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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1435 - (show annotations)
Tue Jan 7 07:47:12 2014 UTC (5 years, 10 months ago) by zherczeg
File size: 260661 byte(s)
Error occurred while calculating annotation data.
JIT: Optimize might be empty checks.
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. Remove these unnecessary checks form JIT.
56
57
58 Version 8.34 15-December-2013
59 -----------------------------
60
61 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
62 executable memory. Patch inspired by Carsten Klein.
63
64 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
65 this macro is never tested and has no effect, because the work to support
66 coverage involves only compiling and linking options and special targets in
67 the Makefile. The comment in config.h implied that defining the macro would
68 enable coverage support, which is totally false. There was also support for
69 setting this macro in the CMake files (my fault, I just copied it from
70 configure). SUPPORT_GCOV has now been removed.
71
72 3. Make a small performance improvement in strlen16() and strlen32() in
73 pcretest.
74
75 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
76 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
77 these cases). There was also one in pcretest.c.
78
79 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
80
81 6. In UTF mode, the code for checking whether a group could match an empty
82 string (which is used for indefinitely repeated groups to allow for
83 breaking an infinite loop) was broken when the group contained a repeated
84 negated single-character class with a character that occupied more than one
85 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
86 UTF-8 mode). The effect was undefined: the group might or might not be
87 deemed as matching an empty string, or the program might have crashed.
88
89 7. The code for checking whether a group could match an empty string was not
90 recognizing that \h, \H, \v, \V, and \R must match a character.
91
92 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
93 an empty string. If it can, pcretest shows this in its information output.
94
95 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
96 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
97 matched by pcre_exec() without using JIT:
98
99 (a) If the rest of the pattern did not match after a maximal run of
100 grapheme clusters, the code for backing up to try with fewer of them
101 did not always back up over a full grapheme when characters that do not
102 have the modifier quality were involved, e.g. Hangul syllables.
103
104 (b) If the match point in a subject started with modifier character, and
105 there was no match, the code could incorrectly back up beyond the match
106 point, and potentially beyond the first character in the subject,
107 leading to a segfault or an incorrect match result.
108
109 10. A conditional group with an assertion condition could lead to PCRE
110 recording an incorrect first data item for a match if no other first data
111 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
112 first data item, and therefore matched "ca" after "c" instead of at the
113 start.
114
115 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
116 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
117
118 12. The source of pcregrep now includes z/OS-specific code so that it can be
119 compiled for z/OS as part of the special z/OS distribution.
120
121 13. Added the -T and -TM options to pcretest.
122
123 14. The code in pcre_compile.c for creating the table of named capturing groups
124 has been refactored. Instead of creating the table dynamically during the
125 actual compiling pass, the information is remembered during the pre-compile
126 pass (on the stack unless there are more than 20 named groups, in which
127 case malloc() is used) and the whole table is created before the actual
128 compile happens. This has simplified the code (it is now nearly 150 lines
129 shorter) and prepared the way for better handling of references to groups
130 with duplicate names.
131
132 15. A back reference to a named subpattern when there is more than one of the
133 same name now checks them in the order in which they appear in the pattern.
134 The first one that is set is used for the reference. Previously only the
135 first one was inspected. This change makes PCRE more compatible with Perl.
136
137 16. Unicode character properties were updated from Unicode 6.3.0.
138
139 17. The compile-time code for auto-possessification has been refactored, based
140 on a patch by Zoltan Herczeg. It now happens after instead of during
141 compilation. The code is cleaner, and more cases are handled. The option
142 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
143 options in pcretest are provided to set it. It can also be set by
144 (*NO_AUTO_POSSESS) at the start of a pattern.
145
146 18. The character VT has been added to the default ("C" locale) set of
147 characters that match \s and are generally treated as white space,
148 following this same change in Perl 5.18. There is now no difference between
149 "Perl space" and "POSIX space". Whether VT is treated as white space in
150 other locales depends on the locale.
151
152 19. The code for checking named groups as conditions, either for being set or
153 for being recursed, has been refactored (this is related to 14 and 15
154 above). Processing unduplicated named groups should now be as fast at
155 numerical groups, and processing duplicated groups should be faster than
156 before.
157
158 20. Two patches to the CMake build system, by Alexander Barkov:
159
160 (1) Replace the "source" command by "." in CMakeLists.txt because
161 "source" is a bash-ism.
162
163 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
164 without these the CMake build does not work on Solaris.
165
166 21. Perl has changed its handling of \8 and \9. If there is no previously
167 encountered capturing group of those numbers, they are treated as the
168 literal characters 8 and 9 instead of a binary zero followed by the
169 literals. PCRE now does the same.
170
171 22. Following Perl, added \o{} to specify codepoints in octal, making it
172 possible to specify values greater than 0777 and also making them
173 unambiguous.
174
175 23. Perl now gives an error for missing closing braces after \x{... instead of
176 treating the string as literal. PCRE now does the same.
177
178 24. RunTest used to grumble if an inappropriate test was selected explicitly,
179 but just skip it when running all tests. This make it awkward to run ranges
180 of tests when one of them was inappropriate. Now it just skips any
181 inappropriate tests, as it always did when running all tests.
182
183 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
184 character types such as \d or \w, too many callouts were inserted, and the
185 data that they returned was rubbish.
186
187 26. In UCP mode, \s was not matching two of the characters that Perl matches,
188 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
189 were matched by \h. The code has now been refactored so that the lists of
190 the horizontal and vertical whitespace characters used for \h and \v (which
191 are defined only in one place) are now also used for \s.
192
193 27. Add JIT support for the 64 bit TileGX architecture.
194 Patch by Jiong Wang (Tilera Corporation).
195
196 28. Possessive quantifiers for classes (both explicit and automatically
197 generated) now use special opcodes instead of wrapping in ONCE brackets.
198
199 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
200 (because it's meaningless), this was not happening when PCRE_CASELESS was
201 set. Not wrong, but inefficient.
202
203 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
204 properties for \w, \d, etc) is present in a test regex. Otherwise if the
205 test contains no characters greater than 255, Perl doesn't realise it
206 should be using Unicode semantics.
207
208 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
209 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
210 does in Unicode mode.
211
212 32. Added the "forbid" facility to pcretest so that putting tests into the
213 wrong test files can sometimes be quickly detected.
214
215 33. There is now a limit (default 250) on the depth of nesting of parentheses.
216 This limit is imposed to control the amount of system stack used at compile
217 time. It can be changed at build time by --with-parens-nest-limit=xxx or
218 the equivalent in CMake.
219
220 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
221 errors. Perl warns for these when in warning mode, but PCRE has no facility
222 for giving warnings.
223
224 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
225 However, this was not working for (?!) because it is optimized to (*FAIL),
226 for which PCRE does not allow quantifiers. The optimization is now disabled
227 when a quantifier follows (?!). I can't see any use for this, but it makes
228 things uniform.
229
230 36. Perl no longer allows group names to start with digits, so I have made this
231 change also in PCRE. It simplifies the code a bit.
232
233 37. In extended mode, Perl ignores spaces before a + that indicates a
234 possessive quantifier. PCRE allowed a space before the quantifier, but not
235 before the possessive +. It now does.
236
237 38. The use of \K (reset reported match start) within a repeated possessive
238 group such as (a\Kb)*+ was not working.
239
240 40. Document that the same character tables must be used at compile time and
241 run time, and that the facility to pass tables to pcre_exec() and
242 pcre_dfa_exec() is for use only with saved/restored patterns.
243
244 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
245 features for Builds with MSVC:
246
247 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
248 builds). The .rc files can be used to set FileDescription and many other
249 attributes.
250
251 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
252 This allows higher-level build scripts which want .pdb files to avoid
253 hard-coding the exact files needed."
254
255 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
256 mean "start of word" and "end of word", respectively, as a transition aid.
257
258 43. A minimizing repeat of a class containing codepoints greater than 255 in
259 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
260 compiled to use the heap for recursion.
261
262 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
263 is configured.
264
265
266 Version 8.33 28-May-2013
267 ------------------------
268
269 1. Added 'U' to some constants that are compared to unsigned integers, to
270 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
271 variables that are added to signed variables, to ensure the result is
272 signed and can be negated.
273
274 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
275 CMake config files.
276
277 3. Revise the creation of config.h.generic so that all boolean macros are
278 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
279 overriding via -D on the command line possible.
280
281 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
282 to unsigned int is reported to make a quite noticeable speed difference in
283 a specific Windows environment. Testing on Linux did also appear to show
284 some benefit (and it is clearly not harmful). Also fixed the definition of
285 Xop which should be unsigned.
286
287 5. Related to (4), changing the definition of the intermediate variable cc
288 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
289 improvements.
290
291 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
292 unnecessary spaces.
293
294 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
295 and later.
296
297 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
298
299 9. Optimizing fast_forward_start_bits in JIT.
300
301 10. Adding support for callouts in JIT, and fixing some issues revealed
302 during this work. Namely:
303
304 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
305
306 (b) Minimum length was not checked before the matching is started.
307
308 11. The value of capture_last that is passed to callouts was incorrect in some
309 cases when there was a capture on one path that was subsequently abandoned
310 after a backtrack. Also, the capture_last value is now reset after a
311 recursion, since all captures are also reset in this case.
312
313 12. The interpreter no longer returns the "too many substrings" error in the
314 case when an overflowing capture is in a branch that is subsequently
315 abandoned after a backtrack.
316
317 13. In the pathological case when an offset vector of size 2 is used, pcretest
318 now prints out the matched string after a yield of 0 or 1.
319
320 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
321 Only supported by the JIT compiler at the moment.
322
323 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
324
325 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
326 offset of the starting point of the matching process, provided the offsets
327 vector is large enough.
328
329 17. The \A escape now records a lookbehind value of 1, though its execution
330 does not actually inspect the previous character. This is to ensure that,
331 in partial multi-segment matching, at least one character from the old
332 segment is retained when a new segment is processed. Otherwise, if there
333 are no lookbehinds in the pattern, \A might match incorrectly at the start
334 of a new segment.
335
336 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
337
338 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
339 gives some modest performance improvement in 8-bit mode.
340
341 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
342 be expressed in certain programming languages using Universal Character
343 Names.
344
345 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
346 which points out that "non characters" are not "characters that may not
347 appear in Unicode strings" but rather "characters that are reserved for
348 internal use and have only local meaning".
349
350 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
351 there was a conditional group that depended on an assertion, if the
352 assertion was false, the callout that immediately followed the alternation
353 in the condition was skipped when pcre_exec() was used for matching.
354
355 23. Allow an explicit callout to be inserted before an assertion that is the
356 condition for a conditional group, for compatibility with automatic
357 callouts, which always insert a callout at this point.
358
359 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
360 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
361
362 25. (*PRUNE) is now supported by the JIT compiler.
363
364 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
365
366 27. Fix the case where there are two or more SKIPs with arguments that may be
367 ignored.
368
369 28. (*SKIP) is now supported by the JIT compiler.
370
371 29. (*THEN) is now supported by the JIT compiler.
372
373 30. Update RunTest with additional test selector options.
374
375 31. The way PCRE handles backtracking verbs has been changed in two ways.
376
377 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
378 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
379 backtracking. In some cases this makes it more Perl-compatible, but Perl's
380 rather obscure rules do not always do the same thing.
381
382 (2) Previously, backtracking verbs were confined within assertions. This is
383 no longer the case for positive assertions, except for (*ACCEPT). Again,
384 this sometimes improves Perl compatibility, and sometimes does not.
385
386 32. A number of tests that were in test 2 because Perl did things differently
387 have been moved to test 1, because either Perl or PCRE has changed, and
388 these tests are now compatible.
389
390 32. Backtracking control verbs are now handled in the same way in JIT and
391 interpreter.
392
393 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
394 contained a forward subroutine reference caused a compile error.
395
396 34. Auto-detect and optimize limited repetitions in JIT.
397
398 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
399 blocking (*UTF) etc.
400
401 36. In the interpreter, maximizing pattern repetitions for characters and
402 character types now use tail recursion, which reduces stack usage.
403
404 37. The value of the max lookbehind was not correctly preserved if a compiled
405 and saved regex was reloaded on a host of different endianness.
406
407 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
408 of the compiled pattern block, expand the flags field from 16 to 32 bits
409 because it was almost full.
410
411 39. Try madvise first before posix_madvise.
412
413 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
414 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
415 original change locked out all matches of empty strings. This has been
416 changed so that one match of an empty string per line is recognized.
417 Subsequent searches on the same line (for colouring or for --only-matching,
418 for example) do not recognize empty strings.
419
420 41. Applied a user patch to fix a number of spelling mistakes in comments.
421
422 42. Data lines longer than 65536 caused pcretest to crash.
423
424 43. Clarified the data type for length and startoffset arguments for pcre_exec
425 and pcre_dfa_exec in the function-specific man pages, where they were
426 explicitly stated to be in bytes, never having been updated. I also added
427 some clarification to the pcreapi man page.
428
429 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
430 a segmentation fault.
431
432
433 Version 8.32 30-November-2012
434 -----------------------------
435
436 1. Improved JIT compiler optimizations for first character search and single
437 character iterators.
438
439 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
440 Patch by Daniel Richard G.
441
442 3. Single character iterator optimizations in the JIT compiler.
443
444 4. Improved JIT compiler optimizations for character ranges.
445
446 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
447 Reported by Giuseppe D'Angelo.
448
449 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
450 of a line, was being set incorrectly in cases where .* appeared inside
451 atomic brackets at the start of a pattern, or where there was a subsequent
452 *PRUNE or *SKIP.
453
454 7. Improved instruction cache flush for POWER/PowerPC.
455 Patch by Daniel Richard G.
456
457 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
458 grep:
459
460 (a) There is now no limit to the number of patterns to be matched.
461
462 (b) An error is given if a pattern is too long.
463
464 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
465 are now supported.
466
467 (d) --exclude-from and --include-from (multiple use) have been added.
468
469 (e) Exclusions and inclusions now apply to all files and directories, not
470 just to those obtained from scanning a directory recursively.
471
472 (f) Multiple uses of -f and --file-list are now supported.
473
474 (g) In a Windows environment, the default for -d has been changed from
475 "read" (the GNU grep default) to "skip", because otherwise the presence
476 of a directory in the file list provokes an error.
477
478 (h) The documentation has been revised and clarified in places.
479
480 9. Improve the matching speed of capturing brackets.
481
482 10. Changed the meaning of \X so that it now matches a Unicode extended
483 grapheme cluster.
484
485 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
486 out POSIX threads when JIT support is configured.
487
488 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
489
490 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
491 structure could go wrong in environments where size_t is not the same size
492 as int.
493
494 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
495 set.
496
497 15. The EBCDIC support had decayed; later updates to the code had included
498 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
499 general tidy up of EBCDIC-related issues, and the documentation was also
500 not quite right. There is now a test that can be run on ASCII systems to
501 check some of the EBCDIC-related things (but is it not a full test).
502
503 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
504 in a small tidy to the code.
505
506 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
507
508 18. If the --only-matching (-o) option in pcregrep is specified multiple
509 times, each one causes appropriate output. For example, -o1 -o2 outputs the
510 substrings matched by the 1st and 2nd capturing parentheses. A separating
511 string can be specified by --om-separator (default empty).
512
513 19. Improving the first n character searches.
514
515 20. Turn case lists for horizontal and vertical white space into macros so that
516 they are defined only once.
517
518 21. This set of changes together give more compatible Unicode case-folding
519 behaviour for characters that have more than one other case when UCP
520 support is available.
521
522 (a) The Unicode property table now has offsets into a new table of sets of
523 three or more characters that are case-equivalent. The MultiStage2.py
524 script that generates these tables (the pcre_ucd.c file) now scans
525 CaseFolding.txt instead of UnicodeData.txt for character case
526 information.
527
528 (b) The code for adding characters or ranges of characters to a character
529 class has been abstracted into a generalized function that also handles
530 case-independence. In UTF-mode with UCP support, this uses the new data
531 to handle characters with more than one other case.
532
533 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
534 whose other case is greater than 256 are now correctly matched
535 caselessly. Previously, the high codepoint matched the low one, but not
536 vice versa.
537
538 (d) The processing of \h, \H, \v, and \ in character classes now makes use
539 of the new class addition function, using character lists defined as
540 macros alongside the case definitions of 20 above.
541
542 (e) Caseless back references now work with characters that have more than
543 one other case.
544
545 (f) General caseless matching of characters with more than one other case
546 is supported.
547
548 22. Unicode character properties were updated from Unicode 6.2.0
549
550 23. Improved CMake support under Windows. Patch by Daniel Richard G.
551
552 24. Add support for 32-bit character strings, and UTF-32
553
554 25. Major JIT compiler update (code refactoring and bugfixing).
555 Experimental Sparc 32 support is added.
556
557 26. Applied a modified version of Daniel Richard G's patch to create
558 pcre.h.generic and config.h.generic by "make" instead of in the
559 PrepareRelease script.
560
561 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
562 pcre_compile.c when checking for a zero character.
563
564 28. Introducing a native interface for JIT. Through this interface, the compiled
565 machine code can be directly executed. The purpose of this interface is to
566 provide fast pattern matching, so several sanity checks are not performed.
567 However, feature tests are still performed. The new interface provides
568 1.4x speedup compared to the old one.
569
570 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
571 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
572 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
573
574 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
575 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
576 the "old" RFC 2279). Instead, it ended up passing a negative length to
577 pcre_exec().
578
579 31. Add support for GCC's visibility feature to hide internal functions.
580
581 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
582 "unknown -C option" after outputting 0 or 1.
583
584 33. There is now support for generating a code coverage report for the test
585 suite in environments where gcc is the compiler and lcov is installed. This
586 is mainly for the benefit of the developers.
587
588 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
589 unaddressable using valgrind annotations, allowing valgrind to detect
590 invalid memory accesses. This is mainly for the benefit of the developers.
591
592 25. (*UTF) can now be used to start a pattern in any of the three libraries.
593
594 26. Give configure error if --enable-cpp but no C++ compiler found.
595
596
597 Version 8.31 06-July-2012
598 -------------------------
599
600 1. Fixing a wrong JIT test case and some compiler warnings.
601
602 2. Removed a bashism from the RunTest script.
603
604 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
605 to unsigned type, result still unsigned" that was given by an MS compiler
606 on encountering the code "-sizeof(xxx)".
607
608 4. Partial matching support is added to the JIT compiler.
609
610 5. Fixed several bugs concerned with partial matching of items that consist
611 of more than one character:
612
613 (a) /^(..)\1/ did not partially match "aba" because checking references was
614 done on an "all or nothing" basis. This also applied to repeated
615 references.
616
617 (b) \R did not give a hard partial match if \r was found at the end of the
618 subject.
619
620 (c) \X did not give a hard partial match after matching one or more
621 characters at the end of the subject.
622
623 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
624 a partial match for the string "\r".
625
626 (e) When newline was set to CRLF, the metacharacter "." did not recognize
627 a partial match for a CR character at the end of the subject string.
628
629 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
630 running pcretest, the text "(JIT)" added to the output whenever JIT is
631 actually used to run the match.
632
633 7. Individual JIT compile options can be set in pcretest by following -s+[+]
634 or /S+[+] with a digit between 1 and 7.
635
636 8. OP_NOT now supports any UTF character not just single-byte ones.
637
638 9. (*MARK) control verb is now supported by the JIT compiler.
639
640 10. The command "./RunTest list" lists the available tests without actually
641 running any of them. (Because I keep forgetting what they all are.)
642
643 11. Add PCRE_INFO_MAXLOOKBEHIND.
644
645 12. Applied a (slightly modified) user-supplied patch that improves performance
646 when the heap is used for recursion (compiled with --disable-stack-for-
647 recursion). Instead of malloc and free for each heap frame each time a
648 logical recursion happens, frames are retained on a chain and re-used where
649 possible. This sometimes gives as much as 30% improvement.
650
651 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
652 call.
653
654 14. As documented, (*COMMIT) is now confined to within a positive assertion.
655
656 15. It is now possible to link pcretest with libedit as an alternative to
657 libreadline.
658
659 16. (*COMMIT) control verb is now supported by the JIT compiler.
660
661 17. The Unicode data tables have been updated to Unicode 6.1.0.
662
663 18. Added --file-list option to pcregrep.
664
665 19. Added binary file support to pcregrep, including the -a, --binary-files,
666 -I, and --text options.
667
668 20. The madvise function is renamed for posix_madvise for QNX compatibility
669 reasons. Fixed by Giuseppe D'Angelo.
670
671 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
672
673 22. Changed the option for creating symbolic links for 16-bit man pages from
674 -s to -sf so that re-installing does not cause issues.
675
676 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
677
678 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
679 matches in certain environments (the workspace was not being correctly
680 retained). Also added to pcre_dfa_exec() a simple plausibility check on
681 some of the workspace data at the beginning of a restart.
682
683 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
684 was not doing so when it should - probably a typo introduced by SVN 528
685 (change 8.10/14).
686
687 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
688 \w+ when the character tables indicated that \x{c4} was a word character.
689 There were several related cases, all because the tests for doing a table
690 lookup were testing for characters less than 127 instead of 255.
691
692 27. If a pattern contains capturing parentheses that are not used in a match,
693 their slots in the ovector are set to -1. For those that are higher than
694 any matched groups, this happens at the end of processing. In the case when
695 there were back references that the ovector was too small to contain
696 (causing temporary malloc'd memory to be used during matching), and the
697 highest capturing number was not used, memory off the end of the ovector
698 was incorrectly being set to -1. (It was using the size of the temporary
699 memory instead of the true size.)
700
701 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
702 ovector size, it uses memory at the end of the block that it has got.
703
704 29. Check for an overlong MARK name and give an error at compile time. The
705 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
706
707 30. JIT compiler update.
708
709 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
710 Rill for the patch.
711
712 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
713
714 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
715
716 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
717 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
718 was enabled.
719
720 35. Improve JIT code generation for greedy plus quantifier.
721
722 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
723 1 to "aa" instead of to an empty string. The bug affected repeated groups
724 that could potentially match an empty string.
725
726 37. Optimizing single character iterators in JIT.
727
728 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
729 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
730 codepoints that are too big for the mode are faulted, and in a UTF mode,
731 disallowed codepoints are also faulted.
732
733 39. If PCRE was compiled with UTF support, in three places in the DFA
734 matcher there was code that should only have been obeyed in UTF mode, but
735 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
736 processing when bytes with values greater than 127 were present. In 16-bit
737 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
738 both cases the values are those that cannot be the first data item in a UTF
739 character. The three items that might have provoked this were recursions,
740 possessively repeated groups, and atomic groups.
741
742 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
743 and pcregrep, because some OS require shared objects to be explicitly
744 passed to ld, causing the link step to fail if they are not.
745
746 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
747 mode, patterns that started with \h* or \R* might be incorrectly matched.
748
749
750 Version 8.30 04-February-2012
751 -----------------------------
752
753 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
754 name is defined in ctype.h.
755
756 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
757 only in quite long subpatterns.
758
759 3. Removed the function pcre_info(), which has been obsolete and deprecated
760 since it was replaced by pcre_fullinfo() in February 2000.
761
762 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
763 match a (*MARK), and the match failed at the start of the subject, a
764 reference to memory before the start of the subject could occur. This bug
765 was introduced by fix 17 of release 8.21.
766
767 5. A reference to an unset group with zero minimum repetition was giving
768 totally wrong answers (in non-JavaScript-compatibility mode). For example,
769 /(another)?(\1?)test/ matched against "hello world test". This bug was
770 introduced in release 8.13.
771
772 6. Add support for 16-bit character strings (a large amount of work involving
773 many changes and refactorings).
774
775 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
776 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
777 from a file.
778
779 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
780 rounding is not applied in this particular case).
781
782 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
783 if they appear, or are escaped, in patterns.
784
785 10. Get rid of a number of -Wunused-but-set-variable warnings.
786
787 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
788 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
789 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
790 also returns the mark "x". This bug applied to capturing parentheses,
791 non-capturing parentheses, and atomic parentheses. It also applied to some
792 assertions.
793
794 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
795 information out of configure.ac instead of relying on pcre.h.generic, which
796 is not stored in the repository.
797
798 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
799 -lreadline.
800
801 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
802
803 15. Applied Graycode's patch to put the top-level frame on the stack rather
804 than the heap when not using the stack for recursion. This gives a
805 performance improvement in many cases when recursion is not deep.
806
807 16. Experimental code added to "pcretest -C" to output the stack frame size.
808
809
810 Version 8.21 12-Dec-2011
811 ------------------------
812
813 1. Updating the JIT compiler.
814
815 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
816 are added as well.
817
818 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
819 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
820 calling _pcre_jit_exec. Some extra comments are added.
821
822 4. (*MARK) settings inside atomic groups that do not contain any capturing
823 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
824 was introduced by change 18 for 8.20.
825
826 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
827 ECMA-262 standard.
828
829 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
830 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
831 This bug was probably introduced by change 9 of 8.13.
832
833 7. While fixing 6 above, I noticed that a number of other items were being
834 incorrectly rejected as "not fixed length". This arose partly because newer
835 opcodes had not been added to the fixed-length checking code. I have (a)
836 corrected the bug and added tests for these items, and (b) arranged for an
837 error to occur if an unknown opcode is encountered while checking for fixed
838 length instead of just assuming "not fixed length". The items that were
839 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
840 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
841 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
842
843 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
844 being incorrectly compiled and would have given unpredicatble results.
845
846 9. A possessively repeated subpattern with minimum repeat count greater than
847 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
848 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
849 the first (A) could occur when it should not.
850
851 10. Add a cast and remove a redundant test from the code.
852
853 11. JIT should use pcre_malloc/pcre_free for allocation.
854
855 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
856 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
857 is anything other than /usr, -L is still shown).
858
859 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
860
861 14. Perl does not support \N without a following name in a [] class; PCRE now
862 also gives an error.
863
864 15. If a forward reference was repeated with an upper limit of around 2000,
865 it caused the error "internal error: overran compiling workspace". The
866 maximum number of forward references (including repeats) was limited by the
867 internal workspace, and dependent on the LINK_SIZE. The code has been
868 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
869 the default depends on LINK_SIZE. There is a new upper limit (for safety)
870 of around 200,000 forward references. While doing this, I also speeded up
871 the filling in of repeated forward references.
872
873 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
874 incorrectly expecting the subject to contain another "a" after the start.
875
876 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
877 in the match, the SKIP should be ignored. This was not happening; instead
878 the SKIP was being treated as NOMATCH. For patterns such as
879 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
880 tested.
881
882 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
883 now much more compatible with Perl, in particular in cases where the result
884 is a non-match for a non-anchored pattern. For example, if
885 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
886 "m", where previously it did not return a name. A side effect of this
887 change is that for partial matches, the last encountered mark name is
888 returned, as for non matches. A number of tests that were previously not
889 Perl-compatible have been moved into the Perl-compatible test files. The
890 refactoring has had the pleasing side effect of removing one argument from
891 the match() function, thus reducing its stack requirements.
892
893 19. If the /S+ option was used in pcretest to study a pattern using JIT,
894 subsequent uses of /S (without +) incorrectly behaved like /S+.
895
896 21. Retrieve executable code size support for the JIT compiler and fixing
897 some warnings.
898
899 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
900 not work when the shorter character appeared right at the end of the
901 subject string.
902
903 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
904 systems.
905
906 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
907 output it when the /M option is used in pcretest.
908
909 25. The CheckMan script was not being included in the distribution. Also, added
910 an explicit "perl" to run Perl scripts from the PrepareRelease script
911 because this is reportedly needed in Windows.
912
913 26. If study data was being save in a file and studying had not found a set of
914 "starts with" bytes for the pattern, the data written to the file (though
915 never used) was taken from uninitialized memory and so caused valgrind to
916 complain.
917
918 27. Updated RunTest.bat as provided by Sheri Pierce.
919
920 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
921
922 29. Computation of memory usage for the table of capturing group names was
923 giving an unnecessarily large value.
924
925
926 Version 8.20 21-Oct-2011
927 ------------------------
928
929 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
930 a POSIX class. After further experiments with Perl, which convinced me that
931 Perl has bugs and confusions, a closing square bracket is no longer allowed
932 in a POSIX name. This bug also affected patterns with classes that started
933 with full stops.
934
935 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
936 captured substring, but while checking the failing first alternative,
937 substring 1 is temporarily captured. If the output vector supplied to
938 pcre_exec() was not big enough for this capture, the yield of the function
939 was still zero ("insufficient space for captured substrings"). This cannot
940 be totally fixed without adding another stack variable, which seems a lot
941 of expense for a edge case. However, I have improved the situation in cases
942 such as /(a)(b)x|abc/ matched against "abc", where the return code
943 indicates that fewer than the maximum number of slots in the ovector have
944 been set.
945
946 3. Related to (2) above: when there are more back references in a pattern than
947 slots in the output vector, pcre_exec() uses temporary memory during
948 matching, and copies in the captures as far as possible afterwards. It was
949 using the entire output vector, but this conflicts with the specification
950 that only 2/3 is used for passing back captured substrings. Now it uses
951 only the first 2/3, for compatibility. This is, of course, another edge
952 case.
953
954 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
955 main code base, and can be used by building with --enable-jit. When this is
956 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
957 runtime --no-jit option is given.
958
959 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
960 ovector, the return from the function was zero, implying that there were
961 other matches that did not fit. The correct "exactly full" value is now
962 returned.
963
964 6. If a subpattern that was called recursively or as a subroutine contained
965 (*PRUNE) or any other control that caused it to give a non-standard return,
966 invalid errors such as "Error -26 (nested recursion at the same subject
967 position)" or even infinite loops could occur.
968
969 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
970 computing the minimum length on reaching *ACCEPT, and so ended up with the
971 wrong value of 1 rather than 0. Further investigation indicates that
972 computing a minimum subject length in the presence of *ACCEPT is difficult
973 (think back references, subroutine calls), and so I have changed the code
974 so that no minimum is registered for a pattern that contains *ACCEPT.
975
976 8. If (*THEN) was present in the first (true) branch of a conditional group,
977 it was not handled as intended. [But see 16 below.]
978
979 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
980 Sheri Pierce.
981
982 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
983 the first byte in a match must be "a".
984
985 11. Change 17 for 8.13 increased the recursion depth for patterns like
986 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
987 pattern contains any instances of (*THEN). If it does not, the old
988 optimizations are restored. It would be nice to do this on a per-group
989 basis, but at the moment that is not feasible.
990
991 12. In some environments, the output of pcretest -C is CRLF terminated. This
992 broke RunTest's code that checks for the link size. A single white space
993 character after the value is now allowed for.
994
995 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
996 For "fr", it uses the Windows-specific input and output files.
997
998 14. If (*THEN) appeared in a group that was called recursively or as a
999 subroutine, it did not work as intended. [But see next item.]
1000
1001 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1002 pattern fragments (but not containing any | characters). If A and B are
1003 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1004 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1005 D. In other words, Perl considers parentheses that do not contain any |
1006 characters to be part of a surrounding alternative, whereas PCRE was
1007 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1008 differently. PCRE now behaves in the same way as Perl, except in the case
1009 of subroutine/recursion calls such as (?1) which have in any case always
1010 been different (but PCRE had them first :-).
1011
1012 16. Related to 15 above: Perl does not treat the | in a conditional group as
1013 creating alternatives. Such a group is treated in the same way as an
1014 ordinary group without any | characters when processing (*THEN). PCRE has
1015 been changed to match Perl's behaviour.
1016
1017 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1018 RunGrepTest script failed.
1019
1020 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1021 inevitable for groups that contain captures, but it can lead to a lot of
1022 stack use in large patterns. The old behaviour has been restored for atomic
1023 groups that do not contain any capturing parentheses.
1024
1025 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1026 suppress the check for a minimum subject length at run time. (If it was
1027 given to pcre_exec() or pcre_dfa_exec() it did work.)
1028
1029 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1030 fail to work when decoding hex characters in data strings in EBCDIC
1031 environments.
1032
1033 21. It appears that in at least one Mac OS environment, the isxdigit() function
1034 is implemented as a macro that evaluates to its argument more than once,
1035 contravening the C 90 Standard (I haven't checked a later standard). There
1036 was an instance in pcretest which caused it to go wrong when processing
1037 \x{...} escapes in subject strings. The has been rewritten to avoid using
1038 things like p++ in the argument of isxdigit().
1039
1040
1041 Version 8.13 16-Aug-2011
1042 ------------------------
1043
1044 1. The Unicode data tables have been updated to Unicode 6.0.0.
1045
1046 2. Two minor typos in pcre_internal.h have been fixed.
1047
1048 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1049 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1050 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1051
1052 4. There were a number of related bugs in the code for matching backrefences
1053 caselessly in UTF-8 mode when codes for the characters concerned were
1054 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1055 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1056 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1057 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1058 2-byte code at the end of the subject (it thought there wasn't enough data
1059 left).
1060
1061 5. Comprehensive information about what went wrong is now returned by
1062 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1063 as the output vector has at least 2 elements. The offset of the start of
1064 the failing character and a reason code are placed in the vector.
1065
1066 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1067 now returned is for the first byte of the failing character, instead of the
1068 last byte inspected. This is an incompatible change, but I hope it is small
1069 enough not to be a problem. It makes the returned offset consistent with
1070 pcre_exec() and pcre_dfa_exec().
1071
1072 7. pcretest now gives a text phrase as well as the error number when
1073 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1074 failure, the offset and reason code are output.
1075
1076 8. When \R was used with a maximizing quantifier it failed to skip backwards
1077 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1078 back over a single character (\n). This seems wrong (because it treated the
1079 two characters as a single entity when going forwards), conflicts with the
1080 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1081 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1082 has been changed.
1083
1084 9. Some internal refactoring has changed the processing so that the handling
1085 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1086 time (the PCRE_DOTALL option was changed this way some time ago: version
1087 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1088 which was always a bit of a fudge. It also means that there is one less
1089 argument for the match() function, which reduces its stack requirements
1090 slightly. This change also fixes an incompatibility with Perl: the pattern
1091 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1092
1093 10. More internal refactoring has drastically reduced the number of recursive
1094 calls to match() for possessively repeated groups such as (abc)++ when
1095 using pcre_exec().
1096
1097 11. While implementing 10, a number of bugs in the handling of groups were
1098 discovered and fixed:
1099
1100 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1101 (a|)*(?1) gave a compile-time internal error.
1102 ((a|)+)+ did not notice that the outer group could match an empty string.
1103 (^a|^)+ was not marked as anchored.
1104 (.*a|.*)+ was not marked as matching at start or after a newline.
1105
1106 12. Yet more internal refactoring has removed another argument from the match()
1107 function. Special calls to this function are now indicated by setting a
1108 value in a variable in the "match data" data block.
1109
1110 13. Be more explicit in pcre_study() instead of relying on "default" for
1111 opcodes that mean there is no starting character; this means that when new
1112 ones are added and accidentally left out of pcre_study(), testing should
1113 pick them up.
1114
1115 14. The -s option of pcretest has been documented for ages as being an old
1116 synonym of -m (show memory usage). I have changed it to mean "force study
1117 for every regex", that is, assume /S for every regex. This is similar to -i
1118 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1119 using it. It makes it easier to run collections of tests with and without
1120 study enabled, and thereby test pcre_study() more easily. All the standard
1121 tests are now run with and without -s (but some patterns can be marked as
1122 "never study" - see 20 below).
1123
1124 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1125 restoration of the capturing data to the outer values was not happening
1126 correctly.
1127
1128 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1129 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1130 pattern had matched an empty string, and so incorrectly returned a no
1131 match.
1132
1133 17. There was optimizing code for the last branch of non-capturing parentheses,
1134 and also for the obeyed branch of a conditional subexpression, which used
1135 tail recursion to cut down on stack usage. Unfortunately, now that there is
1136 the possibility of (*THEN) occurring in these branches, tail recursion is
1137 no longer possible because the return has to be checked for (*THEN). These
1138 two optimizations have therefore been removed. [But see 8.20/11 above.]
1139
1140 18. If a pattern containing \R was studied, it was assumed that \R always
1141 matched two bytes, thus causing the minimum subject length to be
1142 incorrectly computed because \R can also match just one byte.
1143
1144 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1145 was incorrectly computed.
1146
1147 20. If /S is present twice on a test pattern in pcretest input, it now
1148 *disables* studying, thereby overriding the use of -s on the command line
1149 (see 14 above). This is necessary for one or two tests to keep the output
1150 identical in both cases.
1151
1152 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1153 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1154
1155 22. When an atomic group that contained a capturing parenthesis was
1156 successfully matched, but the branch in which it appeared failed, the
1157 capturing was not being forgotten if a higher numbered group was later
1158 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1159 group 1 to "a", when in fact it should be unset. This applied to multi-
1160 branched capturing and non-capturing groups, repeated or not, and also to
1161 positive assertions (capturing in negative assertions does not happen
1162 in PCRE) and also to nested atomic groups.
1163
1164 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1165 subject after a captured substring, to make it easier to tell which of a
1166 number of identical substrings has been captured.
1167
1168 24. The way atomic groups are processed by pcre_exec() has been changed so that
1169 if they are repeated, backtracking one repetition now resets captured
1170 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1171 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1172 "aaa". Previously, it would have been "a". As part of this code
1173 refactoring, the way recursive calls are handled has also been changed.
1174
1175 25. If an assertion condition captured any substrings, they were not passed
1176 back unless some other capturing happened later. For example, if
1177 (?(?=(a))a) was matched against "a", no capturing was returned.
1178
1179 26. When studying a pattern that contained subroutine calls or assertions,
1180 the code for finding the minimum length of a possible match was handling
1181 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1182 group 1 called group 2 while simultaneously a separate group 2 called group
1183 1). A stack overflow occurred in this case. I have fixed this by limiting
1184 the recursion depth to 10.
1185
1186 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1187 Fortmann. This supports explicit test numbers on the command line, and has
1188 argument validation and error reporting.
1189
1190 28. An instance of \X with an unlimited repeat could fail if at any point the
1191 first character it looked at was a mark character.
1192
1193 29. Some minor code refactoring concerning Unicode properties and scripts
1194 should reduce the stack requirement of match() slightly.
1195
1196 30. Added the '=' option to pcretest to check the setting of unused capturing
1197 slots at the end of the pattern, which are documented as being -1, but are
1198 not included in the return count.
1199
1200 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1201 compiled something random. Now it gives a compile-time error (as does
1202 Perl).
1203
1204 32. A *MARK encountered during the processing of a positive assertion is now
1205 recorded and passed back (compatible with Perl).
1206
1207 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1208 had alternative anchored branches, the search for a second match in a line
1209 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1210 matched the line "0102" twice. The same bug affected patterns that started
1211 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1212 twice.
1213
1214 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1215 does, and because of capturing effects, quantifying parenthesized
1216 assertions may at times be useful. Quantifiers are now allowed for
1217 parenthesized assertions.
1218
1219 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1220
1221 36. \g was being checked for fancy things in a character class, when it should
1222 just be a literal "g".
1223
1224 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1225 appearance of a nested POSIX class supersedes an apparent external class.
1226 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1227 unescaped square brackets may also appear as part of class names. For
1228 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1229 more like Perl. (But see 8.20/1 above.)
1230
1231 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1232 was because it thought it was \N{name}, which is not supported).
1233
1234 39. Add minix to OS list not supporting the -S option in pcretest.
1235
1236 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1237 cannot analyze patterns in sufficient detail to catch mutual recursions
1238 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1239 subgroup is called recursively as a subpattern for a second time at the
1240 same position in the subject string. In previous releases this might have
1241 been caught by the recursion limit, or it might have run out of stack.
1242
1243 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1244 happen only once. PCRE was, however incorrectly giving a compile time error
1245 "recursive call could loop indefinitely" because it cannot analyze the
1246 pattern in sufficient detail. The compile time test no longer happens when
1247 PCRE is compiling a conditional subpattern, but actual runaway loops are
1248 now caught at runtime (see 40 above).
1249
1250 42. It seems that Perl allows any characters other than a closing parenthesis
1251 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1252 has been changed to be the same.
1253
1254 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1255 as not to get warnings when autogen.sh is called. Also changed
1256 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1257
1258 44. To help people who use pcregrep to scan files containing exceedingly long
1259 lines, the following changes have been made:
1260
1261 (a) The default value of the buffer size parameter has been increased from
1262 8K to 20K. (The actual buffer used is three times this size.)
1263
1264 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1265 PCRE is built.
1266
1267 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1268 to be set at run time.
1269
1270 (d) Numerical values in pcregrep options can be followed by K or M, for
1271 example --buffer-size=50K.
1272
1273 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1274 given and the return code is set to 2.
1275
1276 45. Add a pointer to the latest mark to the callout data block.
1277
1278 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1279 partial match of an empty string instead of no match. This was specific to
1280 the use of ".".
1281
1282 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1283 complete match instead of a partial match. This bug was dependent on both
1284 the PCRE_UTF8 and PCRE_DOTALL options being set.
1285
1286 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1287 starting byte set, because \b was not being ignored.
1288
1289
1290 Version 8.12 15-Jan-2011
1291 ------------------------
1292
1293 1. Fixed some typos in the markup of the man pages, and wrote a script that
1294 checks for such things as part of the documentation building process.
1295
1296 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1297 --match-limit and --recursion-limit options (added for 8.11). In
1298 particular, this made one of the standard tests fail. (The integer value
1299 went into the wrong half of a long int.)
1300
1301 3. If the --colour option was given to pcregrep with -v (invert match), it
1302 did strange things, either producing crazy output, or crashing. It should,
1303 of course, ignore a request for colour when reporting lines that do not
1304 match.
1305
1306 4. Another pcregrep bug caused similar problems if --colour was specified with
1307 -M (multiline) and the pattern match finished with a line ending.
1308
1309 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1310 matched in multiline mode, the following line was shown as part of the
1311 match. This seems wrong, so I have changed it.
1312
1313 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1314 the check for further matches in the same line (so they could be coloured)
1315 to overrun the end of the current line. If another match was found, it was
1316 incorrectly shown (and then shown again when found in the next line).
1317
1318 7. If pcregrep was compiled under Windows, there was a reference to the
1319 function pcregrep_exit() before it was defined. I am assuming this was
1320 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1321 reported by a user. I've moved the definition above the reference.
1322
1323
1324 Version 8.11 10-Dec-2010
1325 ------------------------
1326
1327 1. (*THEN) was not working properly if there were untried alternatives prior
1328 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1329 backtracked to try for "b" instead of moving to the next alternative branch
1330 at the same level (in this case, to look for "c"). The Perl documentation
1331 is clear that when (*THEN) is backtracked onto, it goes to the "next
1332 alternative in the innermost enclosing group".
1333
1334 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1335 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1336 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1337 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1338 (*THEN).
1339
1340 3. If \s appeared in a character class, it removed the VT character from
1341 the class, even if it had been included by some previous item, for example
1342 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1343 of \s, but is part of the POSIX "space" class.)
1344
1345 4. A partial match never returns an empty string (because you can always
1346 match an empty string at the end of the subject); however the checking for
1347 an empty string was starting at the "start of match" point. This has been
1348 changed to the "earliest inspected character" point, because the returned
1349 data for a partial match starts at this character. This means that, for
1350 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1351 (previously it gave "no match").
1352
1353 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1354 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1355 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1356 has an implication that the given string is incomplete (because a partial
1357 match is preferred over a full match). For this reason, these items now
1358 give a partial match in this situation. [Aside: previously, the one case
1359 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1360 match rather than a full match, which was wrong by the old rules, but is
1361 now correct.]
1362
1363 6. There was a bug in the handling of #-introduced comments, recognized when
1364 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1365 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1366 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1367 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1368 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1369 places in pcre_compile().
1370
1371 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1372 comments when looking ahead for named forward references to subpatterns,
1373 the only newline sequence it recognized was NL. It now handles newlines
1374 according to the set newline convention.
1375
1376 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1377 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1378 cater for a lack of strerror(). These oversights have been fixed.
1379
1380 9. Added --match-limit and --recursion-limit to pcregrep.
1381
1382 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1383
1384 11. When the -o option was used, pcregrep was setting a return code of 1, even
1385 when matches were found, and --line-buffered was not being honoured.
1386
1387 12. Added an optional parentheses number to the -o and --only-matching options
1388 of pcregrep.
1389
1390 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1391 can match an empty string. The code to do it in pcretest and pcredemo
1392 needed fixing:
1393
1394 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1395 only one byte after an empty string match just before CRLF (this case
1396 just got forgotten; "any" and "anycrlf" were OK).
1397
1398 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1399 mode when an empty string match preceded an ASCII character followed by
1400 a non-ASCII character. (The code for advancing by one character rather
1401 than one byte was nonsense.)
1402
1403 (c) The pcredemo.c sample program did not have any code at all to handle
1404 the cases when CRLF is a valid newline sequence.
1405
1406 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1407 as a starting offset was within the subject string. There is now a new
1408 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1409 negative or greater than the length of the string. In order to test this,
1410 pcretest is extended to allow the setting of negative starting offsets.
1411
1412 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1413 starting offset points to the beginning of a UTF-8 character was
1414 unnecessarily clumsy. I tidied it up.
1415
1416 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1417 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1418
1419 17. Nobody had reported that the --include_dir option, which was added in
1420 release 7.7 should have been called --include-dir (hyphen, not underscore)
1421 for compatibility with GNU grep. I have changed it to --include-dir, but
1422 left --include_dir as an undocumented synonym, and the same for
1423 --exclude-dir, though that is not available in GNU grep, at least as of
1424 release 2.5.4.
1425
1426 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1427 characters from a string of bytes) have been redefined so as not to use
1428 loops, in order to improve performance in some environments. At the same
1429 time, I abstracted some of the common code into auxiliary macros to save
1430 repetition (this should not affect the compiled code).
1431
1432 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1433 compile-time error is now given if \c is not followed by an ASCII
1434 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1435 different, and any byte value is allowed.)
1436
1437 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1438 START_OPTIMIZE option, which is now allowed at compile time - but just
1439 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1440 to pcregrep and other applications that have no direct access to PCRE
1441 options. The new /Y option in pcretest sets this option when calling
1442 pcre_compile().
1443
1444 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1445 back references. Groups containing recursive back references were forced to
1446 be atomic by that change, but in the case of named groups, the amount of
1447 memory required was incorrectly computed, leading to "Failed: internal
1448 error: code overflow". This has been fixed.
1449
1450 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1451 pcretest.c, to avoid build problems in some Borland environments.
1452
1453
1454 Version 8.10 25-Jun-2010
1455 ------------------------
1456
1457 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1458 THEN.
1459
1460 2. (*ACCEPT) was not working when inside an atomic group.
1461
1462 3. Inside a character class, \B is treated as a literal by default, but
1463 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1464 causes the error). The code is unchanged, but I tidied the documentation.
1465
1466 4. Inside a character class, PCRE always treated \R and \X as literals,
1467 whereas Perl faults them if its -w option is set. I have changed PCRE so
1468 that it faults them when PCRE_EXTRA is set.
1469
1470 5. Added support for \N, which always matches any character other than
1471 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1472
1473 6. When compiling pcregrep with newer versions of gcc which may have
1474 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1475 declared with attribute warn_unused_result" were given. Just casting the
1476 result to (void) does not stop the warnings; a more elaborate fudge is
1477 needed. I've used a macro to implement this.
1478
1479 7. Minor change to pcretest.c to avoid a compiler warning.
1480
1481 8. Added four artifical Unicode properties to help with an option to make
1482 \s etc use properties (see next item). The new properties are: Xan
1483 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1484
1485 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1486 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1487 this option. Modified pcretest to add /W to test this facility. Added
1488 REG_UCP to make it available via the POSIX interface.
1489
1490 10. Added --line-buffered to pcregrep.
1491
1492 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1493 studied, and the match started with a letter with a code point greater than
1494 127 whose first byte was different to the first byte of the other case of
1495 the letter, the other case of this starting letter was not recognized
1496 (#976).
1497
1498 12. If a pattern that was studied started with a repeated Unicode property
1499 test, for example, \p{Nd}+, there was the theoretical possibility of
1500 setting up an incorrect bitmap of starting bytes, but fortunately it could
1501 not have actually happened in practice until change 8 above was made (it
1502 added property types that matched character-matching opcodes).
1503
1504 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1505 possible starting bytes for non-anchored patterns.
1506
1507 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1508 \R, and also a number of cases that involve Unicode properties, both
1509 explicit and implicit when PCRE_UCP is set.
1510
1511 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1512 input, it could crash or give wrong results if characters with values
1513 greater than 0xc0 were present in the subject string. (Detail: it assumed
1514 UTF-8 input when processing these items.)
1515
1516 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1517 size_t is 64-bit (#991).
1518
1519 17. Added a check for running out of memory when PCRE is compiled with
1520 --disable-stack-for-recursion (#990).
1521
1522 18. If the last data line in a file for pcretest does not have a newline on
1523 the end, a newline was missing in the output.
1524
1525 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1526 less than 128) in its various bitmaps. However, there is a facility for
1527 generating tables according to the current locale when PCRE is compiled. It
1528 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1529 characters, are recognized by isspace() and therefore were getting set in
1530 these tables, and indeed these tables seem to approximate to ISO 8859. This
1531 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1532 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1533 which of course cannot start UTF-8 characters. I have changed the code so
1534 that only real ASCII characters (less than 128) and the correct starting
1535 bytes for UTF-8 encodings are set for characters greater than 127 when in
1536 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1537 altogether.)
1538
1539 20. Added the /T option to pcretest so as to be able to run tests with non-
1540 standard character tables, thus making it possible to include the tests
1541 used for 19 above in the standard set of tests.
1542
1543 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1544 reference to a subpattern the other side of a comment that contains an
1545 opening parenthesis caused either an internal compiling error, or a
1546 reference to the wrong subpattern.
1547
1548
1549 Version 8.02 19-Mar-2010
1550 ------------------------
1551
1552 1. The Unicode data tables have been updated to Unicode 5.2.0.
1553
1554 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1555 configured.
1556
1557 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1558 original author of that file, following a query about its status.
1559
1560 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1561 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1562
1563 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1564 quantifier applied to a forward-referencing subroutine call, could compile
1565 incorrect code or give the error "internal error: previously-checked
1566 referenced subpattern not found".
1567
1568 6. Both MS Visual Studio and Symbian OS have problems with initializing
1569 variables to point to external functions. For these systems, therefore,
1570 pcre_malloc etc. are now initialized to local functions that call the
1571 relevant global functions.
1572
1573 7. There were two entries missing in the vectors called coptable and poptable
1574 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1575 I've fixed the data, and added a kludgy way of testing at compile time that
1576 the lengths are correct (equal to the number of opcodes).
1577
1578 8. Following on from 7, I added a similar kludge to check the length of the
1579 eint vector in pcreposix.c.
1580
1581 9. Error texts for pcre_compile() are held as one long string to avoid too
1582 much relocation at load time. To find a text, the string is searched,
1583 counting zeros. There was no check for running off the end of the string,
1584 which could happen if a new error number was added without updating the
1585 string.
1586
1587 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1588
1589 11. \K was not working if it appeared in an atomic group or in a group that
1590 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1591 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1592 the assertion is positive, but not if it is negative.
1593
1594 12. Change 11 fortuitously reduced the size of the stack frame used in the
1595 "match()" function of pcre_exec.c by one pointer. Forthcoming
1596 implementation of support for (*MARK) will need an extra pointer on the
1597 stack; I have reserved it now, so that the stack frame size does not
1598 decrease.
1599
1600 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1601 item in branch that calls a recursion is a subroutine call - as in the
1602 second branch in the above example - was incorrectly given the compile-
1603 time error "recursive call could loop indefinitely" because pcre_compile()
1604 was not correctly checking the subroutine for matching a non-empty string.
1605
1606 14. The checks for overrunning compiling workspace could trigger after an
1607 overrun had occurred. This is a "should never occur" error, but it can be
1608 triggered by pathological patterns such as hundreds of nested parentheses.
1609 The checks now trigger 100 bytes before the end of the workspace.
1610
1611 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1612
1613
1614 Version 8.01 19-Jan-2010
1615 ------------------------
1616
1617 1. If a pattern contained a conditional subpattern with only one branch (in
1618 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1619 computed the wrong minimum data length (which is of course zero for such
1620 subpatterns). This could cause incorrect "no match" results.
1621
1622 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1623 the pattern is reset in the first branch, pcre_compile() failed with
1624 "internal error: code overflow at offset...". This happened only when
1625 the reset was to the original external option setting. (An optimization
1626 abstracts leading options settings into an external setting, which was the
1627 cause of this.)
1628
1629 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1630 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1631 assertion pattern did not match (meaning that the assertion was true), it
1632 was incorrectly treated as false if the SKIP had been reached during the
1633 matching. This also applied to assertions used as conditions.
1634
1635 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1636 assertion subpattern, including such a pattern used as a condition,
1637 unpredictable results occurred, instead of the error return
1638 PCRE_ERROR_DFA_UITEM.
1639
1640 5. The C++ GlobalReplace function was not working like Perl for the special
1641 situation when an empty string is matched. It now does the fancy magic
1642 stuff that is necessary.
1643
1644 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1645 removed. (These were left over from very, very early versions of PCRE.)
1646
1647 7. Some cosmetic changes to the code to make life easier when compiling it
1648 as part of something else:
1649
1650 (a) Change DEBUG to PCRE_DEBUG.
1651
1652 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1653 called "current" as "current_branch", to prevent a collision with the
1654 Linux macro when compiled as a kernel module.
1655
1656 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1657 prevent a collision with the Linux macro when compiled as a kernel
1658 module.
1659
1660 8. In pcre_compile() there are some checks for integer overflows that used to
1661 cast potentially large values to (double). This has been changed to that
1662 when building, a check for int64_t is made, and if it is found, it is used
1663 instead, thus avoiding the use of floating point arithmetic. (There is no
1664 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1665 double.
1666
1667 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1668 2005 (difference between two addresses compared to an unsigned value).
1669
1670 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1671 custom one, because of the following reported problem in Windows:
1672
1673 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1674 under Win32.
1675 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1676 therefore missing the function definition.
1677 - The compiler thus generates a "C" signature for the test function.
1678 - The linker fails to find the "C" function.
1679 - PCRE fails to configure if asked to do so against libbz2.
1680
1681 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1682 messages were output:
1683
1684 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1685 rerunning libtoolize, to keep the correct libtool macros in-tree.
1686 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1687
1688 I have done both of these things.
1689
1690 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1691 most of the time, it *can* run out if it is given a pattern that contains a
1692 runaway infinite recursion. I updated the discussion in the pcrestack man
1693 page.
1694
1695 13. Now that we have gone to the x.xx style of version numbers, the minor
1696 version may start with zero. Using 08 or 09 is a bad idea because users
1697 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1698 interpreted as invalid octal numbers. I've updated the previous comment in
1699 configure.ac, and also added a check that gives an error if 08 or 09 are
1700 used.
1701
1702 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1703 causing partial matching to fail when the end of the subject matched \W
1704 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1705
1706 15. There were some discrepancies between the declarations in pcre_internal.h
1707 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1708 their definitions. The declarations used "const uschar *" and the
1709 definitions used USPTR. Even though USPTR is normally defined as "const
1710 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1711 reported that: "This difference in casting confuses some C++ compilers, for
1712 example, SunCC recognizes above declarations as different functions and
1713 generates broken code for hbpcre." I have changed the declarations to use
1714 USPTR.
1715
1716 16. GNU libtool is named differently on some systems. The autogen.sh script now
1717 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1718 (FreeBSD).
1719
1720 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1721 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1722 comment: "Figure out how to create a longlong from a string: strtoll and
1723 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1724 instance, but it only takes 2 args instead of 3!"
1725
1726 18. A subtle bug concerned with back references has been fixed by a change of
1727 specification, with a corresponding code fix. A pattern such as
1728 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1729 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1730 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1731 same bug. Such groups have to be quantified to be useful, or contained
1732 inside another quantified group. (If there's no repetition, the reference
1733 can never match.) The problem arises because, having left the group and
1734 moved on to the rest of the pattern, a later failure that backtracks into
1735 the group uses the captured value from the final iteration of the group
1736 rather than the correct earlier one. I have fixed this in PCRE by forcing
1737 any group that contains a reference to itself to be an atomic group; that
1738 is, there cannot be any backtracking into it once it has completed. This is
1739 similar to recursive and subroutine calls.
1740
1741
1742 Version 8.00 19-Oct-09
1743 ----------------------
1744
1745 1. The table for translating pcre_compile() error codes into POSIX error codes
1746 was out-of-date, and there was no check on the pcre_compile() error code
1747 being within the table. This could lead to an OK return being given in
1748 error.
1749
1750 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1751 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1752 in a Windows environment.
1753
1754 3. The pcregrep --count option prints the count for each file even when it is
1755 zero, as does GNU grep. However, pcregrep was also printing all files when
1756 --files-with-matches was added. Now, when both options are given, it prints
1757 counts only for those files that have at least one match. (GNU grep just
1758 prints the file name in this circumstance, but including the count seems
1759 more useful - otherwise, why use --count?) Also ensured that the
1760 combination -clh just lists non-zero counts, with no names.
1761
1762 4. The long form of the pcregrep -F option was incorrectly implemented as
1763 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1764 but it seems right to fix it, and I didn't think it was worth preserving
1765 the old behaviour.
1766
1767 5. The command line items --regex=pattern and --regexp=pattern were not
1768 recognized by pcregrep, which required --regex pattern or --regexp pattern
1769 (with a space rather than an '='). The man page documented the '=' forms,
1770 which are compatible with GNU grep; these now work.
1771
1772 6. No libpcreposix.pc file was created for pkg-config; there was just
1773 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1774
1775 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1776 when UCP support is not needed, by modifying the Python script that
1777 generates it from Unicode data files. This should not matter if the module
1778 is correctly used as a library, but I received one complaint about 50K of
1779 unwanted data. My guess is that the person linked everything into his
1780 program rather than using a library. Anyway, it does no harm.
1781
1782 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1783 was a minimum greater than 1 for a wide character in a possessive
1784 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1785 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1786 character. Chaos in the form of incorrect output or a compiling loop could
1787 result.
1788
1789 9. The restrictions on what a pattern can contain when partial matching is
1790 requested for pcre_exec() have been removed. All patterns can now be
1791 partially matched by this function. In addition, if there are at least two
1792 slots in the offset vector, the offset of the earliest inspected character
1793 for the match and the offset of the end of the subject are set in them when
1794 PCRE_ERROR_PARTIAL is returned.
1795
1796 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1797 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1798 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1799 and may be more useful for multi-segment matching.
1800
1801 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1802 used to be given if ever the end of the subject was reached; now it is
1803 given only if matching could not proceed because another character was
1804 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1805 string "Z", which now yields "no match" instead of "partial match". In the
1806 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1807 final character ended with (*FAIL).
1808
1809 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1810 if the pattern had a "must contain" character that was already found in the
1811 earlier partial match, unless partial matching was again requested. For
1812 example, with the pattern /dog.(body)?/, the "must contain" character is
1813 "g". If the first part-match was for the string "dog", restarting with
1814 "sbody" failed. This bug has been fixed.
1815
1816 13. The string returned by pcre_dfa_exec() after a partial match has been
1817 changed so that it starts at the first inspected character rather than the
1818 first character of the match. This makes a difference only if the pattern
1819 starts with a lookbehind assertion or \b or \B (\K is not supported by
1820 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1821 matching functions compatible, and I think it's the right thing to do.
1822
1823 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1824 so that the demonstration program is easily available in environments where
1825 PCRE has not been installed from source.
1826
1827 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1828 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1829 library.
1830
1831 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1832 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1833 is not the first non-POSIX option to be added. Clearly some people find
1834 these options useful.
1835
1836 17. If a caller to the POSIX matching function regexec() passes a non-zero
1837 value for nmatch with a NULL value for pmatch, the value of
1838 nmatch is forced to zero.
1839
1840 18. RunGrepTest did not have a test for the availability of the -u option of
1841 the diff command, as RunTest does. It now checks in the same way as
1842 RunTest, and also checks for the -b option.
1843
1844 19. If an odd number of negated classes containing just a single character
1845 interposed, within parentheses, between a forward reference to a named
1846 subpattern and the definition of the subpattern, compilation crashed with
1847 an internal error, complaining that it could not find the referenced
1848 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1849 [The bug was that it was starting one character too far in when skipping
1850 over the character class, thus treating the ] as data rather than
1851 terminating the class. This meant it could skip too much.]
1852
1853 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1854 /g option in pcretest when the pattern contains \K, which makes it possible
1855 to have an empty string match not at the start, even when the pattern is
1856 anchored. Updated pcretest and pcredemo to use this option.
1857
1858 21. If the maximum number of capturing subpatterns in a recursion was greater
1859 than the maximum at the outer level, the higher number was returned, but
1860 with unset values at the outer level. The correct (outer level) value is
1861 now given.
1862
1863 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1864 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1865 make it do so. The string so far is captured, making this feature
1866 compatible with Perl.
1867
1868 23. The tests have been re-organized, adding tests 11 and 12, to make it
1869 possible to check the Perl 5.10 features against Perl 5.10.
1870
1871 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1872 pattern matches a fixed length string. PCRE did not allow this; now it
1873 does. Neither allows recursion.
1874
1875 25. I finally figured out how to implement a request to provide the minimum
1876 length of subject string that was needed in order to match a given pattern.
1877 (It was back references and recursion that I had previously got hung up
1878 on.) This code has now been added to pcre_study(); it finds a lower bound
1879 to the length of subject needed. It is not necessarily the greatest lower
1880 bound, but using it to avoid searching strings that are too short does give
1881 some useful speed-ups. The value is available to calling programs via
1882 pcre_fullinfo().
1883
1884 26. While implementing 25, I discovered to my embarrassment that pcretest had
1885 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1886 study optimizations had never been tested with that matching function.
1887 Oops. What is worse, even when it was passed study data, there was a bug in
1888 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1889 were also very few tests of studied patterns with pcre_dfa_exec().
1890
1891 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1892 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1893 on the other side of the coin, they are no longer allowed to have different
1894 names, because these cannot be distinguished in PCRE, and this has caused
1895 confusion. (This is a difference from Perl.)
1896
1897 28. When duplicate subpattern names are present (necessarily with different
1898 numbers, as required by 27 above), and a test is made by name in a
1899 conditional pattern, either for a subpattern having been matched, or for
1900 recursion in such a pattern, all the associated numbered subpatterns are
1901 tested, and the overall condition is true if the condition is true for any
1902 one of them. This is the way Perl works, and is also more like the way
1903 testing by number works.
1904
1905
1906 Version 7.9 11-Apr-09
1907 ---------------------
1908
1909 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1910 (pcretest), all targets were linked against these libraries. This included
1911 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1912 libraries. This caused unwanted dependencies to be created. This problem
1913 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1914 pcretest is linked with readline.
1915
1916 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1917 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1918 moved outside it again, because FALSE and TRUE are already defined in AIX,
1919 but BOOL is not.
1920
1921 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1922 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1923
1924 4. The pcregrep documentation said spaces were inserted as well as colons (or
1925 hyphens) following file names and line numbers when outputting matching
1926 lines. This is not true; no spaces are inserted. I have also clarified the
1927 wording for the --colour (or --color) option.
1928
1929 5. In pcregrep, when --colour was used with -o, the list of matching strings
1930 was not coloured; this is different to GNU grep, so I have changed it to be
1931 the same.
1932
1933 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1934 each matching line was coloured. Now it goes on to look for further matches
1935 of any of the test patterns, which is the same behaviour as GNU grep.
1936
1937 7. A pattern that could match an empty string could cause pcregrep to loop; it
1938 doesn't make sense to accept an empty string match in pcregrep, so I have
1939 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1940 seems to be how GNU grep behaves. [But see later change 40 for release
1941 8.33.]
1942
1943 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1944 start or after a newline", because the conditional assertion was not being
1945 correctly handled. The rule now is that both the assertion and what follows
1946 in the first alternative must satisfy the test.
1947
1948 9. If auto-callout was enabled in a pattern with a conditional group whose
1949 condition was an assertion, PCRE could crash during matching, both with
1950 pcre_exec() and pcre_dfa_exec().
1951
1952 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1953 used for matching.
1954
1955 11. Unicode property support in character classes was not working for
1956 characters (bytes) greater than 127 when not in UTF-8 mode.
1957
1958 12. Added the -M command line option to pcretest.
1959
1960 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1961
1962 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1963
1964 16. Added comments and documentation about mis-use of no_arg in the C++
1965 wrapper.
1966
1967 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1968 from Martin Jerabek that uses macro names for all relevant character and
1969 string constants.
1970
1971 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1972 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1973 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1974 these, but not everybody uses configure.
1975
1976 19. A conditional group that had only one branch was not being correctly
1977 recognized as an item that could match an empty string. This meant that an
1978 enclosing group might also not be so recognized, causing infinite looping
1979 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1980 with the subject "ab", where knowledge that the repeated group can match
1981 nothing is needed in order to break the loop.
1982
1983 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1984 exec(), but without supplying a callout function, matching went wrong.
1985
1986 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1987 leak if the size of the offset vector was greater than 30. When the vector
1988 is smaller, the saved offsets during recursion go onto a local stack
1989 vector, but for larger vectors malloc() is used. It was failing to free
1990 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1991 error, in fact).
1992
1993 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1994 heapframe that is used only when UTF-8 support is enabled. This caused no
1995 problem, but was untidy.
1996
1997 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1998 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1999 included within another project.
2000
2001 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2002 slightly modified by me:
2003
2004 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2005 not building pcregrep.
2006
2007 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2008 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2009
2010 25. Forward references, both numeric and by name, in patterns that made use of
2011 duplicate group numbers, could behave incorrectly or give incorrect errors,
2012 because when scanning forward to find the reference group, PCRE was not
2013 taking into account the duplicate group numbers. A pattern such as
2014 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2015
2016 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2017 the feature of a custom pointer more persuasive (as requested by a user).
2018
2019 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2020 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2021 pre-defined.
2022
2023 28. Added support for (*UTF8) at the start of a pattern.
2024
2025 29. Arrange for flags added by the "release type" setting in CMake to be shown
2026 in the configuration summary.
2027
2028
2029 Version 7.8 05-Sep-08
2030 ---------------------
2031
2032 1. Replaced UCP searching code with optimized version as implemented for Ad
2033 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2034 stage table and inline lookup instead of a function, giving speed ups of 2
2035 to 5 times on some simple patterns that I tested. Permission was given to
2036 distribute the MultiStage2.py script that generates the tables (it's not in
2037 the tarball, but is in the Subversion repository).
2038
2039 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2040 scripts.
2041
2042 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2043 a group with a zero qualifier. The result of the study could be incorrect,
2044 or the function might crash, depending on the pattern.
2045
2046 4. Caseless matching was not working for non-ASCII characters in back
2047 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2048 It now works when Unicode Property Support is available.
2049
2050 5. In pcretest, an escape such as \x{de} in the data was always generating
2051 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2052 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2053 truncation.
2054
2055 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2056
2057 7. Added two (int) casts to pcregrep when printing the difference of two
2058 pointers, in case they are 64-bit values.
2059
2060 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2061 test 2 if it fails.
2062
2063 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2064 and a #define of that name to empty if it is not externally set. This is to
2065 allow users of MSVC to set it if necessary.
2066
2067 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2068 the convenience functions in the pcre_get.c source file.
2069
2070 11. An option change at the start of a pattern that had top-level alternatives
2071 could cause overwriting and/or a crash. This command provoked a crash in
2072 some environments:
2073
2074 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2075
2076 This potential security problem was recorded as CVE-2008-2371.
2077
2078 12. For a pattern where the match had to start at the beginning or immediately
2079 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2080 pcre_dfa_exec() could read past the end of the passed subject if there was
2081 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2082 pcretest so that it places the subject at the end of its malloc-ed buffer.
2083
2084 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2085 exec() might read past the end of the data buffer in UTF-8 mode.
2086
2087 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2088 the data contained the byte 0x85 as part of a UTF-8 character within its
2089 first line. This applied both to normal and DFA matching.
2090
2091 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2092 /^[^d]*?$/8 failed to match "abc".
2093
2094 16. Added a missing copyright notice to pcrecpp_internal.h.
2095
2096 17. Make it more clear in the documentation that values returned from
2097 pcre_exec() in ovector are byte offsets, not character counts.
2098
2099 18. Tidied a few places to stop certain compilers from issuing warnings.
2100
2101 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2102 supplied by Stefan Weber. I made a further small update for 7.8 because
2103 there is a change of source arrangements: the pcre_searchfuncs.c module is
2104 replaced by pcre_ucd.c.
2105
2106
2107 Version 7.7 07-May-08
2108 ---------------------
2109
2110 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2111 a string to a long long, pretend we don't even have a long long." This is
2112 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2113
2114 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2115 pre-7.6 versions, which defined a global no_arg variable instead of putting
2116 it in the RE class. (See also #8 below.)
2117
2118 3. Remove a line of dead code, identified by coverity and reported by Nuno
2119 Lopes.
2120
2121 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2122
2123 (1) The include/exclude patterns were being applied to the whole pathnames
2124 of files, instead of just to the final components.
2125
2126 (2) If there was more than one level of directory, the subdirectories were
2127 skipped unless they satisfied the include/exclude conditions. This is
2128 inconsistent with GNU grep (and could even be seen as contrary to the
2129 pcregrep specification - which I improved to make it absolutely clear).
2130 The action now is always to scan all levels of directory, and just
2131 apply the include/exclude patterns to regular files.
2132
2133 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2134 --exclude_dir in the tests to avoid scanning .svn directories.
2135
2136 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2137 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2138 doesn't support NULs in patterns.
2139
2140 7. Added some missing "const"s to declarations of static tables in
2141 pcre_compile.c and pcre_dfa_exec.c.
2142
2143 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2144 caused by fix #2 above. (Subsequently also a second patch to fix the
2145 first patch. And a third patch - this was a messy problem.)
2146
2147 9. Applied Craig's patch to remove the use of push_back().
2148
2149 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2150 matching function regexec().
2151
2152 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2153 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2154 references. PCRE supports relative numbers with this syntax (I don't think
2155 Oniguruma does).
2156
2157 12. Previously, a group with a zero repeat such as (...){0} was completely
2158 omitted from the compiled regex. However, this means that if the group
2159 was called as a subroutine from elsewhere in the pattern, things went wrong
2160 (an internal error was given). Such groups are now left in the compiled
2161 pattern, with a new opcode that causes them to be skipped at execution
2162 time.
2163
2164 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2165 to the way PCRE behaves:
2166
2167 (a) A lone ] character is dis-allowed (Perl treats it as data).
2168
2169 (b) A back reference to an unmatched subpattern matches an empty string
2170 (Perl fails the current match path).
2171
2172 (c) A data ] in a character class must be notated as \] because if the
2173 first data character in a class is ], it defines an empty class. (In
2174 Perl it is not possible to have an empty class.) The empty class []
2175 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2176 The negative empty class [^] matches any one character, independently
2177 of the DOTALL setting.
2178
2179 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2180 non-existent subpattern following a character class starting with ']' and
2181 containing () gave an internal compiling error instead of "reference to
2182 non-existent subpattern". Fortunately, when the pattern did exist, the
2183 compiled code was correct. (When scanning forwards to check for the
2184 existence of the subpattern, it was treating the data ']' as terminating
2185 the class, so got the count wrong. When actually compiling, the reference
2186 was subsequently set up correctly.)
2187
2188 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2189 it was being rejected as not supported by pcre_dfa_exec(), even though
2190 other assertions are supported. I have made pcre_dfa_exec() support
2191 (*FAIL).
2192
2193 16. The implementation of 13c above involved the invention of a new opcode,
2194 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2195 cannot be changed at match time, I realized I could make a small
2196 improvement to matching performance by compiling OP_ALLANY instead of
2197 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2198 on the OP_ANY path.
2199
2200 17. Compiling pcretest on Windows with readline support failed without the
2201 following two fixes: (1) Make the unistd.h include conditional on
2202 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2203
2204 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2205 ncurses library to be included for pcretest when ReadLine support is
2206 requested, but also to allow for it to be overridden. This patch came from
2207 Daniel Bergström.
2208
2209 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2210 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2211 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2212 spotting this.
2213
2214
2215 Version 7.6 28-Jan-08
2216 ---------------------
2217
2218 1. A character class containing a very large number of characters with
2219 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2220 overflow.
2221
2222 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2223 HAVE_LONG_LONG is not defined.
2224
2225 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2226 bring them up to date and include new features. This patch includes:
2227
2228 - Fixed PH's badly added libz and libbz2 support.
2229 - Fixed a problem with static linking.
2230 - Added pcredemo. [But later removed - see 7 below.]
2231 - Fixed dftables problem and added an option.
2232 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2233 HAVE_LONG_LONG.
2234 - Added readline support for pcretest.
2235 - Added an listing of the option settings after cmake has run.
2236
2237 4. A user submitted a patch to Makefile that makes it easy to create
2238 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2239 Makefile.am that cause it to include this special target, without
2240 affecting anything else. Note that the same mingw target plus all
2241 the other distribution libraries and programs are now supported
2242 when configuring with CMake (see 6 below) instead of with
2243 Configure/Make.
2244
2245 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2246 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2247 exported in the Windows port". It has not yet been confirmed that the patch
2248 solves the problem, but it does no harm.
2249
2250 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2251 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2252 with CMake, and also correct the comment about stack recursion.
2253
2254 7. Remove the automatic building of pcredemo from the ./configure system and
2255 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2256 of a program that users should build themselves after PCRE is installed, so
2257 building it automatically is not really right. What is more, it gave
2258 trouble in some build environments.
2259
2260 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2261
2262
2263 Version 7.5 10-Jan-08
2264 ---------------------
2265
2266 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2267 values in parens when parsing an RE using the C++ wrapper."
2268
2269 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2270 Characters greater than 255 were excluded from the class instead of being
2271 included.
2272
2273 3. The same bug as (2) above applied to negated POSIX classes such as
2274 [:^space:].
2275
2276 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2277 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2278 I have changed it.
2279
2280 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2281 first named subpattern) and a construct such as (?&a) would reference the
2282 first named subpattern whose name started with "a" (in other words, the
2283 length check was missing). Both these problems are fixed. "Subpattern name
2284 expected" is now given for (?&) (a zero-length name), and this patch also
2285 makes it give the same error for \k'' (previously it complained that that
2286 was a reference to a non-existent subpattern).
2287
2288 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2289 this is right because (?- can be followed by option settings as well as by
2290 digits. I have, however, made the messages clearer.
2291
2292 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2293 than the number used in the conditional) now cause a compile-time error.
2294 This is actually not compatible with Perl, which accepts such patterns, but
2295 treats the conditional as always being FALSE (as PCRE used to), but it
2296 seems to me that giving a diagnostic is better.
2297
2298 8. Change "alphameric" to the more common word "alphanumeric" in comments
2299 and messages.
2300
2301 9. Fix two occurrences of "backslash" in comments that should have been
2302 "backspace".
2303
2304 10. Remove two redundant lines of code that can never be obeyed (their function
2305 was moved elsewhere).
2306
2307 11. The program that makes PCRE's Unicode character property table had a bug
2308 which caused it to generate incorrect table entries for sequences of
2309 characters that have the same character type, but are in different scripts.
2310 It amalgamated them into a single range, with the script of the first of
2311 them. In other words, some characters were in the wrong script. There were
2312 thirteen such cases, affecting characters in the following ranges:
2313
2314 U+002b0 - U+002c1
2315 U+0060c - U+0060d
2316 U+0061e - U+00612
2317 U+0064b - U+0065e
2318 U+0074d - U+0076d
2319 U+01800 - U+01805
2320 U+01d00 - U+01d77
2321 U+01d9b - U+01dbf
2322 U+0200b - U+0200f
2323 U+030fc - U+030fe
2324 U+03260 - U+0327f
2325 U+0fb46 - U+0fbb1
2326 U+10450 - U+1049d
2327
2328 12. The -o option (show only the matching part of a line) for pcregrep was not
2329 compatible with GNU grep in that, if there was more than one match in a
2330 line, it showed only the first of them. It now behaves in the same way as
2331 GNU grep.
2332
2333 13. If the -o and -v options were combined for pcregrep, it printed a blank
2334 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2335 does the same. The return code can be used to tell if there were any
2336 non-matching lines.
2337
2338 14. Added --file-offsets and --line-offsets to pcregrep.
2339
2340 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2341 infinitely looping recursion. The bug was that positive lookaheads were not
2342 being skipped when checking for a possible empty match (negative lookaheads
2343 and both kinds of lookbehind were skipped).
2344
2345 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2346 inclusion of <windows.h> to before rather than after the definition of
2347 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2348
2349 17. Specifying a possessive quantifier with a specific limit for a Unicode
2350 character property caused pcre_compile() to compile bad code, which led at
2351 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2352 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2353 caused the error; without that there was no problem.
2354
2355 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2356
2357 19. Added --enable-pcretest-libreadline.
2358
2359 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2360 RE::GlobalReplace(). As a result, the number of replacements returned was
2361 double what it should be. I removed one of the increments, but Craig sent a
2362 later patch that removed the other one (the right fix) and added unit tests
2363 that check the return values (which was not done before).
2364
2365 21. Several CMake things:
2366
2367 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2368 the names libpcre and libpcreposix, not just pcre and pcreposix.
2369
2370 (2) The above change means that pcretest and pcregrep are now correctly
2371 linked with the newly-built libraries, not previously installed ones.
2372
2373 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2374
2375 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2376 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2377 UTF-8 newline character). The key issue is that the pattern starts .*;
2378 this means that the match must be either at the beginning, or after a
2379 newline. The bug was in the code for advancing after a failed match and
2380 checking that the new position followed a newline. It was not taking
2381 account of UTF-8 characters correctly.
2382
2383 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2384 character classes. PCRE was not treating the sequence [:...:] as a
2385 character class unless the ... were all letters. Perl, however, seems to
2386 allow any characters between [: and :], though of course it rejects as
2387 unknown any "names" that contain non-letters, because all the known class
2388 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2389 for example, whereas PCRE did not - it did not recognize a POSIX character
2390 class. This seemed a bit dangerous, so the code has been changed to be
2391 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2392 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2393 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2394 Perl does, and where it didn't before.
2395
2396 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2397 Windows environments %n is disabled by default.
2398
2399
2400 Version 7.4 21-Sep-07
2401 ---------------------
2402
2403 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2404 means that a class such as [\s] counted as "explicit reference to CR or
2405 LF". That isn't really right - the whole point of the change was to try to
2406 help when there was an actual mention of one of the two characters. So now
2407 the change happens only if \r or \n (or a literal CR or LF) character is
2408 encountered.
2409
2410 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2411 of both had grown to the point where there were only 3 bits left.
2412 Fortunately, there was spare space in the data structure, and so I have
2413 moved the internal flags into a new 16-bit field to free up more option
2414 bits.
2415
2416 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2417 but did not set the internal JCHANGED flag - either of these is enough to
2418 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2419 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2420 start sets both bits.
2421
2422 4. Added options (at build time, compile time, exec time) to change \R from
2423 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2424
2425 5. doc/pcresyntax.html was missing from the distribution.
2426
2427 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2428 compatibility, even though it is no longer used.
2429
2430 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2431 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2432 windows.h file is present (where different names are used). [This was
2433 reversed later after testing - see 16 below.]
2434
2435 8. Changed all #include <config.h> to #include "config.h". There were also
2436 some further <pcre.h> cases that I changed to "pcre.h".
2437
2438 9. When pcregrep was used with the --colour option, it missed the line ending
2439 sequence off the lines that it output.
2440
2441 10. It was pointed out to me that arrays of string pointers cause lots of
2442 relocations when a shared library is dynamically loaded. A technique of
2443 using a single long string with a table of offsets can drastically reduce
2444 these. I have refactored PCRE in four places to do this. The result is
2445 dramatic:
2446
2447 Originally: 290
2448 After changing UCP table: 187
2449 After changing error message table: 43
2450 After changing table of "verbs" 36
2451 After changing table of Posix names 22
2452
2453 Thanks to the folks working on Gregex for glib for this insight.
2454
2455 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2456 unicode-properties was also set.
2457
2458 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2459
2460 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2461 checked only for CRLF.
2462
2463 14. Added casts to pcretest.c to avoid compiler warnings.
2464
2465 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2466
2467 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2468 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2469 entirely. This removes changes made in 7 above.
2470
2471 17. The CMake files have been updated, and there is now more information about
2472 building with CMake in the NON-UNIX-USE document.
2473
2474
2475 Version 7.3 28-Aug-07
2476 ---------------------
2477
2478 1. In the rejigging of the build system that eventually resulted in 7.1, the
2479 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2480 brackets there is not right, since it causes compilers to look for an
2481 installed pcre.h, not the version that is in the source that is being
2482 compiled (which of course may be different). I have changed it back to:
2483
2484 #include "pcre.h"
2485
2486 I have a vague recollection that the change was concerned with compiling in
2487 different directories, but in the new build system, that is taken care of
2488 by the VPATH setting the Makefile.
2489
2490 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2491 when the subject happened to end in the byte 0x85 (e.g. if the last
2492 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2493 characters but of course it shouldn't be taken as a newline when it is part
2494 of another character. The bug was that, for an unlimited repeat of . in
2495 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2496 characters when looking for a newline.
2497
2498 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2499
2500 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2501 in debug output.
2502
2503 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2504 long printing in the pcrecpp unittest when running under MinGW.
2505
2506 6. ESC_K was left out of the EBCDIC table.
2507
2508 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2509 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2510 limit also applies to "virtual nesting" when a pattern is recursive, and in
2511 this case 1000 isn't so big. I have been able to remove this limit at the
2512 expense of backing off one optimization in certain circumstances. Normally,
2513 when pcre_exec() would call its internal match() function recursively and
2514 immediately return the result unconditionally, it uses a "tail recursion"
2515 feature to save stack. However, when a subpattern that can match an empty
2516 string has an unlimited repetition quantifier, it no longer makes this
2517 optimization. That gives it a stack frame in which to save the data for
2518 checking that an empty string has been matched. Previously this was taken
2519 from the 1000-entry workspace that had been reserved. So now there is no
2520 explicit limit, but more stack is used.
2521
2522 8. Applied Daniel's patches to solve problems with the import/export magic
2523 syntax that is required for Windows, and which was going wrong for the
2524 pcreposix and pcrecpp parts of the library. These were overlooked when this
2525 problem was solved for the main library.
2526
2527 9. There were some crude static tests to avoid integer overflow when computing
2528 the size of patterns that contain repeated groups with explicit upper
2529 limits. As the maximum quantifier is 65535, the maximum group length was
2530 set at 30,000 so that the product of these two numbers did not overflow a
2531 32-bit integer. However, it turns out that people want to use groups that
2532 are longer than 30,000 bytes (though not repeat them that many times).
2533 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2534 made it possible to implement the integer overflow checks in a much more
2535 dynamic way, which I have now done. The artificial limitation on group
2536 length has been removed - we now have only the limit on the total length of
2537 the compiled pattern, which depends on the LINK_SIZE setting.
2538
2539 10. Fixed a bug in the documentation for get/copy named substring when
2540 duplicate names are permitted. If none of the named substrings are set, the
2541 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2542 empty string.
2543
2544 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2545 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2546 because the ] is interpreted as the first data character and the
2547 terminating ] is not found. PCRE has been made compatible with Perl in this
2548 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2549 cause memory overwriting.
2550
2551 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2552 string has been matched (to stop an infinite loop). It was not recognizing
2553 a conditional subpattern that could match an empty string if that
2554 subpattern was within another subpattern. For example, it looped when
2555 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2556 condition was not nested. This bug has been fixed.
2557
2558 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2559 past the start of the subject in the presence of bytes with the top bit
2560 set, for example "\x8aBCD".
2561
2562 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2563 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2564
2565 14. Optimized (?!) to (*FAIL).
2566
2567 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2568 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2569 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2570 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2571 does: it's just the validity check that is more restrictive.
2572
2573 16. Inserted checks for integer overflows during escape sequence (backslash)
2574 processing, and also fixed erroneous offset values for syntax errors during
2575 backslash processing.
2576
2577 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2578 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2579
2580 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2581 caused an overrun.
2582
2583 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2584 something other than just ASCII characters) inside a group that had an
2585 unlimited repeat caused a loop at compile time (while checking to see
2586 whether the group could match an empty string).
2587
2588 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2589 [\P{Any}] did so. (Error in the code for printing property names.)
2590
2591 21. An orphan \E inside a character class could cause a crash.
2592
2593 22. A repeated capturing bracket such as (A)? could cause a wild memory
2594 reference during compilation.
2595
2596 23. There are several functions in pcre_compile() that scan along a compiled
2597 expression for various reasons (e.g. to see if it's fixed length for look
2598 behind). There were bugs in these functions when a repeated \p or \P was
2599 present in the pattern. These operators have additional parameters compared
2600 with \d, etc, and these were not being taken into account when moving along
2601 the compiled data. Specifically:
2602
2603 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2604 length.
2605
2606 (b) An item such as \pL+ within a repeated group could cause crashes or
2607 loops.
2608
2609 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2610 "reference to non-existent subpattern" error.
2611
2612 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2613
2614 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2615 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2616
2617 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2618
2619 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2620 character were causing crashes (broken optimization).
2621
2622 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2623 \p or \P) caused a compile-time loop.
2624
2625 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2626 break. For example, the unstudied pattern [\r\n]A does not match the string
2627 "\r\nA" because change 7.0/46 below moves the current point on by two
2628 characters after failing to match at the start. However, the pattern \nA
2629 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2630 the same is true. There doesn't seem any very clean way out of this, but
2631 what I have chosen to do makes the common cases work: PCRE now takes note
2632 of whether there can be an explicit match for \r or \n anywhere in the
2633 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2634 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2635 pattern has explicit CR or LF references.
2636
2637 29. Added (*CR) etc for changing newline setting at start of pattern.
2638
2639
2640 Version 7.2 19-Jun-07
2641 ---------------------
2642
2643 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2644 which is apparently normally available under Windows.
2645
2646 2. Re-jig the pcregrep tests with different newline settings in an attempt
2647 to make them independent of the local environment's newline setting.
2648
2649 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2650
2651 4. Some of the "internals" tests were previously cut out when the link size
2652 was not 2, because the output contained actual offsets. The recent new
2653 "Z" feature of pcretest means that these can be cut out, making the tests
2654 usable with all link sizes.
2655
2656 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2657 stack recursion. This gives a massive performance boost under BSD, but just
2658 a small improvement under Linux. However, it saves one field in the frame
2659 in all cases.
2660
2661 6. Added more features from the forthcoming Perl 5.10:
2662
2663 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2664 recursion call. It refers to the nth most recently opened parentheses.
2665
2666 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2667 to be opened parentheses.
2668
2669 (c) Conditions that refer to capturing parentheses can be specified
2670 relatively, for example, (?(-2)... or (?(+3)...
2671
2672 (d) \K resets the start of the current match so that everything before
2673 is not part of it.
2674
2675 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2676
2677 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2678 reference syntax.
2679
2680 (g) (?| introduces a group in which the numbering of parentheses in each
2681 alternative starts with the same number.
2682
2683 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2684
2685 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2686 PCRE_INFO_JCHANGED.
2687
2688 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2689 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2690 for detecting groups that can match an empty string.
2691
2692 9. A pattern with a very large number of alternatives (more than several
2693 hundred) was running out of internal workspace during the pre-compile
2694 phase, where pcre_compile() figures out how much memory will be needed. A
2695 bit of new cunning has reduced the workspace needed for groups with
2696 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2697 workspace instead of running out of the 4096 that are available.
2698
2699 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2700
2701 11. Applied patch from Google to remove an optimization that didn't quite work.
2702 The report of the bug said:
2703
2704 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2705 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2706 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2707
2708 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2709 it matched the wrong number of bytes.
2710
2711
2712 Version 7.1 24-Apr-07
2713 ---------------------
2714
2715 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2716 that is more "standard", making use of automake and other Autotools. There
2717 is some re-arrangement of the files and adjustment of comments consequent
2718 on this.
2719
2720 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2721 for recursive directory scanning broke on some systems because the files
2722 are not scanned in any specific order and on different systems the order
2723 was different. A call to "sort" has been inserted into RunGrepTest for the
2724 approprate test as a short-term fix. In the longer term there may be an
2725 alternative.
2726
2727 3. I had an email from Eric Raymond about problems translating some of PCRE's
2728 man pages to HTML (despite the fact that I distribute HTML pages, some
2729 people do their own conversions for various reasons). The problems
2730 concerned the use of low-level troff macros .br and .in. I have therefore
2731 removed all such uses from the man pages (some were redundant, some could
2732 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2733 HTML has been updated to handle .nf/.fi and to complain if it encounters
2734 .br or .in.
2735
2736 4. Updated comments in configure.ac that get placed in config.h.in and also
2737 arranged for config.h to be included in the distribution, with the name
2738 config.h.generic, for the benefit of those who have to compile without
2739 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2740
2741 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2742 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2743 makevp.bat for the current PCRE, using the additional files
2744 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2745
2746 6. A Windows user reported a minor discrepancy with test 2, which turned out
2747 to be caused by a trailing space on an input line that had got lost in his
2748 copy. The trailing space was an accident, so I've just removed it.
2749
2750 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2751 that is needed.
2752
2753 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2754 as "const" (a) because they are and (b) because it helps the PHP
2755 maintainers who have recently made a script to detect big data structures
2756 in the php code that should be moved to the .rodata section. I remembered
2757 to update Builducptable as well, so it won't revert if ucptable.h is ever
2758 re-created.
2759
2760 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2761 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2762 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2763 support is not required. This saves 1.5-2K of code, which is important in
2764 some applications.
2765
2766 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2767 so as not to refer to the tables, even though these functions will never be
2768 called when UTF-8 support is disabled. Otherwise there are problems with a
2769 shared library.
2770
2771 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2772
2773 (a) It was defining its arguments as char * instead of void *.
2774
2775 (b) It was assuming that all moves were upwards in memory; this was true
2776 a long time ago when I wrote it, but is no longer the case.
2777
2778 The emulated memove() is provided for those environments that have neither
2779 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2780 is clearly not the case, as these two bugs were recently reported.
2781
2782 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2783 and Detrail to create the HTML documentation, the .txt form of the man
2784 pages, and it removes trailing spaces from listed files. It also creates
2785 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2786 case, it wraps all the #defines with #ifndefs. This script should be run
2787 before "make dist".
2788
2789 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2790 with Unicode property support.
2791
2792 (a) For a maximizing quantifier, if the two different cases of the
2793 character were of different lengths in their UTF-8 codings (there are
2794 some cases like this - I found 11), and the matching function had to
2795 back up over a mixture of the two cases, it incorrectly assumed they
2796 were both the same length.
2797
2798 (b) When PCRE was configured to use the heap rather than the stack for
2799 recursion during matching, it was not correctly preserving the data for
2800 the other case of a UTF-8 character when checking ahead for a match
2801 while processing a minimizing repeat. If the check also involved
2802 matching a wide character, but failed, corruption could cause an
2803 erroneous result when trying to check for a repeat of the original
2804 character.
2805
2806 13. Some tidying changes to the testing mechanism:
2807
2808 (a) The RunTest script now detects the internal link size and whether there
2809 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2810 values substituted by "configure". (The RunGrepTest script already did
2811 this for UTF-8.) The configure.ac script no longer substitutes the
2812 relevant variables.
2813
2814 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2815 with length and offset values. This means that the output is different
2816 for different internal link sizes. Test 2 is skipped for link sizes
2817 other than 2 because of this, bypassing the problem. Unfortunately,
2818 there was also a test in test 3 (the locale tests) that used /B and
2819 failed for link sizes other than 2. Rather than cut the whole test out,
2820 I have added a new /Z option to pcretest that replaces the length and
2821 offset values with spaces. This is now used to make test 3 independent
2822 of link size. (Test 2 will be tidied up later.)
2823
2824 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2825 segmentation fault instead of returning the appropriate error message.
2826
2827 15. In multiline mode when the newline sequence was set to "any", the pattern
2828 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2829 This doesn't seem right; it now treats the CRLF combination as the line
2830 ending, and so does not match in that case. It's only a pattern such as ^$
2831 that would hit this one: something like ^ABC$ would have failed after \r
2832 and then tried again after \r\n.
2833
2834 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2835 in an attempt to make files that differ only in their line terminators
2836 compare equal. This works on Linux.
2837
2838 17. Under certain error circumstances pcregrep might try to free random memory
2839 as it exited. This is now fixed, thanks to valgrind.
2840
2841 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2842 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2843 was because its rules for how to advance for /g after matching an empty
2844 string at the end of a line did not allow for this case. They now check for
2845 it specially.
2846
2847 20. pcretest is supposed to handle patterns and data of any length, by
2848 extending its buffers when necessary. It was getting this wrong when the
2849 buffer for a data line had to be extended.
2850
2851 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2852 CRLF as a newline sequence.
2853
2854 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2855 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2856 I have nevertheless tidied it up.
2857
2858 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2859
2860 24. Added a man page for pcre-config.
2861
2862
2863 Version 7.0 19-Dec-06
2864 ---------------------
2865
2866 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2867 moving to gcc 4.1.1.
2868
2869 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2870 sys/time.h, which is documented as needed for this function. It doesn't
2871 seem to matter on Linux, but it showed up on some releases of OS X.
2872
2873 3. It seems that there are systems where bytes whose values are greater than
2874 127 match isprint() in the "C" locale. The "C" locale should be the
2875 default when a C program starts up. In most systems, only ASCII printing
2876 characters match isprint(). This difference caused the output from pcretest
2877 to vary, making some of the tests fail. I have changed pcretest so that:
2878
2879 (a) When it is outputting text in the compiled version of a pattern, bytes
2880 other than 32-126 are always shown as hex escapes.
2881
2882 (b) When it is outputting text that is a matched part of a subject string,
2883 it does the same, unless a different locale has been set for the match
2884 (using the /L modifier). In this case, it uses isprint() to decide.
2885
2886 4. Fixed a major bug that caused incorrect computation of the amount of memory
2887 required for a compiled pattern when options that changed within the
2888 pattern affected the logic of the preliminary scan that determines the
2889 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2890 that the computed length was too small. The symptoms of this bug were
2891 either the PCRE error "internal error: code overflow" from pcre_compile(),
2892 or a glibc crash with a message such as "pcretest: free(): invalid next
2893 size (fast)". Examples of patterns that provoked this bug (shown in
2894 pcretest format) are:
2895
2896 /(?-x: )/x
2897 /(?x)(?-x: \s*#\s*)/
2898 /((?i)[\x{c0}])/8
2899 /(?i:[\x{c0}])/8
2900
2901 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2902 is now done differently.
2903
2904 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2905 wrapper classes; (b) implement a new function in the C++ scanner that is
2906 more efficient than the old way of doing things because it avoids levels of
2907 recursion in the regex matching; (c) add a paragraph to the documentation
2908 for the FullMatch() function.
2909
2910 6. The escape sequence \n was being treated as whatever was defined as
2911 "newline". Not only was this contrary to the documentation, which states
2912 that \n is character 10 (hex 0A), but it also went horribly wrong when
2913 "newline" was defined as CRLF. This has been fixed.
2914
2915 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2916 was being set to -1 for the "end of line" case (supposedly a value that no
2917 character can have). Though this value is never used (the check for end of
2918 line is "zero bytes in current character"), it caused compiler complaints.
2919 I've changed it to 0xffffffff.
2920
2921 8. In pcre_version.c, the version string was being built by a sequence of
2922 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2923 string (as it is for production releases) called a macro with an empty
2924 argument. The C standard says the result of this is undefined. The gcc
2925 compiler treats it as an empty string (which was what was wanted) but it is
2926 reported that Visual C gives an error. The source has been hacked around to
2927 avoid this problem.
2928
2929 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2930 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2931 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2932 of them did).
2933
2934 10. Originally, pcretest opened its input and output without "b"; then I was
2935 told that "b" was needed in some environments, so it was added for release
2936 5.0 to both the input and output. (It makes no difference on Unix-like
2937 systems.) Later I was told that it is wrong for the input on Windows. I've
2938 now abstracted the modes into two macros, to make it easier to fiddle with
2939 them, and removed "b" from the input mode under Windows.
2940
2941 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2942
2943 12. Added -help and --help to pcretest as an official way of being reminded
2944 of the options.
2945
2946 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2947 and pcrecpp.cc because they annoy compilers at high warning levels.
2948
2949 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2950
2951 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2952 scripts are not C programs :-) and which was not noticed because it works
2953 on Linux.
2954
2955 16. pcretest is supposed to handle any length of pattern and data line (as one
2956 line or as a continued sequence of lines) by extending its input buffer if
2957 necessary. This feature was broken for very long pattern lines, leading to
2958 a string of junk being passed to pcre_compile() if the pattern was longer
2959 than about 50K.
2960
2961 17. I have done a major re-factoring of the way pcre_compile() computes the
2962 amount of memory needed for a compiled pattern. Previously, there was code
2963 that made a preliminary scan of the pattern in order to do this. That was
2964 OK when PCRE was new, but as the facilities have expanded, it has become
2965 harder and harder to keep it in step with the real compile phase, and there
2966 have been a number of bugs (see for example, 4 above). I have now found a
2967 cunning way of running the real compile function in a "fake" mode that
2968 enables it to compute how much memory it would need, while actually only
2969 ever using a few hundred bytes of working memory and without too many
2970 tests of the mode. This should make future maintenance and development
2971 easier. A side effect of this work is that the limit of 200 on the nesting
2972 depth of parentheses has been removed (though this was never a serious
2973 limitation, I suspect). However, there is a downside: pcre_compile() now
2974 runs more slowly than before (30% or more, depending on the pattern). I
2975 hope this isn't a big issue. There is no effect on runtime performance.
2976
2977 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2978 newline (only possible for the last line of a file) and it was a
2979 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2980
2981 19. Added additional timing features to pcretest. (1) The -tm option now times
2982 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2983 separate command line item, by a number that specifies the number of
2984 repeats to use when timing. The default is 50000; this gives better
2985 precision, but takes uncomfortably long for very large patterns.
2986
2987 20. Extended pcre_study() to be more clever in cases where a branch of a
2988 subpattern has no definite first character. For example, (a*|b*)[cd] would
2989 previously give no result from pcre_study(). Now it recognizes that the
2990 first character must be a, b, c, or d.
2991
2992 21. There was an incorrect error "recursive call could loop indefinitely" if
2993 a subpattern (or the entire pattern) that was being tested for matching an
2994 empty string contained only one non-empty item after a nested subpattern.
2995 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2996 incorrectly, because the \d was being skipped in the check.
2997
2998 22. The pcretest program now has a new pattern option /B and a command line
2999 option -b, which is equivalent to adding /B to every pattern. This causes
3000 it to show the compiled bytecode, without the additional information that
3001 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3002 is the same as /B/I).
3003
3004 23. A new optimization is now able automatically to treat some sequences such
3005 as a*b as a*+b. More specifically, if something simple (such as a character
3006 or a simple class like \d) has an unlimited quantifier, and is followed by
3007 something that cannot possibly match the quantified thing, the quantifier
3008 is automatically "possessified".
3009
3010 24. A recursive reference to a subpattern whose number was greater than 39
3011 went wrong under certain circumstances in UTF-8 mode. This bug could also
3012 have affected the operation of pcre_study().
3013
3014 25. Realized that a little bit of performance could be had by replacing
3015 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3016
3017 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3018
3019 27. Possessive quantifiers such as a++ were previously implemented by turning
3020 them into atomic groups such as ($>a+). Now they have their own opcodes,
3021 which improves performance. This includes the automatically created ones
3022 from 23 above.
3023
3024 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3025 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3026 the first matched character to be a colon. This applied both to named and
3027 numbered groups.
3028
3029 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3030
3031 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3032 building PCRE on Windows easier, so I have included it in the distribution.
3033
3034 31. There is now a check in pcretest against a ridiculously large number being
3035 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3036 loop, the loop is abandoned.
3037
3038 32. Forward references to subpatterns in conditions such as (?(2)...) where
3039 subpattern 2 is defined later cause pcre_compile() to search forwards in
3040 the pattern for the relevant set of parentheses. This search went wrong
3041 when there were unescaped parentheses in a character class, parentheses
3042 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3043
3044 33. "Subroutine" calls and backreferences were previously restricted to
3045 referencing subpatterns earlier in the regex. This restriction has now
3046 been removed.
3047
3048 34. Added a number of extra features that are going to be in Perl 5.10. On the
3049 whole, these are just syntactic alternatives for features that PCRE had
3050 previously implemented using the Python syntax or my own invention. The
3051 other formats are all retained for compatibility.
3052
3053 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3054 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3055 also .NET compatible.
3056
3057 (b) A recursion or subroutine call to a named group can now be defined as
3058 (?&name) as well as (?P>name).
3059
3060 (c) A backreference to a named group can now be defined as \k<name> or
3061 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3062 5.10, are also .NET compatible.
3063
3064 (d) A conditional reference to a named group can now use the syntax
3065 (?(<name>) or (?('name') as well as (?(name).
3066
3067 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3068 groups (named and numbered) that are never evaluated inline, but can be
3069 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3070 is always false. There may be only one alternative in such a group.
3071
3072 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3073 as the simple (?(R). The condition is true only if the most recent
3074 recursion is that of the given number or name. It does not search out
3075 through the entire recursion stack.
3076
3077 (g) The escape \gN or \g{N} has been added, where N is a positive or
3078 negative number, specifying an absolute or relative reference.
3079
3080 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3081 some "unreachable code" warnings.
3082
3083 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3084 things, this adds five new scripts.
3085
3086 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3087 There were also incompatibilities regarding the handling of \Q..\E inside
3088 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3089 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3090
3091 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3092 matches an empty string, and forcibly breaks the loop. There were bugs in
3093 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3094 against aaaa the result was just "a" rather than "aaaa", for example. Two
3095 separate and independent bugs (that affected different cases) have been
3096 fixed.
3097
3098 39. Refactored the code to abolish the use of different opcodes for small
3099 capturing bracket numbers. This is a tidy that I avoided doing when I
3100 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3101 The new approach is not only tidier, it makes it possible to reduce the
3102 memory needed to fix the previous bug (38).
3103
3104 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3105 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3106 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3107 mode.
3108
3109 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3110 report.
3111
3112 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3113 copy construction and assignment in the C++ wrapper.
3114
3115 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3116 couple of bugs that could have given wrong results in the "--newline=crlf"
3117 case.
3118
3119 44. Added a number of casts and did some reorganization of signed/unsigned int
3120 variables following suggestions from Dair Grant. Also renamed the variable
3121 "this" as "item" because it is a C++ keyword.
3122
3123 45. Arranged for dftables to add
3124
3125 #include "pcre_internal.h"
3126
3127 to pcre_chartables.c because without it, gcc 4.x may remove the array
3128 definition from the final binary if PCRE is built into a static library and
3129 dead code stripping is activated.
3130
3131 46. For an unanchored pattern, if a match attempt fails at the start of a
3132 newline sequence, and the newline setting is CRLF or ANY, and the next two
3133 characters are CRLF, advance by two characters instead of one.
3134
3135
3136 Version 6.7 04-Jul-06
3137 ---------------------
3138
3139 1. In order to handle tests when input lines are enormously long, pcretest has
3140 been re-factored so that it automatically extends its buffers when
3141 necessary. The code is crude, but this _is_ just a test program. The
3142 default size has been increased from 32K to 50K.
3143
3144 2. The code in pcre_study() was using the value of the re argument before
3145 testing it for NULL. (Of course, in any sensible call of the function, it
3146 won't be NULL.)
3147
3148 3. The memmove() emulation function in pcre_internal.h, which is used on
3149 systems that lack both memmove() and bcopy() - that is, hardly ever -
3150 was missing a "static" storage class specifier.
3151
3152 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3153 containing an extended class (one that cannot be represented by a bitmap
3154 because it contains high-valued characters or Unicode property items, e.g.
3155 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3156 pattern, but PCRE should not loop if you do not (it no longer does).
3157 [Detail: two cases were found: (a) a repeated subpattern containing an
3158 extended class; (b) a recursive reference to a subpattern that followed a
3159 previous extended class. It wasn't skipping over the extended class
3160 correctly when UTF-8 mode was not set.]
3161
3162 5. A negated single-character class was not being recognized as fixed-length
3163 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3164 compile error "lookbehind assertion is not fixed length".
3165
3166 6. The RunPerlTest auxiliary script was showing an unexpected difference
3167 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3168 write a Perl script that can interpret lines of an input file either as
3169 byte characters or as UTF-8, which is what "perltest" was being required to
3170 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3171 can't do is switch easily at run time between having the "use utf8;" pragma
3172 or not. In the end, I fudged it by using the RunPerlTest script to insert
3173 "use utf8;" explicitly for the UTF-8 tests.
3174
3175 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3176 the end of the subject string, contrary to the documentation and to what
3177 Perl does. This was true of both matching functions. Now it matches only at
3178 the start of the subject and immediately after *internal* newlines.
3179
3180 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3181 a pointer to an int instead of a pointer to an unsigned long int. This
3182 caused problems on 64-bit systems.
3183
3184 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3185 instance of the 'standard' template library not being so standard".
3186
3187 10. There was no check on the number of named subpatterns nor the maximum
3188 length of a subpattern name. The product of these values is used to compute
3189 the size of the memory block for a compiled pattern. By supplying a very
3190 long subpattern name and a large number of named subpatterns, the size
3191 computation could be caused to overflow. This is now prevented by limiting
3192 the length of names to 32 characters, and the number of named subpatterns
3193 to 10,000.
3194
3195 11. Subpatterns that are repeated with specific counts have to be replicated in
3196 the compiled pattern. The size of memory for this was computed from the
3197 length of the subpattern and the repeat count. The latter is limited to
3198 65535, but there was no limit on the former, meaning that integer overflow
3199 could in principle occur. The compiled length of a repeated subpattern is
3200 now limited to 30,000 bytes in order to prevent this.
3201
3202 12. Added the optional facility to have named substrings with the same name.
3203
3204 13. Added the ability to use a named substring as a condition, using the
3205 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3206 are numbers (not recommended). Forward references are permitted.
3207
3208 14. Added forward references in named backreferences (if you see what I mean).
3209
3210 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3211 pattern could run off the end of the subject. For example, the pattern
3212 "(?s)(.{1,5})"8 did this with the subject "ab".
3213
3214 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3215 PCRE_CASELESS was set when matching characters that were quantified with ?
3216 or *.
3217
3218 17. A character class other than a single negated character that had a minimum
3219 but no maximum quantifier - for example [ab]{6,} - was not handled
3220 correctly by pce_dfa_exec(). It would match only one character.
3221
3222 18. A valid (though odd) pattern that looked like a POSIX character
3223 class but used an invalid character after [ (for example [[,abc,]]) caused
3224 pcre_compile() to give the error "Failed: internal error: code overflow" or
3225 in some cases to crash with a glibc free() error. This could even happen if
3226 the pattern terminated after [[ but there just happened to be a sequence of
3227 letters, a binary zero, and a closing ] in the memory that followed.
3228
3229 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3230 over the years. Originally (before any Unicode support), just the bottom 8
3231 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3232 output from "man perlunicode" includes this:
3233
3234 The regular expression compiler produces polymorphic opcodes. That
3235 is, the pattern adapts to the data and automatically switches to
3236 the Unicode character scheme when presented with Unicode data--or
3237 instead uses a traditional byte scheme when presented with byte
3238 data.
3239
3240 Sadly, a wide octal escape does not cause a switch, and in a string with
3241 no other multibyte characters, these octal escapes are treated as before.
3242 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3243 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3244 Unicode string.
3245
3246 I have not perpetrated such confusion in PCRE. Up till now, it took just
3247 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3248 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3249 translate to the appropriate multibyte character.
3250
3251 29. Applied some refactoring to reduce the number of warnings from Microsoft
3252 and Borland compilers. This has included removing the fudge introduced
3253 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3254 a warning about an unused variable.
3255
3256 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3257 characters since release 4.0, because Perl (from release 5.004) does not.
3258 [Or at least, is documented not to: some releases seem to be in conflict
3259 with the documentation.] However, when a pattern was studied with
3260 pcre_study() and all its branches started with \s, PCRE still included VT
3261 as a possible starting character. Of course, this did no harm; it just
3262 caused an unnecessary match attempt.
3263
3264 22. Removed a now-redundant internal flag bit that recorded the fact that case
3265 dependency changed within the pattern. This was once needed for "required
3266 byte" processing, but is no longer used. This recovers a now-scarce options
3267 bit. Also moved the least significant internal flag bit to the most-
3268 significant bit of the word, which was not previously used (hangover from
3269 the days when it was an int rather than a uint) to free up another bit for
3270 the future.
3271
3272 23. Added support for CRLF line endings as well as CR and LF. As well as the
3273 default being selectable at build time, it can now be changed at runtime
3274 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3275 specify that it is scanning data with non-default line endings.
3276
3277 24. Changed the definition of CXXLINK to make it agree with the definition of
3278 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3279
3280 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3281 recursions. This makes a big different to stack usage for some patterns.
3282
3283 26. If a subpattern containing a named recursion or subroutine reference such
3284 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3285 the space required for the compiled pattern went wrong and gave too small a
3286 value. Depending on the environment, this could lead to "Failed: internal
3287 error: code overflow at offset 49" or "glibc detected double free or
3288 corruption" errors.
3289
3290 27. Applied patches from Google (a) to support the new newline modes and (b) to
3291 advance over multibyte UTF-8 characters in GlobalReplace.
3292
3293 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3294 difference for some implementation of PCRE in some Windows version.
3295
3296 29. Added some extra testing facilities to pcretest:
3297
3298 \q<number> in a data line sets the "match limit" value
3299 \Q<number> in a data line sets the "match recursion limt" value
3300 -S <number> sets the stack size, where <number> is in megabytes
3301
3302 The -S option isn't available for Windows.
3303
3304
3305 Version 6.6 06-Feb-06
3306 ---------------------
3307
3308 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3309 in pcreposix.h. I have copied the definition from pcre.h.
3310
3311 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3312 because pcre.h is no longer a built file.
3313
3314 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3315 not normally included in the compiled code.
3316
3317
3318 Version 6.5 01-Feb-06
3319 ---------------------
3320
3321 1. When using the partial match feature with pcre_dfa_exec(), it was not
3322 anchoring the second and subsequent partial matches at the new starting
3323 point. This could lead to incorrect results. For example, with the pattern
3324 /1234/, partially matching against "123" and then "a4" gave a match.
3325
3326 2. Changes to pcregrep:
3327
3328 (a) All non-match returns from pcre_exec() were being treated as failures
3329 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3330 error message is output. Some extra information is given for the
3331 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3332 probably the only errors that are likely to be caused by users (by
3333 specifying a regex that has nested indefinite repeats, for instance).
3334 If there are more than 20 of these errors, pcregrep is abandoned.
3335
3336 (b) A binary zero was treated as data while matching, but terminated the
3337 output line if it was written out. This has been fixed: binary zeroes
3338 are now no different to any other data bytes.
3339
3340 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3341 used to set a locale for matching. The --locale=xxxx long option has
3342 been added (no short equivalent) to specify a locale explicitly on the
3343 pcregrep command, overriding the environment variables.
3344
3345 (d) When -B was used with -n, some line numbers in the output were one less
3346 than they should have been.
3347
3348 (e) Added the -o (--only-matching) option.
3349
3350 (f) If -A or -C was used with -c (count only), some lines of context were
3351 accidentally printed for the final match.
3352
3353 (g) Added the -H (--with-filename) option.
3354
3355 (h) The combination of options -rh failed to suppress file names for files
3356 that were found from directory arguments.
3357
3358 (i) Added the -D (--devices) and -d (--directories) options.
3359
3360 (j) Added the -F (--fixed-strings) option.
3361
3362 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3363
3364 (l) Added the --colo(u)r option.
3365
3366 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3367 is not present by default.
3368
3369 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3370 items such as (?R) or (?1), when the recursion could match a number of
3371 alternatives. If it matched one of the alternatives, but subsequently,
3372 outside the recursion, there was a failure, the code tried to back up into
3373 the recursion. However, because of the way PCRE is implemented, this is not
3374 possible, and the result was an incorrect result from the match.
3375
3376 In order to prevent this happening, the specification of recursion has
3377 been changed so that all such subpatterns are automatically treated as
3378 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3379
3380 4. I had overlooked the fact that, in some locales, there are characters for
3381 which isalpha() is true but neither isupper() nor islower() are true. In
3382 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3383 and ordfeminine) are like this. This affected the treatment of \w and \W
3384 when they appeared in character classes, but not when they appeared outside
3385 a character class. The bit map for "word" characters is now created
3386 separately from the results of isalnum() instead of just taking it from the
3387 upper, lower, and digit maps. (Plus the underscore character, of course.)
3388
3389 5. The above bug also affected the handling of POSIX character classes such as
3390 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3391 permanent tables. Instead, the bit maps for such a class were previously
3392 created as the appropriate unions of the upper, lower, and digit bitmaps.
3393 Now they are created by subtraction from the [[:word:]] class, which has
3394 its own bitmap.
3395
3396 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3397 It is created by subtracting the vertical space characters (\x09, \x0a,
3398 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3399 subtraction was done in the overall bitmap for a character class, meaning
3400 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3401 be recognized. This bug has been fixed.
3402
3403 7. Patches from the folks at Google:
3404
3405 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3406 real life, but is still worth protecting against".
3407
3408 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3409 regular expressions".
3410
3411 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3412 have it.
3413
3414 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3415 "configure" and the latter not, in order to fix a problem somebody had
3416 with compiling the Arg class on HP-UX.
3417
3418 (e) Improve the error-handling of the C++ wrapper a little bit.
3419
3420 (f) New tests for checking recursion limiting.
3421
3422 8. The pcre_memmove() function, which is used only if the environment does not
3423 have a standard memmove() function (and is therefore rarely compiled),
3424 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3425 returning a result (though PCRE never actually uses the result).
3426
3427 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3428 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3429 returned instead of calling malloc() with an overflowing number that would
3430 most likely cause subsequent chaos.
3431
3432 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3433
3434 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3435 with this option is matched, the nmatch and pmatch options of regexec() are
3436 ignored.
3437
3438 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3439 provided in case anyone wants to the the POSIX interface with UTF-8
3440 strings.
3441
3442 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3443 C++ linking (needed for some HP-UX environments).
3444
3445 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3446 (unused parameter) and in the pcre_printint() function (omitted "default"
3447 switch label when the default is to do nothing).
3448
3449 15. Added some code to make it possible, when PCRE is compiled as a C++
3450 library, to replace subject pointers for pcre_exec() with a smart pointer
3451 class, thus making it possible to process discontinuous strings.
3452
3453 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3454 much the same function. They were added by different people who were trying
3455 to make PCRE easy to compile on non-Unix systems. It has been suggested
3456 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3457 for compiling on Windows systems. I have therefore replaced it with
3458 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3459 defaults to "extern" for C or "extern C" for C++, which works fine on
3460 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3461 SCOPE with something explicit in config.h. In addition:
3462
3463 (a) pcreposix.h still had just "extern" instead of either of these macros;
3464 I have replaced it with PCRE_DATA_SCOPE.
3465
3466 (b) Functions such as _pcre_xclass(), which are internal to the library,
3467 but external in the C sense, all had PCRE_EXPORT in their definitions.
3468 This is apparently wrong for the Windows case, so I have removed it.
3469 (It makes no difference on Unix-like systems.)
3470
3471 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3472 of recursive calls to match(). This is different to MATCH_LIMIT because
3473 that limits the total number of calls to match(), not all of which increase
3474 the depth of recursion. Limiting the recursion depth limits the amount of
3475 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3476 when PCRE is compiled, and changed at run time. A patch from Google adds
3477 this functionality to the C++ interface.
3478
3479 18. Changes to the handling of Unicode character properties:
3480
3481 (a) Updated the table to Unicode 4.1.0.
3482
3483 (b) Recognize characters that are not in the table as "Cn" (undefined).
3484
3485 (c) I revised the way the table is implemented to a much improved format
3486 which includes recognition of ranges. It now supports the ranges that
3487 are defined in UnicodeData.txt, and it also amalgamates other
3488 characters into ranges. This has reduced the number of entries in the
3489 table from around 16,000 to around 3,000, thus reducing its size
3490 considerably. I realized I did not need to use a tree structure after
3491 all - a binary chop search is just as efficient. Having reduced the
3492 number of entries, I extended their size from 6 bytes to 8 bytes to
3493 allow for more data.
3494
3495 (d) Added support for Unicode script names via properties such as \p{Han}.
3496
3497 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3498 matching that character.
3499
3500 20. When matching a repeated Unicode property with a minimum greater than zero,
3501 (for example \pL{2,}), PCRE could look past the end of the subject if it
3502 reached it while seeking the minimum number of characters. This could
3503 happen only if some of the characters were more than one byte long, because
3504 there is a check for at least the minimum number of bytes.
3505
3506 21. Refactored the implementation of \p and \P so as to be more general, to
3507 allow for more different types of property in future. This has changed the
3508 compiled form incompatibly. Anybody with saved compiled patterns that use
3509 \p or \P will have to recompile them.
3510
3511 22. Added "Any" and "L&" to the supported property types.
3512
3513 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3514 but give a compile time error if the value is greater than 0xff.
3515
3516 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3517 accidentally not being installed or uninstalled.
3518
3519 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3520 made were to insert the current release number. This seemed silly, because
3521 it made things harder for people building PCRE on systems that don't run
3522 "configure". I have turned pcre.h into a distributed file, no longer built
3523 by "configure", with the version identification directly included. There is
3524 no longer a pcre.h.in file.
3525
3526 However, this change necessitated a change to the pcre-config script as
3527 well. It is built from pcre-config.in, and one of the substitutions was the
3528 release number. I have updated configure.ac so that ./configure now finds
3529 the release number by grepping pcre.h.
3530
3531 26. Added the ability to run the tests under valgrind.
3532
3533
3534 Version 6.4 05-Sep-05
3535 ---------------------
3536
3537 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3538 "--" to be printed when multiple files were scanned, even when none of the
3539 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3540 consider it to be a bug, and have restored the previous behaviour.
3541
3542 2. A couple of code tidies to get rid of compiler warnings.
3543
3544 3. The pcretest program used to cheat by referring to symbols in the library
3545 whose names begin with _pcre_. These are internal symbols that are not
3546 really supposed to be visible externally, and in some environments it is
3547 possible to suppress them. The cheating is now confined to including
3548 certain files from the library's source, which is a bit cleaner.
3549
3550 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3551 file's purpose clearer.
3552
3553 5. Reorganized pcre_ucp_findchar().
3554
3555
3556 Version 6.3 15-Aug-05
3557 ---------------------
3558
3559 1. The file libpcre.pc.in did not have general read permission in the tarball.
3560
3561 2. There were some problems when building without C++ support:
3562
3563 (a) If C++ support was not built, "make install" and "make test" still
3564 tried to test it.
3565
3566 (b) There were problems when the value of CXX was explicitly set. Some
3567 changes have been made to try to fix these, and ...
3568
3569 (c) --disable-cpp can now be used to explicitly disable C++ support.
3570
3571 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3572 backslash in a target when C++ was disabled. This confuses some
3573 versions of "make", apparently. Using an intermediate variable solves
3574 this. (Same for CPP_LOBJ.)
3575
3576 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3577 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3578 necessary on certain architectures.
3579
3580 4. Added a setting of -export-symbols-regex to the link command to remove
3581 those symbols that are exported in the C sense, but actually are local
3582 within the library, and not documented. Their names all begin with
3583 "_pcre_". This is not a perfect job, because (a) we have to except some
3584 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3585 available (and never for static libraries). I have made a note to try to
3586 find a way round (a) in the future.
3587
3588
3589 Version 6.2 01-Aug-05
3590 ---------------------
3591
3592 1. There was no test for integer overflow of quantifier values. A construction
3593 such as {1111111111111111} would give undefined results. What is worse, if
3594 a minimum quantifier for a parenthesized subpattern overflowed and became
3595 negative, the calculation of the memory size went wrong. This could have
3596 led to memory overwriting.
3597
3598 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3599
3600 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3601 operating environments where this matters.
3602
3603 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3604 PCRE options from within the C++ wrapper.
3605
3606 5. Named capturing subpatterns were not being correctly counted when a pattern
3607 was compiled. This caused two problems: (a) If there were more than 100
3608 such subpatterns, the calculation of the memory needed for the whole
3609 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3610 back references of the form \12, where the number was greater than 9, were
3611 not recognized as back references, even though there were sufficient
3612 previous subpatterns.
3613
3614 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3615 versions of gcc, e.g. 2.95.4.
3616
3617
3618 Version 6.1 21-Jun-05
3619 ---------------------
3620
3621 1. There was one reference to the variable "posix" in pcretest.c that was not
3622 surrounded by "#if !defined NOPOSIX".
3623
3624 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3625 the cross-check on the old pcre_info() function, for the benefit of the
3626 cut-down version of PCRE that is currently imported into Exim.
3627
3628 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3629 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3630 patterns that start (?i) but I don't think that matters. The use of (?i) is
3631 just an example; this all applies to the other options as well.
3632
3633 4. Since libtool seems to echo the compile commands it is issuing, the output
3634 from "make" can be reduced a bit by putting "@" in front of each libtool
3635 compile command.
3636
3637 5. Patch from the folks at Google for configure.in to be a bit more thorough
3638 in checking for a suitable C++ installation before trying to compile the
3639 C++ stuff. This should fix a reported problem when a compiler was present,
3640 but no suitable headers.
3641
3642 6. The man pages all had just "PCRE" as their title. I have changed them to
3643 be the relevant file name. I have also arranged that these names are
3644 retained in the file doc/pcre.txt, which is a concatenation in text format
3645 of all the man pages except the little individual ones for each function.
3646
3647 7. The NON-UNIX-USE file had not been updated for the different set of source
3648 files that come with release 6. I also added a few comments about the C++
3649 wrapper.
3650
3651
3652 Version 6.0 07-Jun-05
3653 ---------------------
3654
3655 1. Some minor internal re-organization to help with my DFA experiments.
3656
3657 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3658 didn't matter for the library itself when fully configured, but did matter
3659 when compiling without UCP support, or within Exim, where the ucp files are
3660 not imported.
3661
3662 3. Refactoring of the library code to split up the various functions into
3663 different source modules. The addition of the new DFA matching code (see
3664 below) to a single monolithic source would have made it really too
3665 unwieldy, quite apart from causing all the code to be include in a
3666 statically linked application, when only some functions are used. This is
3667 relevant even without the DFA addition now that patterns can be compiled in
3668 one application and matched in another.
3669
3670 The downside of splitting up is that there have to be some external
3671 functions and data tables that are used internally in different modules of
3672 the library but which are not part of the API. These have all had their
3673 names changed to start with "_pcre_" so that they are unlikely to clash
3674 with other external names.
3675
3676 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3677 a different (DFA) algorithm. Although it is slower than the original
3678 function, it does have some advantages for certain types of matching
3679 problem.
3680
3681 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3682 including restarting after a partial match.
3683
3684 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3685 defined when compiling for Windows was sent to me. I have put it into the
3686 code, though I have no means of testing or verifying it.
3687
3688 7. Added the pcre_refcount() auxiliary function.
3689
3690 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3691 match before or at the first newline in the subject string. In pcretest,
3692 the /f option on a pattern can be used to set this.
3693
3694 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3695 would behave wrongly. This has been present in PCRE since release 4.0.
3696
3697 10. A number of changes to the pcregrep command:
3698
3699 (a) Refactored how -x works; insert ^(...)$ instead of setting
3700 PCRE_ANCHORED and checking the length, in preparation for adding
3701 something similar for -w.
3702
3703 (b) Added the -w (match as a word) option.
3704
3705 (c) Refactored the way lines are read and buffered so as to have more
3706 than one at a time available.
3707
3708 (d) Implemented a pcregrep test script.
3709
3710 (e) Added the -M (multiline match) option. This allows patterns to match
3711 over several lines of the subject. The buffering ensures that at least
3712 8K, or the rest of the document (whichever is the shorter) is available
3713 for matching (and similarly the previous 8K for lookbehind assertions).
3714
3715 (f) Changed the --help output so that it now says
3716
3717 -w, --word-regex(p)
3718
3719 instead of two lines, one with "regex" and the other with "regexp"
3720 because that confused at least one person since the short forms are the
3721 same. (This required a bit of code, as the output is generated
3722 automatically from a table. It wasn't just a text change.)
3723
3724 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3725 option but starts with a hyphen. Could be a pattern or a path name
3726 starting with a hyphen, for instance.
3727
3728 (h) "-" can be given as a file name to represent stdin.
3729
3730 (i) When file names are being printed, "(standard input)" is used for
3731 the standard input, for compatibility with GNU grep. Previously
3732 "<stdin>" was used.
3733
3734 (j) The option --label=xxx can be used to supply a name to be used for
3735 stdin when file names are being printed. There is no short form.
3736
3737 (k) Re-factored the options decoding logic because we are going to add
3738 two more options that take data. Such options can now be given in four
3739 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3740
3741 (l) Added the -A, -B, and -C options for requesting that lines of context
3742 around matches be printed.
3743
3744 (m) Added the -L option to print the names of files that do not contain
3745 any matching lines, that is, the complement of -l.
3746
3747 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3748 continue to scan other files.
3749
3750 (o) The -s option was incorrectly implemented. For compatibility with other
3751 greps, it now suppresses the error message for a non-existent or non-
3752 accessible file (but not the return code). There is a new option called
3753 -q that suppresses the output of matching lines, which was what -s was
3754 previously doing.
3755
3756 (p) Added --include and --exclude options to specify files for inclusion
3757 and exclusion when recursing.
3758
3759 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3760 Hopefully, it now does.
3761
3762 12. Missing cast in pcre_study().
3763
3764 13. Added an "uninstall" target to the makefile.
3765
3766 14. Replaced "extern" in the function prototypes in Makefile.in with
3767 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3768 world, but is set differently for Windows.
3769
3770 15. Added a second compiling function called pcre_compile2(). The only
3771 difference is that it has an extra argument, which is a pointer to an
3772 integer error code. When there is a compile-time failure, this is set
3773 non-zero, in addition to the error test pointer being set to point to an
3774 error message. The new argument may be NULL if no error number is required
3775 (but then you may as well call pcre_compile(), which is now just a
3776 wrapper). This facility is provided because some applications need a
3777 numeric error indication, but it has also enabled me to tidy up the way
3778 compile-time errors are handled in the POSIX wrapper.
3779
3780 16. Added VPATH=.libs to the makefile; this should help when building with one
3781 prefix path and installing with another. (Or so I'm told by someone who
3782 knows more about this stuff than I do.)
3783
3784 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3785 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3786 match everything, including newlines. This is not POSIX-compatible, but
3787 somebody wanted the feature. From pcretest it can be activated by using
3788 both the P and the s flags.
3789
3790 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3791
3792 19. libpcre.pc was being incorrectly installed as executable.
3793
3794 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3795 it now also looks for '\r' so that it will work unmodified on Windows.
3796
3797 21. Added Google's contributed C++ wrapper to the distribution.
3798
3799 22. Added some untidy missing memory free() calls in pcretest, to keep
3800 Electric Fence happy when testing.
3801
3802
3803
3804 Version 5.0 13-Sep-04
3805 ---------------------
3806
3807 1. Internal change: literal characters are no longer packed up into items
3808 containing multiple characters in a single byte-string. Each character
3809 is now matched using a separate opcode. However, there may be more than one
3810 byte in the character in UTF-8 mode.
3811
3812 2. The pcre_callout_block structure has two new fields: pattern_position and
3813 next_item_length. These contain the offset in the pattern to the next match
3814 item, and its length, respectively.
3815
3816 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3817 insertion of callouts before each pattern item. Added the /C option to
3818 pcretest to make use of this.
3819
3820 4. On the advice of a Windows user, the lines
3821
3822 #if defined(_WIN32) || defined(WIN32)
3823 _setmode( _fileno( stdout ), 0x8000 );
3824 #endif /* defined(_WIN32) || defined(WIN32) */
3825
3826 have been added to the source of pcretest. This apparently does useful
3827 magic in relation to line terminators.
3828
3829 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3830 for the benefit of those environments where the "b" makes a difference.
3831
3832 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3833 to know about it. I have put a hack into configure.in that adds in code
3834 to set GCC=yes if CC=icc. This seems to end up at a point in the
3835 generated configure script that is early enough to affect the setting of
3836 compiler options, which is what is needed, but I have no means of testing
3837 whether it really works. (The user who reported this had patched the
3838 generated configure script, which of course I cannot do.)
3839
3840 LATER: After change 22 below (new libtool files), the configure script
3841 seems to know about icc (and also ecc). Therefore, I have commented out
3842 this hack in configure.in.
3843
3844 7. Added support for pkg-config (2 patches were sent in).
3845
3846 8. Negated POSIX character classes that used a combination of internal tables
3847 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3848 [[:^ascii]]. Typically, they would match almost any characters. The other
3849 POSIX classes were not broken in this way.
3850
3851 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3852 to find the match, as PCRE was deluded into thinking that the match had to
3853 start at the start point or following a newline. The same bug applied to
3854 patterns with negative forward assertions or any backward assertions
3855 preceding ".*" at the start, unless the pattern required a fixed first
3856 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3857
3858 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3859 starting at the last subject character, bytes beyond the end of the subject
3860 string were read.
3861
3862 11. Renamed the variable "class" as "classbits" to make life easier for C++
3863 users. (Previously there was a macro definition, but it apparently wasn't
3864 enough.)
3865
3866 12. Added the new field "tables" to the extra data so that tables can be passed
3867 in at exec time, or the internal tables can be re-selected. This allows
3868 a compiled regex to be saved and re-used at a later time by a different
3869 program that might have everything at different addresses.
3870
3871 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3872 -R library as well as a -L library.
3873
3874 14. The debugging options of pcretest (-d on the command line or D on a
3875 pattern) showed incorrect output for anything following an extended class
3876 that contained multibyte characters and which was followed by a quantifier.
3877
3878 15. Added optional support for general category Unicode character properties
3879 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3880 support. It adds about 90K to the size of the library. The meanings of the
3881 inbuilt class escapes such as \d and \s have NOT been changed.
3882
3883 16. Updated pcredemo.c to include calls to free() to release the memory for the
3884 compiled pattern.
3885
3886 17. The generated file chartables.c was being created in the source directory
3887 instead of in the building directory. This caused the build to fail if the
3888 source directory was different from the building directory, and was
3889 read-only.
3890
3891 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3892 file. No doubt somebody will tell me if they don't make sense... Also added
3893 Dan Mooney's comments about building on OpenVMS.
3894
3895 19. Added support for partial matching via the PCRE_PARTIAL option for
3896 pcre_exec() and the \P data escape in pcretest.
3897
3898 20. Extended pcretest with 3 new pattern features:
3899
3900 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3901 write the compiled pattern to the file whose name is "rest-of-line".
3902 This is a straight binary dump of the data, with the saved pointer to
3903 the character tables forced to be NULL. The study data, if any, is
3904 written too. After writing, pcretest reads a new pattern.
3905
3906 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3907 compiled pattern from the given file. There must not be any
3908 occurrences of "<" in the file name (pretty unlikely); if there are,
3909 pcretest will instead treat the initial "<" as a pattern delimiter.
3910 After reading in the pattern, pcretest goes on to read data lines as
3911 usual.
3912
3913 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3914 and 16-bit fields in a compiled pattern, to simulate a pattern that
3915 was compiled on a host of opposite endianness.
3916
3917 21. The pcre-exec() function can now cope with patterns that were compiled on
3918 hosts of opposite endianness, with this restriction:
3919
3920 As for any compiled expression that is saved and used later, the tables
3921 pointer field cannot be preserved; the extra_data field in the arguments
3922 to pcre_exec() should be used to pass in a tables address if a value
3923 other than the default internal tables were used at compile time.
3924
3925 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3926 now diagnosed as an error. Previously, most of the time, a negative number
3927 would have been treated as zero, but if in addition "ovector" was passed as
3928 NULL, a crash could occur.
3929
3930 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3931 new versions from the libtool 1.5 distribution (the last one is a copy of
3932 a file called libtool.m4). This seems to have fixed the need to patch
3933 "configure" to support Darwin 1.3 (which I used to do). However, I still
3934 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3935 workstation).
3936
3937 24. Changed the PCRE licence to be the more standard "BSD" licence.
3938
3939
3940 Version 4.5 01-Dec-03
3941 ---------------------
3942
3943 1. There has been some re-arrangement of the code for the match() function so
3944 that it can be compiled in a version that does not call itself recursively.
3945 Instead, it keeps those local variables that need separate instances for
3946 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3947 needs to "recurse". Keeping track of where control must go is done by means
3948 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3949 hide most of the details from the main code, and operates only if
3950 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3951 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3952 operating.
3953
3954 To make it easier for callers to provide specially tailored get/free
3955 functions for this usage, two new functions, pcre_stack_malloc, and
3956 pcre_stack_free, are used. They are always called in strict stacking order,
3957 and the size of block requested is always the same.
3958
3959 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3960 PCRE has been compiled to use the stack or the heap for recursion. The
3961 -C option of pcretest uses this to show which version is compiled.
3962
3963 A new data escape \S, is added to pcretest; it causes the amounts of store
3964 obtained and freed by both kinds of malloc/free at match time to be added
3965 to the output.
3966
3967 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3968 what's available on my current Linux desktop machine.
3969
3970 3. When matching a UTF-8 string, the test for a valid string at the start has
3971 been extended. If start_offset is not zero, PCRE now checks that it points
3972 to a byte that is the start of a UTF-8 character. If not, it returns
3973 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3974 this is necessary because there may be backward assertions in the pattern.
3975 When matching the same subject several times, it may save resources to use
3976 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3977
3978 4. The code for checking the validity of UTF-8 strings has been tightened so
3979 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3980 containing "overlong sequences".
3981
3982 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3983 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3984 should have been "&", but it just so happened that all the cases this let
3985 through by mistake were picked up later in the function.
3986
3987 6. I had used a variable called "isblank" - this is a C99 function, causing
3988 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3989
3990 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3991 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3992 so the version is shown; (c) stop "make test" echoing "./RunTest".
3993
3994 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3995
3996 9. The prototype for memmove() for systems that don't have it was using
3997 size_t, but the inclusion of the header that defines size_t was later. I've
3998 moved the #includes for the C headers earlier to avoid this.
3999
4000 10. Added some adjustments to the code to make it easier to compiler on certain
4001 special systems:
4002
4003 (a) Some "const" qualifiers were missing.
4004 (b) Added the macro EXPORT before all exported functions; by default this
4005 is defined to be empty.
4006 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4007 that it reads its output file name as an argument instead of writing
4008 to the standard output and assuming this can be redirected.
4009
4010 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4011 class containing characters with values greater than 255, PCRE compilation
4012 went into a loop.
4013
4014 12. A recursive reference to a subpattern that was within another subpattern
4015 that had a minimum quantifier of zero caused PCRE to crash. For example,
4016 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4017 recursion. If the recursively-called subpattern itself had a zero repeat,
4018 that was OK.
4019
4020 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4021 buffer into which it was copied (for escape processing) was still set at
4022 1024, so long lines caused crashes.
4023
4024 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4025 "internal error: code overflow...". This applied to any character class
4026 that was followed by a possessive quantifier.
4027
4028 15. Modified the Makefile to add libpcre.la as a prerequisite for
4029 libpcreposix.la because I was told this is needed for a parallel build to
4030 work.
4031
4032 16. If a pattern that contained .* following optional items at the start was
4033 studied, the wrong optimizing data was generated, leading to matching
4034 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4035 matching string must start with a or b or c. The correct conclusion for
4036 this pattern is that a match can start with any character.
4037
4038
4039 Version 4.4 13-Aug-03
4040 ---------------------
4041
4042 1. In UTF-8 mode, a character class containing characters with values between
4043 127 and 255 was not handled correctly if the compiled pattern was studied.
4044 In fixing this, I have also improved the studying algorithm for such
4045 classes (slightly).
4046
4047 2. Three internal functions had redundant arguments passed to them. Removal
4048 might give a very teeny performance improvement.
4049
4050 3. Documentation bug: the value of the capture_top field in a callout is *one
4051 more than* the number of the hightest numbered captured substring.
4052
4053 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4054 in incorrectly linking with a previously installed version. They now link
4055 explicitly with libpcre.la.
4056
4057 5. configure.in no longer needs to recognize Cygwin specially.
4058
4059 6. A problem in pcre.in for Windows platforms is fixed.
4060
4061 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4062 pcretest, it used to include the size of the study block as part of its
4063 output. Unfortunately, the structure contains a field that has a different
4064 size on different hardware architectures. This meant that the tests that
4065 showed this size failed. As the block is currently always of a fixed size,
4066 this information isn't actually particularly useful in pcretest output, so
4067 I have just removed it.
4068
4069 8. Three pre-processor statements accidentally did not start in column 1.
4070 Sadly, there are *still* compilers around that complain, even though
4071 standard C has not required this for well over a decade. Sigh.
4072
4073 9. In pcretest, the code for checking callouts passed small integers in the
4074 callout_data field, which is a void * field. However, some picky compilers
4075 complained about the casts involved for this on 64-bit systems. Now
4076 pcretest passes the address of the small integer instead, which should get
4077 rid of the warnings.
4078
4079 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4080 both compile and run time, and gives an error if an invalid UTF-8 sequence
4081 is found. There is a option for disabling this check in cases where the
4082 string is known to be correct and/or the maximum performance is wanted.
4083
4084 11. In response to a bug report, I changed one line in Makefile.in from
4085
4086 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4087 to
4088 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4089
4090 to look similar to other lines, but I have no way of telling whether this
4091 is the right thing to do, as I do not use Windows. No doubt I'll get told
4092 if it's wrong...
4093
4094
4095 Version 4.3 21-May-03
4096 ---------------------
4097
4098 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4099 Makefile.
4100
4101 2. Some refactoring to improve the quality of the code:
4102
4103 (i) The utf8_table... variables are now declared "const".
4104
4105 (ii) The code for \cx, which used the "case flipping" table to upper case
4106 lower case letters, now just substracts 32. This is ASCII-specific,
4107 but the whole concept of \cx is ASCII-specific, so it seems
4108 reasonable.
4109
4110 (iii) PCRE was using its character types table to recognize decimal and
4111 hexadecimal digits in the pattern. This is silly, because it handles
4112 only 0-9, a-f, and A-F, but the character types table is locale-
4113 specific, which means strange things might happen. A private
4114 table is now used for this - though it costs 256 bytes, a table is
4115 much faster than multiple explicit tests. Of course, the standard
4116 character types table is still used for matching digits in subject
4117 strings against \d.
4118
4119 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4120 ending in _t are). So I've renamed it as ESC_tee.
4121
4122 3. The first argument for regexec() in the POSIX wrapper should have been
4123 defined as "const".
4124
4125 4. Changed pcretest to use malloc() for its buffers so that they can be
4126 Electric Fenced for debugging.
4127
4128 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4129 to read one or more bytes before the start of the subject string. Often this
4130 had no effect on PCRE's behaviour, but in some circumstances it could
4131 provoke a segmentation fault.
4132
4133 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4134 to try to read one or more bytes before the start of the subject string.
4135
4136 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4137 UTF-8 support could misbehave in various ways if the subject string
4138 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4139 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4140 back over UTF-8 characters.)
4141
4142
4143 Version 4.2 14-Apr-03
4144 ---------------------
4145
4146 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4147
4148 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4149 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4150 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4151 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4152 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4153 and BUILD_EXEEXT
4154 Note: automatic setting of the BUILD variables is not yet working
4155 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4156 compile-time but not at link-time
4157 [LINK]: use for linking executables only
4158 make different versions for Windows and non-Windows
4159 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4160 libraries
4161 [LINK_FOR_BUILD]: new variable
4162 [OBJEXT]: use throughout
4163 [EXEEXT]: use throughout
4164 <winshared>: new target
4165 <wininstall>: new target
4166 <dftables.o>: use native compiler
4167 <dftables>: use native linker
4168 <install>: handle Windows platform correctly
4169 <clean>: ditto
4170 <check>: ditto
4171 copy DLL to top builddir before testing
4172
4173 As part of these changes, -no-undefined was removed again. This was reported
4174 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4175 in any case.
4176
4177 3. Some tidies to get rid of compiler warnings:
4178
4179 . In the match_data structure, match_limit was an unsigned long int, whereas
4180 match_call_count was an int. I've made them both unsigned long ints.
4181
4182 . In pcretest the fact that a const uschar * doesn't automatically cast to
4183 a void * provoked a warning.
4184
4185 . Turning on some more compiler warnings threw up some "shadow" variables
4186 and a few more missing casts.
4187
4188 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4189 option, a class that contained a single character with a value between 128
4190 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4191
4192 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4193 option, a class that contained several characters, but with at least one
4194 whose value was between 128 and 255 caused PCRE to crash.
4195
4196
4197 Version 4.1 12-Mar-03
4198 ---------------------
4199
4200 1. Compiling with gcc -pedantic found a couple of places where casts were
4201 needed, and a string in dftables.c that was longer than standard compilers are
4202 required to support.
4203
4204 2. Compiling with Sun's compiler found a few more places where the code could
4205 be tidied up in order to avoid warnings.
4206
4207 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4208 first of these names is deprecated in the latest Autoconf in favour of the name
4209 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4210 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4211 analogy I have changed it to CFLAGS_FOR_BUILD.
4212
4213 4. Added -no-undefined to the linking command in the Makefile, because this is
4214 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4215 linking step for the pcreposix library.
4216
4217 5. PCRE was failing to diagnose the case of two named groups with the same
4218 name.
4219
4220 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4221 literal character that is needed in the subject for a match, and scans along to
4222 ensure that it is present before embarking on the full matching process. This
4223 saves time in cases of nested unlimited repeats that are never going to match.
4224 Problem: the scan can take a lot of time if the subject is very long (e.g.
4225 megabytes), thus penalizing straightforward matches. It is now done only if the
4226 amount of subject to be scanned is less than 1000 bytes.
4227
4228 7. A lesser problem with the same optimization is that it was recording the
4229 first character of an anchored pattern as "needed", thus provoking a search
4230 right along the subject, even when the first match of the pattern was going to
4231 fail. The "needed" character is now not set for anchored patterns, unless it
4232 follows something in the pattern that is of non-fixed length. Thus, it still
4233 fulfils its original purpose of finding quick non-matches in cases of nested
4234 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4235
4236
4237 Version 4.0 17-Feb-03
4238 ---------------------
4239
4240 1. If a comment in an extended regex that started immediately after a meta-item
4241 extended to the end of string, PCRE compiled incorrect data. This could lead to
4242 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4243
4244 2. Moved to autoconf 2.53 and libtool 1.4.2.
4245
4246 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4247 the special perltest8 script is no longer needed - all the tests can be run
4248 from a single perltest script.
4249
4250 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4251 by \s. It has now been removed in PCRE. This means it isn't recognized as
4252 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4253 class [:space:] *does* include VT, thereby creating a mess.
4254
4255 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4256 space and tab.
4257
4258 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4259 its new features into the main test script, reducing the number of scripts.
4260
4261 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4262 were backward compatible, and made the (?i) apply to the whole pattern, as if
4263 /i were given. Now it behaves more logically, and applies the option setting
4264 only to what follows. PCRE has been changed to follow suit. However, if it
4265 finds options settings right at the start of the pattern, it extracts them into
4266 the global options, as before. Thus, they show up in the info data.
4267
4268 8. Added support for the \Q...\E escape sequence. Characters in between are
4269 treated as literals. This is slightly different from Perl in that $ and @ are
4270 also handled as literals inside the quotes. In Perl, they will cause variable
4271 interpolation. Note the following examples:
4272
4273 Pattern PCRE matches Perl matches
4274
4275 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4276 \Qabc\$xyz\E abc\$xyz abc\$xyz
4277 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4278
4279 For compatibility with Perl, \Q...\E sequences are recognized inside character
4280 classes as well as outside them.
4281
4282 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4283 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4284 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4285 signed/unsigned warnings.
4286
4287 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4288 option for pcretest, so I've replaced it by a simple function that does just
4289 that job.
4290
4291 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4292 "pcregrep -".
4293
4294 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4295 Java package. This provides some syntactic sugar for simple cases of what my
4296 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4297 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4298 item, you can use this simplified notation. Note that only makes sense with
4299 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4300 greediness, whatever the setting of the PCRE_UNGREEDY option.
4301
4302 13. A change of greediness default within a pattern was not taking effect at
4303 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4304 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4305 was abstracted outside.
4306
4307 14. PCRE now supports the \G assertion. It is true when the current matching
4308 position is at the start point of the match. This differs from \A when the
4309 starting offset is non-zero. Used with the /g option of pcretest (or similar
4310 code), it works in the same way as it does for Perl's /g option. If all
4311 alternatives of a regex begin with \G, the expression is anchored to the start
4312 match position, and the "anchored" flag is set in the compiled expression.
4313
4314 15. Some bugs concerning the handling of certain option changes within patterns
4315 have been fixed. These applied to options other than (?ims). For example,
4316 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4317 been the other way round. Some of this was related to change 7 above.
4318
4319 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4320 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4321 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4322 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4323
4324 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4325 mode. Unlike ".", it always matches newline, whatever the setting of
4326 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4327 assertions. Perl allows it, but it doesn't (in general) work because it can't
4328 calculate the length of the lookbehind. At least, that's the case for Perl
4329 5.8.0 - I've been told they are going to document that it doesn't work in
4330 future.
4331
4332 18. Added an error diagnosis for escapes that PCRE does not support: these are
4333 \L, \l, \N, \P, \p, \U, \u, and \X.
4334
4335 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4336 reading past the end of the pattern in cases such as /[abcd/.
4337
4338 20. PCRE was getting more memory than necessary for patterns with classes that
4339 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4340
4341 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4342 compiling PCRE for use with Virtual Pascal.
4343
4344 22. Small fix to the Makefile to make it work properly if the build is done
4345 outside the source tree.
4346
4347 23. Added a new extension: a condition to go with recursion. If a conditional
4348 subpattern starts with (?(R) the "true" branch is used if recursion has
4349 happened, whereas the "false" branch is used only at the top level.
4350
4351 24. When there was a very long string of literal characters (over 255 bytes
4352 without UTF support, over 250 bytes with UTF support), the computation of how
4353 much memory was required could be incorrect, leading to segfaults or other
4354 strange effects.
4355
4356 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4357 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4358 there was a subsequent back reference to those brackets. This meant that, for
4359 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4360 possible to check for precisely this case. All we can do is abandon the
4361 optimization if .* occurs inside capturing brackets when there are any back
4362 references whatsoever. (See below for a better fix that came later.)
4363
4364 26. The handling of the optimization for finding the first character of a
4365 non-anchored pattern, and for finding a character that is required later in the
4366 match were failing in some cases. This didn't break the matching; it just
4367 failed to optimize when it could. The way this is done has been re-implemented.
4368
4369 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4370
4371 28. Added a new feature that provides some of the functionality that Perl
4372 provides with (?{...}). The facility is termed a "callout". The way it is done
4373 in PCRE is for the caller to provide an optional function, by setting
4374 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4375 global variable. By default it is unset, which disables all calling out. To get
4376 the function called, the regex must include (?C) at appropriate points. This
4377 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4378 This provides a means of identifying different callout points. When PCRE
4379 reaches such a point in the regex, if pcre_callout has been set, the external
4380 function is called. It is provided with data in a structure called
4381 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4382 matching continues; if it returns a non-zero value, the match at the current
4383 point fails. However, backtracking will occur if possible. [This was changed
4384 later and other features added - see item 49 below.]
4385
4386 29. pcretest is upgraded to test the callout functionality. It provides a
4387 callout function that displays information. By default, it shows the start of
4388 the match and the current position in the text. There are some new data escapes
4389 to vary what happens:
4390
4391 \C+ in addition, show current contents of captured substrings
4392 \C- do not supply a callout function
4393 \C!n return 1 when callout number n is reached
4394 \C!n!m return 1 when callout number n is reached for the mth time
4395
4396 30. If pcregrep was called with the -l option and just a single file name, it
4397 output "<stdin>" if a match was found, instead of the file name.
4398
4399 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4400 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4401 pcre_exec(). This saves a malloc/free per call. The default value of
4402 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4403 when configuring.
4404
4405 32. The default maximum size of a compiled pattern is 64K. There have been a
4406 few cases of people hitting this limit. The code now uses macros to handle the
4407 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4408 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4409 configuring. Tests 2 and 5 work only with 2-byte links because they output
4410 debugging information about compiled patterns.
4411
4412 33. Internal code re-arrangements:
4413
4414 (a) Moved the debugging function for printing out a compiled regex into
4415 its own source file (printint.c) and used #include to pull it into
4416 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4417 separate copies.
4418
4419 (b) Defined the list of op-code names for debugging as a macro in
4420 internal.h so that it is next to the definition of the opcodes.
4421
4422 (c) Defined a table of op-code lengths for simpler skipping along compiled
4423 code. This is again a macro in internal.h so that it is next to the
4424 definition of the opcodes.
4425
4426 34. Added support for recursive calls to individual subpatterns, along the
4427 lines of Robin Houston's patch (but implemented somewhat differently).
4428
4429 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4430 allow it to read and process whole directories in Win32. This code was
4431 contributed by Lionel Fourquaux; it has not been tested by me.
4432
4433 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4434 used to name a group. Names consist of alphanumerics and underscores, and must
4435 be unique. Back references use the syntax (?P=name) and recursive calls use
4436 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4437 numbers. The function pcre_fullinfo() can be used after compilation to extract
4438 a name/number map. There are three relevant calls:
4439
4440 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4441 PCRE_INFO_NAMECOUNT yields the number of entries
4442 PCRE_INFO_NAMETABLE yields a pointer to the map.
4443
4444 The map is a vector of fixed-size entries. The size of each entry depends on
4445 the length of the longest name used. The first two bytes of each entry are the
4446 group number, most significant byte first. There follows the corresponding
4447 name, zero terminated. The names are in alphabetical order.
4448
4449 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4450 case instead of 255. Making it the same both with and without UTF-8 support
4451 means that the same test output works with both.
4452
4453 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4454 calling malloc() with a zero argument.
4455
4456 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4457 optimization. I've improved things by keeping a bitmap of backreferences with
4458 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4459 fact referenced, the optimization can be applied. It is unlikely that a
4460 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4461 the match to follow \n) will appear inside brackets with a number greater than
4462 31, but if it does, any back reference > 31 suppresses the optimization.
4463
4464 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4465 of disabling numbered capturing parentheses. Any opening parenthesis that is
4466 not followed by ? behaves as if it were followed by ?: but named parentheses
4467 can still be used for capturing (and they will acquire numbers in the usual
4468 way).
4469
4470 41. Redesigned the return codes from the match() function into yes/no/error so
4471 that errors can be passed back from deep inside the nested calls. A malloc
4472 failure while inside a recursive subpattern call now causes the
4473 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4474
4475 42. It is now possible to set a limit on the number of times the match()
4476 function is called in a call to pcre_exec(). This facility makes it possible to
4477 limit the amount of recursion and backtracking, though not in a directly
4478 obvious way, because the match() function is used in a number of different
4479 circumstances. The count starts from zero for each position in the subject
4480 string (for non-anchored patterns). The default limit is, for compatibility, a
4481 large number, namely 10 000 000. You can change this in two ways:
4482
4483 (a) When configuring PCRE before making, you can use --with-match-limit=n
4484 to set a default value for the compiled library.
4485
4486 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4487 a different value is set. See 45 below.
4488
4489 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4490
4491 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4492 of things that can be changed at compile time. The first argument specifies
4493 what is wanted and the second points to where the information is to be placed.
4494 The current list of available information is:
4495
4496 PCRE_CONFIG_UTF8
4497
4498 The output is an integer that is set to one if UTF-8 support is available;
4499 otherwise it is set to zero.
4500
4501 PCRE_CONFIG_NEWLINE
4502
4503 The output is an integer that it set to the value of the code that is used for
4504 newline. It is either LF (10) or CR (13).
4505
4506 PCRE_CONFIG_LINK_SIZE
4507
4508 The output is an integer that contains the number of bytes used for internal
4509 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4510
4511 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4512
4513 The output is an integer that contains the threshold above which the POSIX
4514 interface uses malloc() for output vectors. See item 31 above.
4515
4516 PCRE_CONFIG_MATCH_LIMIT
4517
4518 The output is an unsigned integer that contains the default limit of the number
4519 of match() calls in a pcre_exec() execution. See 42 above.
4520
4521 44. pcretest has been upgraded by the addition of the -C option. This causes it
4522 to extract all the available output from the new pcre_config() function, and to
4523 output it. The program then exits immediately.
4524
4525 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4526 order to support additional features. One way would have been to define
4527 pcre_exec2() (for example) with extra arguments, but this would not have been
4528 extensible, and would also have required all calls to the original function to
4529 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4530 is used for passing in "extra" data from pcre_study().
4531
4532 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4533 contains the following fields:
4534
4535 flags a bitmap indicating which of the following fields are set
4536 study_data opaque data from pcre_study()
4537 match_limit a way of specifying a limit on match() calls for a specific
4538 call to pcre_exec()
4539 callout_data data for callouts (see 49 below)
4540
4541 The flag bits are also defined in pcre.h, and are
4542
4543 PCRE_EXTRA_STUDY_DATA
4544 PCRE_EXTRA_MATCH_LIMIT
4545 PCRE_EXTRA_CALLOUT_DATA
4546
4547 The pcre_study() function now returns one of these new pcre_extra blocks, with
4548 the actual study data pointed to by the study_data field, and the
4549 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4550 before. That is, this change is entirely upwards-compatible and requires no
4551 change to existing code.
4552
4553 If you want to pass in additional data to pcre_exec(), you can either place it
4554 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4555 block.
4556
4557 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4558 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4559 times with different match limits, until it finds the minimum value needed for
4560 pcre_exec() to complete. The value is then output. This can be instructive; for
4561 most simple matches the number is quite small, but for pathological cases it
4562 gets very large very quickly.
4563
4564 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4565 returns the size of the data block pointed to by the study_data field in a
4566 pcre_extra block, that is, the value that was passed as the argument to
4567 pcre_malloc() when PCRE was getting memory in which to place the information
4568 created by pcre_study(). The fourth argument should point to a size_t variable.
4569 pcretest has been extended so that this information is shown after a successful
4570 pcre_study() call when information about the compiled regex is being displayed.
4571
4572 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4573 because what follows is always an absolute path. (Later: it turns out that this
4574 is more than cosmetic for MinGW, because it doesn't like empty path
4575 components.)
4576
4577 49. Some changes have been made to the callout feature (see 28 above):
4578
4579 (i) A callout function now has three choices for what it returns:
4580
4581 0 => success, carry on matching
4582 > 0 => failure at this point, but backtrack if possible
4583 < 0 => serious error, return this value from pcre_exec()
4584
4585 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4586 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4587 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4588 use by callout functions. It will never be used by PCRE itself.
4589
4590 (ii) The pcre_extra structure (see 45 above) has a void * field called
4591 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4592 pcre_callout_block structure has a field of the same name. The contents of
4593 the field passed in the pcre_extra structure are passed to the callout
4594 function in the corresponding field in the callout block. This makes it
4595 easier to use the same callout-containing regex from multiple threads. For
4596 testing, the pcretest program has a new data escape
4597
4598 \C*n pass the number n (may be negative) as callout_data
4599
4600 If the callout function in pcretest receives a non-zero value as
4601 callout_data, it returns that value.
4602
4603 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4604 there were some redundant $(CFLAGS) in commands that are now specified as
4605 $(LINK), which already includes $(CFLAGS).
4606
4607 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4608 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4609 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4610 one-byte characters throughout. Note that case-insensitive matching applies
4611 only to characters whose values are less than 256. PCRE doesn't support the
4612 notion of cases for higher-valued characters.
4613
4614 (i) A character class whose characters are all within 0-255 is handled as
4615 a bit map, and the map is inverted for negative classes. Previously, a
4616 character > 255 always failed to match such a class; however it should
4617 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4618
4619 (ii) A negated character class with a single character < 255 is coded as
4620 "not this character" (OP_NOT). This wasn't working properly when the test
4621 character was multibyte, either singly or repeated.
4622
4623 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4624 mode, for example: \x{100}{2,3}.
4625
4626 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4627 singly or repeated) now correctly test multibyte characters. However,
4628 PCRE doesn't recognize any characters with values greater than 255 as
4629 digits, spaces, or word characters. Such characters always match \D, \S,
4630 and \W, and never match \d, \s, or \w.
4631
4632 (v) Classes may now contain characters and character ranges with values
4633 greater than 255. For example: [ab\x{100}-\x{400}].
4634
4635 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4636 PCRE in UTF-8 mode.
4637
4638 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4639 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4640 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4641 value.)
4642
4643 53. The single man page has become too large. I have therefore split it up into
4644 a number of separate man pages. These also give rise to individual HTML pages;
4645 these are now put in a separate directory, and there is an index.html page that
4646 lists them all. Some hyperlinking between the pages has been installed.
4647
4648 54. Added convenience functions for handling named capturing parentheses.
4649
4650 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4651 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4652 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4653 are faulted.
4654
4655 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4656 calling configure. These values are used when compiling the dftables.c program
4657 which is run to generate the source of the default character tables. They
4658 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4659 you will need to set these values.
4660
4661 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4662
4663
4664 Version 3.9 02-Jan-02
4665 ---------------------
4666
4667 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4668
4669 2. If --disable-static was given, the building process failed when trying to
4670 build pcretest and pcregrep. (For some reason it was using libtool to compile
4671 them, which is not right, as they aren't part of the library.)
4672
4673
4674 Version 3.8 18-Dec-01
4675 ---------------------
4676
4677 1. The experimental UTF-8 code was completely screwed up. It was packing the
4678 bytes in the wrong order. How dumb can you get?
4679
4680
4681 Version 3.7 29-Oct-01
4682 ---------------------
4683
4684 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4685 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4686 this didn't happen under Solaris 8, where I normally test things.
4687
4688 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4689 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4690 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4691 this fix an hour or so after the initial 3.7 release.)
4692
4693
4694 Version 3.6 23-Oct-01
4695 ---------------------
4696
4697 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4698 offsets passed as NULL with zero offset count.
4699
4700 2. The config.guess and config.sub files had not been updated when I moved to
4701 the latest autoconf.
4702
4703
4704 Version 3.5 15-Aug-01
4705 ---------------------
4706
4707 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4708 had been forgotten.
4709
4710 2. By using declared but undefined structures, we can avoid using "void"
4711 definitions in pcre.h while keeping the internal definitions of the structures
4712 private.
4713
4714 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4715 user point of view, this means that both static and shared libraries are built
4716 by default, but this can be individually controlled. More of the work of
4717 handling this static/shared cases is now inside libtool instead of PCRE's make
4718 file.
4719
4720 4. The pcretest utility is now installed along with pcregrep because it is
4721 useful for users (to test regexs) and by doing this, it automatically gets
4722 relinked by libtool. The documentation has been turned into a man page, so
4723 there are now .1, .txt, and .html versions in /doc.
4724
4725 5. Upgrades to pcregrep:
4726 (i) Added long-form option names like gnu grep.
4727 (ii) Added --help to list all options with an explanatory phrase.
4728 (iii) Added -r, --recursive to recurse into sub-directories.
4729 (iv) Added -f, --file to read patterns from a file.
4730
4731 6. pcre_exec() was referring to its "code" argument before testing that
4732 argument for NULL (and giving an error if it was NULL).
4733
4734 7. Upgraded Makefile.in to allow for compiling in a different directory from
4735 the source directory.
4736
4737 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4738 options bits, the pointer it was passed was to an int instead of to an unsigned
4739 long int. This mattered only on 64-bit systems.
4740
4741 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4742 generated) instead of pcre.in, which it its source. Also made the same change
4743 in several of the .c files.
4744
4745 10. A new release of gcc defines printf() as a macro, which broke pcretest
4746 because it had an ifdef in the middle of a string argument for printf(). Fixed
4747 by using separate calls to printf().
4748
4749 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4750 script, to force use of CR or LF instead of \n in the source. On non-Unix
4751 systems, the value can be set in config.h.
4752
4753 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4754 absolute limit. Changed the text of the error message to make this clear, and
4755 likewise updated the man page.
4756
4757 13. The limit of 99 on the number of capturing subpatterns has been removed.
4758 The new limit is 65535, which I hope will not be a "real" limit.
4759
4760
4761 Version 3.4 22-Aug-00
4762 ---------------------
4763
4764 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4765
4766 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4767
4768
4769 Version 3.3 01-Aug-00
4770 ---------------------
4771
4772 1. If an octal character was given, but the value was greater than \377, it
4773 was not getting masked to the least significant bits, as documented. This could
4774 lead to crashes in some systems.
4775
4776 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4777 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4778
4779 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4780 These just pass their arguments on to (pcre_free)(), but they are provided
4781 because some uses of PCRE bind it to non-C systems that can call its functions,
4782 but cannot call free() or pcre_free() directly.
4783
4784 4. Add "make test" as a synonym for "make check". Corrected some comments in
4785 the Makefile.
4786
4787 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4788 Makefile.
4789
4790 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4791 command called pgrep for grepping around the active processes.
4792
4793 7. Added the beginnings of support for UTF-8 character strings.
4794
4795 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4796 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4797 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4798 out for the ar command.)
4799
4800
4801 Version 3.2 12-May-00
4802 ---------------------
4803
4804 This is purely a bug fixing release.
4805
4806 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4807 of ZA. This was just one example of several cases that could provoke this bug,
4808 which was introduced by change 9 of version 2.00. The code for breaking
4809 infinite loops after an iteration that matches an empty string was't working
4810 correctly.
4811
4812 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4813 when matched against abbab (for example). After matching an empty string, it
4814 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4815 caused it to match further down the string than it should.
4816
4817 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4818 was there because it doesn't seem to be needed, and it causes trouble on some
4819 systems, as it is not a Standard C header. It has been removed.
4820
4821 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4822 were reported on the Macintosh. The changes were from
4823
4824 while ((c = *(++ptr)) != 0 && c != '\n');
4825 to
4826 while ((c = *(++ptr)) != 0 && c != '\n') ;
4827
4828 Totally extraordinary, but if that's what it takes...
4829
4830 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4831 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4832 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4833 assumes the way PCRE uses memmove() (always moving upwards).
4834
4835 6. PCRE is being used in one environment where strchr() is not available. There
4836 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4837 faster code anyway.
4838
4839
4840 Version 3.1 09-Feb-00
4841 ---------------------
4842
4843 The only change in this release is the fixing of some bugs in Makefile.in for
4844 the "install" target:
4845
4846 (1) It was failing to install pcreposix.h.
4847
4848 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4849
4850
4851 Version 3.0 01-Feb-00
4852 ---------------------
4853
4854 1. Add support for the /+ modifier to perltest (to output $` like it does in
4855 pcretest).
4856
4857 2. Add support for the /g modifier to perltest.
4858
4859 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4860 matches null strings.
4861
4862 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4863 pattern. Perl treats empty patterns specially - it reuses the most recent
4864 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4865 effect.
4866
4867 5. The POSIX interface was broken in that it was just handing over the POSIX
4868 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4869 required a bigger vector, with some working space on the end. This means that
4870 the POSIX wrapper now has to get and free some memory, and copy the results.
4871
4872 6. Added some simple autoconf support, placing the test data and the
4873 documentation in separate directories, re-organizing some of the
4874 information files, and making it build pcre-config (a GNU standard). Also added
4875 libtool support for building PCRE as a shared library, which is now the
4876 default.
4877
4878 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4879 09 are not valid octal constants. Single digits will be used for minor values
4880 less than 10.
4881
4882 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4883 existing programs that set these in the POSIX interface can use PCRE without
4884 modification.
4885
4886 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4887 return all that pcre_info() returns, plus additional data. The pcre_info()
4888 function is retained for compatibility, but is considered to be obsolete.
4889
4890 10. Added experimental recursion feature (?R) to handle one common case that
4891 Perl 5.6 will be able to do with (?p{...}).
4892
4893 11. Added support for POSIX character classes like [:alpha:], which Perl is
4894 adopting.
4895
4896
4897 Version 2.08 31-Aug-99
4898 ----------------------
4899
4900 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4901 trying to match at the startoffset position, but instead was moving forward to
4902 the next newline as if a previous match had failed.
4903
4904 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4905 and could get into a loop if a null string was matched other than at the start
4906 of the subject.
4907
4908 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4909 be distinguished at compile time, and for completeness also added PCRE_DATE.
4910
4911 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4912 in GnuWin32 environments.
4913
4914
4915 Version 2.07 29-Jul-99
4916 ----------------------
4917
4918 1. The documentation is now supplied in plain text form and HTML as well as in
4919 the form of man page sources.
4920
4921 2. C++ compilers don't like assigning (void *) values to other pointer types.
4922 In particular this affects malloc(). Although there is no problem in Standard
4923 C, I've put in casts to keep C++ compilers happy.
4924
4925 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4926 should be (const char *).
4927
4928 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4929 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4930 However, I haven't made this a standard facility. The documentation doesn't
4931 mention it, and the Makefile doesn't support it.
4932
4933 5. The Makefile now contains an "install" target, with editable destinations at
4934 the top of the file. The pcretest program is not installed.
4935
4936 6. pgrep -V now gives the PCRE version number and date.
4937
4938 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4939 causing the entire string to be ignored, instead of just the last character.
4940
4941 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4942 non-matching string, it can take a very, very long time, even for strings of
4943 quite modest length, because of the nested recursion. PCRE now does better in
4944 some of these cases. It does this by remembering the last required literal
4945 character in the pattern, and pre-searching the subject to ensure it is present
4946 before running the real match. In other words, it applies a heuristic to detect
4947 some types of certain failure quickly, and in the above example, if presented
4948 with a string that has no trailing " it gives "no match" very quickly.
4949
4950 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4951 other alternatives are tried instead.
4952
4953
4954 Version 2.06 09-Jun-99
4955 ----------------------
4956
4957 1. Change pcretest's output for amount of store used to show just the code
4958 space, because the remainder (the data block) varies in size between 32-bit and
4959 64-bit systems.
4960
4961 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4962 start matching at. This allows lookbehinds to work when searching for multiple
4963 occurrences in a string.
4964
4965 3. Added additional options to pcretest for testing multiple occurrences:
4966
4967 /+ outputs the rest of the string that follows a match
4968 /g loops for multiple occurrences, using the new startoffset argument
4969 /G loops for multiple occurrences by passing an incremented pointer
4970
4971 4. PCRE wasn't doing the "first character" optimization for patterns starting
4972 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4973 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4974 the letter 'x'. On long subject strings, this gives a significant speed-up.
4975
4976
4977 Version 2.05 21-Apr-99
4978 ----------------------
4979
4980 1. Changed the type of magic_number from int to long int so that it works
4981 properly on 16-bit systems.
4982
4983 2. Fixed a bug which caused patterns starting with .* not to work correctly
4984 when the subject string contained newline characters. PCRE was assuming
4985 anchoring for such patterns in all cases, which is not correct because .* will
4986 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4987 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4988 must be retried after every newline in the subject.
4989
4990
4991 Version 2.04 18-Feb-99
4992 ----------------------
4993
4994 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4995 computation of the store needed to hold the pattern was incorrect (too large).
4996 If such patterns were nested a few deep, this could multiply and become a real
4997 problem.
4998
4999 2. Added /M option to pcretest to show the memory requirement of a specific
5000 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5001
5002 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5003 compiled in such a way that the backtracking after subsequent failure was
5004 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5005 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5006
5007
5008 Version 2.03 02-Feb-99
5009 ----------------------
5010
5011 1. Fixed typo and small mistake in man page.
5012
5013 2. Added 4th condition (GPL supersedes if conflict) and created separate
5014 LICENCE file containing the conditions.
5015
5016 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5017 Perl, that is the internal \ allows the delimiter to be included in the
5018 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5019 the final delimiter, add \ to the end of the pattern (to test the error).
5020
5021 4. Added the convenience functions for extracting substrings after a successful
5022 match. Updated pcretest to make it able to test these functions.
5023
5024
5025 Version 2.02 14-Jan-99
5026 ----------------------
5027
5028 1. Initialized the working variables associated with each extraction so that
5029 their saving and restoring doesn't refer to uninitialized store.
5030
5031 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5032 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5033 fix the problem.
5034
5035 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5036 calls, and wasn't printing the correct value for compiling calls. Increased the
5037 default value of LOOPREPEAT, and the number of significant figures in the
5038 times.
5039
5040 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5041
5042 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5043 a building problem on Windows NT with a FAT file system.
5044
5045
5046 Version 2.01 21-Oct-98
5047 ----------------------
5048
5049 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5050 to character tables built by pcre_maketables() in the current locale. If NULL
5051 is passed, the default tables are used.
5052
5053
5054 Version 2.00 24-Sep-98
5055 ----------------------
5056
5057 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5058 it any more.
5059
5060 2. Allow quantification of (?>) groups, and make it work correctly.
5061
5062 3. The first character computation wasn't working for (?>) groups.
5063
5064 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5065 end of the subject) and add 5.005's \z, which really does match only at the
5066 very end of the subject.
5067
5068 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5069
5070 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5071 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5072 localized options. All options to pcre_study() were also removed.
5073
5074 7. Add other new features from 5.005:
5075
5076 $(?<= positive lookbehind
5077 $(?<! negative lookbehind
5078 (?imsx-imsx) added the unsetting capability
5079 such a setting is global if at outer level; local otherwise
5080 (?imsx-imsx:) non-capturing groups with option setting
5081 (?(cond)re|re) conditional pattern matching
5082
5083 A backreference to itself in a repeated group matches the previous
5084 captured string.
5085
5086 8. General tidying up of studying (both automatic and via "study")
5087 consequential on the addition of new assertions.
5088
5089 9. As in 5.005, unlimited repeated groups that could match an empty substring
5090 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5091 runtime if any iteration does actually match an empty substring.
5092
5093 10. Include the RunTest script in the distribution.
5094
5095 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5096 discrepancies, some of which were old and were also with respect to 5.004. They
5097 have now been fixed.
5098
5099
5100 Version 1.09 28-Apr-98
5101 ----------------------
5102
5103 1. A negated single character class followed by a quantifier with a minimum
5104 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5105 program crashes, or just wrong answers. This did not apply to negated classes
5106 containing more than one character, or to minima other than one.
5107
5108
5109 Version 1.08 27-Mar-98
5110 ----------------------
5111
5112 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5113
5114 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5115 latter must appear before anything that relies on it in the pattern.
5116
5117
5118 Version 1.07 16-Feb-98
5119 ----------------------
5120
5121 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5122 repeat of a potentially empty string).
5123
5124
5125 Version 1.06 23-Jan-98
5126 ----------------------
5127
5128 1. Added Markus Oberhumer's little patches for C++.
5129
5130 2. Literal strings longer than 255 characters were broken.
5131
5132
5133 Version 1.05 23-Dec-97
5134 ----------------------
5135
5136 1. Negated character classes containing more than one character were failing if
5137 PCRE_CASELESS was set at run time.
5138
5139
5140 Version 1.04 19-Dec-97
5141 ----------------------
5142
5143 1. Corrected the man page, where some "const" qualifiers had been omitted.
5144
5145 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5146 input syntax.
5147
5148 3. Fixed memory leak which occurred when a regex with back references was
5149 matched with an offsets vector that wasn't big enough. The temporary memory
5150 that is used in this case wasn't being freed if the match failed.
5151
5152 4. Tidied pcretest to ensure it frees memory that it gets.
5153
5154 5. Temporary memory was being obtained in the case where the passed offsets
5155 vector was exactly big enough.
5156
5157 6. Corrected definition of offsetof() from change 5 below.
5158
5159 7. I had screwed up change 6 below and broken the rules for the use of
5160 setjmp(). Now fixed.
5161
5162
5163 Version 1.03 18-Dec-97
5164 ----------------------
5165
5166 1. A erroneous regex with a missing opening parenthesis was correctly
5167 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5168 on some systems.
5169
5170 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5171 it was reported that one broken compiler failed on the former because "code" is
5172 also an independent variable.
5173
5174 3. The erroneous regex a[]b caused an array overrun reference.
5175
5176 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5177 fail on data ending with that character. (It was going on too far, and checking
5178 the next character, typically a binary zero.) This was specific to the
5179 optimized code for single-character negative classes.
5180
5181 5. Added a contributed patch from the TIN world which does the following:
5182
5183 + Add an undef for memmove, in case the the system defines a macro for it.
5184
5185 + Add a definition of offsetof(), in case there isn't one. (I don't know
5186 the reason behind this - offsetof() is part of the ANSI standard - but
5187 it does no harm).
5188
5189 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5190 most of the places where whitespace preceded '#'. I have given up and
5191 allowed the remaining 2 cases to be at the margin.
5192
5193 + Rename some variables in pcre to eliminate shadowing. This seems very
5194 pedantic, but does no harm, of course.
5195
5196 6. Moved the call to setjmp() into its own function, to get rid of warnings
5197 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5198
5199 7. Constructs such as \d{8,} were compiling into the equivalent of
5200 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5201 outcome, but in this particular case used more store than had been allocated,
5202 which caused the bug to be discovered because it threw up an internal error.
5203
5204 8. The debugging code in both pcre and pcretest for outputting the compiled
5205 form of a regex was going wrong in the case of back references followed by
5206 curly-bracketed repeats.
5207
5208
5209 Version 1.02 12-Dec-97
5210 ----------------------
5211
5212 1. Typos in pcre.3 and comments in the source fixed.
5213
5214 2. Applied a contributed patch to get rid of places where it used to remove
5215 'const' from variables, and fixed some signed/unsigned and uninitialized
5216 variable warnings.
5217
5218 3. Added the "runtest" target to Makefile.
5219
5220 4. Set default compiler flag to -O2 rather than just -O.
5221
5222
5223 Version 1.01 19-Nov-97
5224 ----------------------
5225
5226 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5227 like /([ab]*)*/, that is, for classes with more than one character in them.
5228
5229 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5230 as /((?>a*))*/ (a PCRE_EXTRA facility).
5231
5232
5233 Version 1.00 18-Nov-97
5234 ----------------------
5235
5236 1. Added compile-time macros to support systems such as SunOS4 which don't have
5237 memmove() or strerror() but have other things that can be used instead.
5238
5239 2. Arranged that "make clean" removes the executables.
5240
5241
5242 Version 0.99 27-Oct-97
5243 ----------------------
5244
5245 1. Fixed bug in code for optimizing classes with only one character. It was
5246 initializing a 32-byte map regardless, which could cause it to run off the end
5247 of the memory it had got.
5248
5249 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5250
5251
5252 Version 0.98 22-Oct-97
5253 ----------------------
5254
5255 1. Fixed bug in code for handling temporary memory usage when there are more
5256 back references than supplied space in the ovector. This could cause segfaults.
5257
5258
5259 Version 0.97 21-Oct-97
5260 ----------------------
5261
5262 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5263
5264 2. Optimized negated single characters not to use a bit map.
5265
5266 3. Brought error texts together as macro definitions; clarified some of them;
5267 fixed one that was wrong - it said "range out of order" when it meant "invalid
5268 escape sequence".
5269
5270 4. Changed some char * arguments to const char *.
5271
5272 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5273
5274 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5275 pcretest.
5276
5277
5278 Version 0.96 16-Oct-97
5279 ----------------------
5280
5281 1. Added a simple "pgrep" utility to the distribution.
5282
5283 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5284 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5285 where "ddd" means "one or more decimal digits".
5286
5287 3. Fixed serious bug. If a pattern had a back reference, but the call to
5288 pcre_exec() didn't supply a large enough ovector to record the related
5289 identifying subpattern, the match always failed. PCRE now remembers the number
5290 of the largest back reference, and gets some temporary memory in which to save
5291 the offsets during matching if necessary, in order to ensure that
5292 backreferences always work.
5293
5294 4. Increased the compatibility with Perl in a number of ways:
5295
5296 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5297 to request this handling. The option can be set at compile or exec time.
5298
5299 (b) $ matches before a terminating newline by default; an option
5300 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5301 mode). The option can be set at compile or exec time.
5302
5303 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5304 the same as Perl's. If the decimal number it represents is less than 10
5305 or there aren't that many previous left capturing parentheses, an octal
5306 escape is read. Inside a character class, it's always an octal escape,
5307 even if it is a single digit.
5308
5309 (d) An escaped but undefined alphabetic character is taken as a literal,
5310 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5311 escapes.
5312
5313 (e) {0} is now permitted. (The previous item is removed from the compiled
5314 pattern).
5315
5316 5. Changed all the names of code files so that the basic parts are no longer
5317 than 10 characters, and abolished the teeny "globals.c" file.
5318
5319 6. Changed the handling of character classes; they are now done with a 32-byte
5320 bit map always.
5321
5322 7. Added the -d and /D options to pcretest to make it possible to look at the
5323 internals of compilation without having to recompile pcre.
5324
5325
5326 Version 0.95 23-Sep-97
5327 ----------------------
5328
5329 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5330 \x20 at the start of a run of normal characters. These were being treated as
5331 real characters, instead of the source characters being re-checked.
5332
5333
5334 Version 0.94 18-Sep-97
5335 ----------------------
5336
5337 1. The functions are now thread-safe, with the caveat that the global variables
5338 containing pointers to malloc() and free() or alternative functions are the
5339 same for all threads.
5340
5341 2. Get pcre_study() to generate a bitmap of initial characters for non-
5342 anchored patterns when this is possible, and use it if passed to pcre_exec().
5343
5344
5345 Version 0.93 15-Sep-97
5346 ----------------------
5347
5348 1. /(b)|(:+)/ was computing an incorrect first character.
5349
5350 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5351 but not actually doing anything yet.
5352
5353 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5354 as Perl does (e.g. [-az] or [az-]).
5355
5356 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5357 all possible positions.
5358
5359 5. Split up into different modules to avoid including unneeded functions in a
5360 compiled binary. However, compile and exec are still in one module. The "study"
5361 function is split off.
5362
5363 6. The character tables are now in a separate module whose source is generated
5364 by an auxiliary program - but can then be edited by hand if required. There are
5365 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5366 toupper() in the code.
5367
5368 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5369 make them global. Abolish the function for setting them, as the caller can now
5370 set them directly.
5371
5372
5373 Version 0.92 11-Sep-97
5374 ----------------------
5375
5376 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5377 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5378
5379 2. Caseless matching was not working in character classes if the characters in
5380 the pattern were in upper case.
5381
5382 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5383
5384 4. Make PCRE_ANCHORED public and accept as a compile option.
5385
5386 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5387 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5388 pass them.
5389
5390 6. Give an error if bad option bits passed at compile or run time.
5391
5392 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5393 pcretest to cause it to pass that flag.
5394
5395 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5396 options, and the first character, if set.
5397
5398 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5399
5400
5401 Version 0.91 10-Sep-97
5402 ----------------------
5403
5404 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5405 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5406
5407 2. PCRE was looping on encountering an indefinitely repeated back reference to
5408 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5409 Perl does - treats the match as successful.
5410
5411 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5