowing. This seems very
5221 pedantic, but does no harm, of course.
5222
5223 6. Moved the call to setjmp() into its own function, to get rid of warnings
5224 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5225
5226 7. Constructs such as \d{8,} were compiling into the equivalent of
5227 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5228 outcome, but in this particular case used more store than had been allocated,
5229 which caused the bug to be discovered because it threw up an internal error.
5230
5231 8. The debugging code in both pcre and pcretest for outputting the compiled
5232 form of a regex was going wrong in the case of back references followed by
5233 curly-bracketed repeats.
5234
5235
5236 Version 1.02 12-Dec-97
5237 ----------------------
5238
5239 1. Typos in pcre.3 and comments in the source fixed.
5240
5241 2. Applied a contributed patch to get rid of places where it used to remove
5242 'const' from variables, and fixed some signed/unsigned and uninitialized
5243 variable warnings.
5244
5245 3. Added the "runtest" target to Makefile.
5246
5247 4. Set default compiler flag to -O2 rather than just -O.
5248
5249
5250 Version 1.01 19-Nov-97
5251 ----------------------
5252
5253 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5254 like /([ab]*)*/, that is, for classes with more than one character in them.
5255
5256 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5257 as /((?>a*))*/ (a PCRE_EXTRA facility).
5258
5259
5260 Version 1.00 18-Nov-97
5261 ----------------------
5262
5263 1. Added compile-time macros to support systems such as SunOS4 which don't have
5264 memmove() or strerror() but have other things that can be used instead.
5265
5266 2. Arranged that "make clean" removes the executables.
5267
5268
5269 Version 0.99 27-Oct-97
5270 ----------------------
5271
5272 1. Fixed bug in code for optimizing classes with only one character. It was
5273 initializing a 32-byte map regardless, which could cause it to run off the end
5274 of the memory it had got.
5275
5276 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5277
5278
5279 Version 0.98 22-Oct-97
5280 ----------------------
5281
5282 1. Fixed bug in code for handling temporary memory usage when there are more
5283 back references than supplied space in the ovector. This could cause segfaults.
5284
5285
5286 Version 0.97 21-Oct-97
5287 ----------------------
5288
5289 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5290
5291 2. Optimized negated single characters not to use a bit map.
5292
5293 3. Brought error texts together as macro definitions; clarified some of them;
5294 fixed one that was wrong - it said "range out of order" when it meant "invalid
5295 escape sequence".
5296
5297 4. Changed some char * arguments to const char *.
5298
5299 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5300
5301 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5302 pcretest.
5303
5304
5305 Version 0.96 16-Oct-97
5306 ----------------------
5307
5308 1. Added a simple "pgrep" utility to the distribution.
5309
5310 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5311 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5312 where "ddd" means "one or more decimal digits".
5313
5314 3. Fixed serious bug. If a pattern had a back reference, but the call to
5315 pcre_exec() didn't supply a large enough ovector to record the related
5316 identifying subpattern, the match always failed. PCRE now remembers the number
5317 of the largest back reference, and gets some temporary memory in which to save
5318 the offsets during matching if necessary, in order to ensure that
5319 backreferences always work.
5320
5321 4. Increased the compatibility with Perl in a number of ways:
5322
5323 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5324 to request this handling. The option can be set at compile or exec time.
5325
5326 (b) $ matches before a terminating newline by default; an option
5327 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5328 mode). The option can be set at compile or exec time.
5329
5330 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5331 the same as Perl's. If the decimal number it represents is less than 10
5332 or there aren't that many previous left capturing parentheses, an octal
5333 escape is read. Inside a character class, it's always an octal escape,
5334 even if it is a single digit.
5335
5336 (d) An escaped but undefined alphabetic character is taken as a literal,
5337 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5338 escapes.
5339
5340 (e) {0} is now permitted. (The previous item is removed from the compiled
5341 pattern).
5342
5343 5. Changed all the names of code files so that the basic parts are no longer
5344 than 10 characters, and abolished the teeny "globals.c" file.
5345
5346 6. Changed the handling of character classes; they are now done with a 32-byte
5347 bit map always.
5348
5349 7. Added the -d and /D options to pcretest to make it possible to look at the
5350 internals of compilation without having to recompile pcre.
5351
5352
5353 Version 0.95 23-Sep-97
5354 ----------------------
5355
5356 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5357 \x20 at the start of a run of normal characters. These were being treated as
5358 real characters, instead of the source characters being re-checked.
5359
5360
5361 Version 0.94 18-Sep-97
5362 ----------------------
5363
5364 1. The functions are now thread-safe, with the caveat that the global variables
5365 containing pointers to malloc() and free() or alternative functions are the
5366 same for all threads.
5367
5368 2. Get pcre_study() to generate a bitmap of initial characters for non-
5369 anchored patterns when this is possible, and use it if passed to pcre_exec().
5370
5371
5372 Version 0.93 15-Sep-97
5373 ----------------------
5374
5375 1. /(b)|(:+)/ was computing an incorrect first character.
5376
5377 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5378 but not actually doing anything yet.
5379
5380 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5381 as Perl does (e.g. [-az] or [az-]).
5382
5383 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5384 all possible positions.
5385
5386 5. Split up into different modules to avoid including unneeded functions in a
5387 compiled binary. However, compile and exec are still in one module. The "study"
5388 function is split off.
5389
5390 6. The character tables are now in a separate module whose source is generated
5391 by an auxiliary program - but can then be edited by hand if required. There are
5392 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5393 toupper() in the code.
5394
5395 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5396 make them global. Abolish the function for setting them, as the caller can now
5397 set them directly.
5398
5399
5400 Version 0.92 11-Sep-97
5401 ----------------------
5402
5403 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5404 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5405
5406 2. Caseless matching was not working in character classes if the characters in
5407 the pattern were in upper case.
5408
5409 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5410
5411 4. Make PCRE_ANCHORED public and accept as a compile option.
5412
5413 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5414 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5415 pass them.
5416
5417 6. Give an error if bad option bits passed at compile or run time.
5418
5419 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5420 pcretest to cause it to pass that flag.
5421
5422 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5423 options, and the first character, if set.
5424
5425 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5426
5427
5428 Version 0.91 10-Sep-97
5429 ----------------------
5430
5431 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5432 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5433
5434 2. PCRE was looping on encountering an indefinitely repeated back reference to
5435 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5436 Perl does - treats the match as successful.
5437
5438 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5