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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1492 - (show annotations)
Tue Jul 8 16:16:14 2014 UTC (5 years, 3 months ago) by ph10
File size: 267648 byte(s)
Fixed several memory leaks in pcregrep.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.36 xx-xxx-2014
5 ------------------------
6
7 1. Got rid of some compiler warnings in the C++ modules that were shown up by
8 -Wmissing-field-initializers and -Wunused-parameter.
9
10 2. The tests for quantifiers being too big (greater than 65535) were being
11 applied after reading the number, and stupidly assuming that integer
12 overflow would give a negative number. The tests are now applied as the
13 numbers are read.
14
15 3. Tidy code in pcre_exec.c where two branches that used to be different are
16 now the same.
17
18 4. The JIT compiler did not generate match limit checks for certain
19 bracketed expressions with quantifiers. This may lead to exponential
20 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
21 issue should be resolved now.
22
23 5. Fixed an issue, which occures when nested alternatives are optimized
24 with table jumps.
25
26 6. Inserted two casts and changed some ints to size_t in the light of some
27 reported 64-bit compiler warnings (Bugzilla 1477).
28
29 7. Fixed a bug concerned with zero-minimum possessive groups that could match
30 an empty string, which sometimes were behaving incorrectly in the
31 interpreter (though correctly in the JIT matcher). This pcretest input is
32 an example:
33
34 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
35 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
36
37 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
38 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
39 for an empty string was breaking the inner loop and carrying on at a lower
40 level, when possessive repeated groups should always return to a higher
41 level as they have no backtrack points in them. The empty string test now
42 occurs at the outer level.
43
44 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
45 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
46
47 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
48 doesn't).
49
50 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
51 to hold a pattern to be incorrectly computed (too small) when there were
52 named back references to duplicated names. This could cause "internal
53 error: code overflow" or "double free or corruption" or other memory
54 handling errors.
55
56 11. When named subpatterns had the same prefixes, back references could be
57 confused. For example, in this pattern:
58
59 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
60
61 the reference to 'Name' was incorrectly treated as a reference to a
62 duplicate name.
63
64 12. A pattern such as /^s?c/mi8 where the optional character has more than
65 one "other case" was incorrectly compiled such that it would only try to
66 match starting at "c".
67
68 13. When a pattern starting with \s was studied, VT was not included in the
69 list of possible starting characters; this should have been part of the
70 8.34/18 patch.
71
72 14. If a character class started [\Qx]... where x is any character, the class
73 was incorrectly terminated at the ].
74
75 15. If a pattern that started with a caseless match for a character with more
76 than one "other case" was studied, PCRE did not set up the starting code
77 unit bit map for the list of possible characters. Now it does. This is an
78 optimization improvement, not a bug fix.
79
80 16. The Unicode data tables have been updated to Unicode 7.0.0.
81
82 17. Fixed a number of memory leaks in pcregrep.
83
84
85 Version 8.35 04-April-2014
86 --------------------------
87
88 1. A new flag is set, when property checks are present in an XCLASS.
89 When this flag is not set, PCRE can perform certain optimizations
90 such as studying these XCLASS-es.
91
92 2. The auto-possessification of character sets were improved: a normal
93 and an extended character set can be compared now. Furthermore
94 the JIT compiler optimizes more character set checks.
95
96 3. Got rid of some compiler warnings for potentially uninitialized variables
97 that show up only when compiled with -O2.
98
99 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
100 of a match later then the end of the match. The pcretest program was not
101 handling the case sensibly - it was outputting from the start to the next
102 binary zero. It now reports this situation in a message, and outputs the
103 text from the end to the start.
104
105 5. Fast forward search is improved in JIT. Instead of the first three
106 characters, any three characters with fixed position can be searched.
107 Search order: first, last, middle.
108
109 6. Improve character range checks in JIT. Characters are read by an inprecise
110 function now, which returns with an unknown value if the character code is
111 above a certain treshold (e.g: 256). The only limitation is that the value
112 must be bigger than the treshold as well. This function is useful, when
113 the characters above the treshold are handled in the same way.
114
115 7. The macros whose names start with RAWUCHAR are placeholders for a future
116 mode in which only the bottom 21 bits of 32-bit data items are used. To
117 make this more memorable for those maintaining the code, the names have
118 been changed to start with UCHAR21, and an extensive comment has been added
119 to their definition.
120
121 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
122 to the export list in Makefile.am (they were accidentally omitted from the
123 8.34 tarball).
124
125 9. The informational output from pcretest used the phrase "starting byte set"
126 which is inappropriate for the 16-bit and 32-bit libraries. As the output
127 for "first char" and "need char" really means "non-UTF-char", I've changed
128 "byte" to "char", and slightly reworded the output. The documentation about
129 these values has also been (I hope) clarified.
130
131 10. Another JIT related optimization: use table jumps for selecting the correct
132 backtracking path, when more than four alternatives are present inside a
133 bracket.
134
135 11. Empty match is not possible, when the minimum length is greater than zero,
136 and there is no \K in the pattern. JIT should avoid empty match checks in
137 such cases.
138
139 12. In a caseless character class with UCP support, when a character with more
140 than one alternative case was not the first character of a range, not all
141 the alternative cases were added to the class. For example, s and \x{17f}
142 are both alternative cases for S: the class [RST] was handled correctly,
143 but [R-T] was not.
144
145 13. The configure.ac file always checked for pthread support when JIT was
146 enabled. This is not used in Windows, so I have put this test inside a
147 check for the presence of windows.h (which was already tested for).
148
149 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
150 The algorithm provides a way to skip certain starting offsets, and usually
151 faster than linear prefix searches.
152
153 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
154 as for 'fr_FR' and 'french'. For some reason, however, it then used the
155 Windows-specific input and output files, which have 'french' screwed in.
156 So this could never have worked. One of the problems with locales is that
157 they aren't always the same. I have now updated RunTest so that it checks
158 the output of the locale test (test 3) against three different output
159 files, and it allows the test to pass if any one of them matches. With luck
160 this should make the test pass on some versions of Solaris where it was
161 failing. Because of the uncertainty, the script did not used to stop if
162 test 3 failed; it now does. If further versions of a French locale ever
163 come to light, they can now easily be added.
164
165 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
166 there was a message during ./configure, but it did not stop. This now
167 provokes an error. The invalid example in README has been corrected.
168 If a value less than the minimum is given, the minimum value has always
169 been used, but now a warning is given.
170
171 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
172 was a bug in the test system, which is now fixed. Also, the list of various
173 configurations that are tested for each release did not have one with both
174 16/32 bits and --enable-bar-anycrlf. It now does.
175
176 18. pcretest was missing "-C bsr" for displaying the \R default setting.
177
178 19. Little endian PowerPC systems are supported now by the JIT compiler.
179
180 20. The fast forward newline mechanism could enter to an infinite loop on
181 certain invalid UTF-8 input. Although we don't support these cases
182 this issue can be fixed by a performance optimization.
183
184 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
185 not take account if existing stack usage. There is now a new global
186 variable called pcre_stack_guard that can be set to point to an external
187 function to check stack availability. It is called at the start of
188 processing every parenthesized group.
189
190 22. A typo in the code meant that in ungreedy mode the max/min qualifier
191 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
192 not match "ab".
193
194 23. When UTF was disabled, the JIT program reported some incorrect compile
195 errors. These messages are silenced now.
196
197 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
198 compiler.
199
200 25. Change all the temporary files used in RunGrepTest to be different to those
201 used by RunTest so that the tests can be run simultaneously, for example by
202 "make -j check".
203
204
205 Version 8.34 15-December-2013
206 -----------------------------
207
208 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
209 executable memory. Patch inspired by Carsten Klein.
210
211 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
212 this macro is never tested and has no effect, because the work to support
213 coverage involves only compiling and linking options and special targets in
214 the Makefile. The comment in config.h implied that defining the macro would
215 enable coverage support, which is totally false. There was also support for
216 setting this macro in the CMake files (my fault, I just copied it from
217 configure). SUPPORT_GCOV has now been removed.
218
219 3. Make a small performance improvement in strlen16() and strlen32() in
220 pcretest.
221
222 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
223 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
224 these cases). There was also one in pcretest.c.
225
226 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
227
228 6. In UTF mode, the code for checking whether a group could match an empty
229 string (which is used for indefinitely repeated groups to allow for
230 breaking an infinite loop) was broken when the group contained a repeated
231 negated single-character class with a character that occupied more than one
232 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
233 UTF-8 mode). The effect was undefined: the group might or might not be
234 deemed as matching an empty string, or the program might have crashed.
235
236 7. The code for checking whether a group could match an empty string was not
237 recognizing that \h, \H, \v, \V, and \R must match a character.
238
239 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
240 an empty string. If it can, pcretest shows this in its information output.
241
242 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
243 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
244 matched by pcre_exec() without using JIT:
245
246 (a) If the rest of the pattern did not match after a maximal run of
247 grapheme clusters, the code for backing up to try with fewer of them
248 did not always back up over a full grapheme when characters that do not
249 have the modifier quality were involved, e.g. Hangul syllables.
250
251 (b) If the match point in a subject started with modifier character, and
252 there was no match, the code could incorrectly back up beyond the match
253 point, and potentially beyond the first character in the subject,
254 leading to a segfault or an incorrect match result.
255
256 10. A conditional group with an assertion condition could lead to PCRE
257 recording an incorrect first data item for a match if no other first data
258 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
259 first data item, and therefore matched "ca" after "c" instead of at the
260 start.
261
262 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
263 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
264
265 12. The source of pcregrep now includes z/OS-specific code so that it can be
266 compiled for z/OS as part of the special z/OS distribution.
267
268 13. Added the -T and -TM options to pcretest.
269
270 14. The code in pcre_compile.c for creating the table of named capturing groups
271 has been refactored. Instead of creating the table dynamically during the
272 actual compiling pass, the information is remembered during the pre-compile
273 pass (on the stack unless there are more than 20 named groups, in which
274 case malloc() is used) and the whole table is created before the actual
275 compile happens. This has simplified the code (it is now nearly 150 lines
276 shorter) and prepared the way for better handling of references to groups
277 with duplicate names.
278
279 15. A back reference to a named subpattern when there is more than one of the
280 same name now checks them in the order in which they appear in the pattern.
281 The first one that is set is used for the reference. Previously only the
282 first one was inspected. This change makes PCRE more compatible with Perl.
283
284 16. Unicode character properties were updated from Unicode 6.3.0.
285
286 17. The compile-time code for auto-possessification has been refactored, based
287 on a patch by Zoltan Herczeg. It now happens after instead of during
288 compilation. The code is cleaner, and more cases are handled. The option
289 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
290 options in pcretest are provided to set it. It can also be set by
291 (*NO_AUTO_POSSESS) at the start of a pattern.
292
293 18. The character VT has been added to the default ("C" locale) set of
294 characters that match \s and are generally treated as white space,
295 following this same change in Perl 5.18. There is now no difference between
296 "Perl space" and "POSIX space". Whether VT is treated as white space in
297 other locales depends on the locale.
298
299 19. The code for checking named groups as conditions, either for being set or
300 for being recursed, has been refactored (this is related to 14 and 15
301 above). Processing unduplicated named groups should now be as fast at
302 numerical groups, and processing duplicated groups should be faster than
303 before.
304
305 20. Two patches to the CMake build system, by Alexander Barkov:
306
307 (1) Replace the "source" command by "." in CMakeLists.txt because
308 "source" is a bash-ism.
309
310 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
311 without these the CMake build does not work on Solaris.
312
313 21. Perl has changed its handling of \8 and \9. If there is no previously
314 encountered capturing group of those numbers, they are treated as the
315 literal characters 8 and 9 instead of a binary zero followed by the
316 literals. PCRE now does the same.
317
318 22. Following Perl, added \o{} to specify codepoints in octal, making it
319 possible to specify values greater than 0777 and also making them
320 unambiguous.
321
322 23. Perl now gives an error for missing closing braces after \x{... instead of
323 treating the string as literal. PCRE now does the same.
324
325 24. RunTest used to grumble if an inappropriate test was selected explicitly,
326 but just skip it when running all tests. This make it awkward to run ranges
327 of tests when one of them was inappropriate. Now it just skips any
328 inappropriate tests, as it always did when running all tests.
329
330 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
331 character types such as \d or \w, too many callouts were inserted, and the
332 data that they returned was rubbish.
333
334 26. In UCP mode, \s was not matching two of the characters that Perl matches,
335 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
336 were matched by \h. The code has now been refactored so that the lists of
337 the horizontal and vertical whitespace characters used for \h and \v (which
338 are defined only in one place) are now also used for \s.
339
340 27. Add JIT support for the 64 bit TileGX architecture.
341 Patch by Jiong Wang (Tilera Corporation).
342
343 28. Possessive quantifiers for classes (both explicit and automatically
344 generated) now use special opcodes instead of wrapping in ONCE brackets.
345
346 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
347 (because it's meaningless), this was not happening when PCRE_CASELESS was
348 set. Not wrong, but inefficient.
349
350 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
351 properties for \w, \d, etc) is present in a test regex. Otherwise if the
352 test contains no characters greater than 255, Perl doesn't realise it
353 should be using Unicode semantics.
354
355 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
356 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
357 does in Unicode mode.
358
359 32. Added the "forbid" facility to pcretest so that putting tests into the
360 wrong test files can sometimes be quickly detected.
361
362 33. There is now a limit (default 250) on the depth of nesting of parentheses.
363 This limit is imposed to control the amount of system stack used at compile
364 time. It can be changed at build time by --with-parens-nest-limit=xxx or
365 the equivalent in CMake.
366
367 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
368 errors. Perl warns for these when in warning mode, but PCRE has no facility
369 for giving warnings.
370
371 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
372 However, this was not working for (?!) because it is optimized to (*FAIL),
373 for which PCRE does not allow quantifiers. The optimization is now disabled
374 when a quantifier follows (?!). I can't see any use for this, but it makes
375 things uniform.
376
377 36. Perl no longer allows group names to start with digits, so I have made this
378 change also in PCRE. It simplifies the code a bit.
379
380 37. In extended mode, Perl ignores spaces before a + that indicates a
381 possessive quantifier. PCRE allowed a space before the quantifier, but not
382 before the possessive +. It now does.
383
384 38. The use of \K (reset reported match start) within a repeated possessive
385 group such as (a\Kb)*+ was not working.
386
387 40. Document that the same character tables must be used at compile time and
388 run time, and that the facility to pass tables to pcre_exec() and
389 pcre_dfa_exec() is for use only with saved/restored patterns.
390
391 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
392 features for Builds with MSVC:
393
394 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
395 builds). The .rc files can be used to set FileDescription and many other
396 attributes.
397
398 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
399 This allows higher-level build scripts which want .pdb files to avoid
400 hard-coding the exact files needed."
401
402 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
403 mean "start of word" and "end of word", respectively, as a transition aid.
404
405 43. A minimizing repeat of a class containing codepoints greater than 255 in
406 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
407 compiled to use the heap for recursion.
408
409 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
410 is configured.
411
412
413 Version 8.33 28-May-2013
414 ------------------------
415
416 1. Added 'U' to some constants that are compared to unsigned integers, to
417 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
418 variables that are added to signed variables, to ensure the result is
419 signed and can be negated.
420
421 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
422 CMake config files.
423
424 3. Revise the creation of config.h.generic so that all boolean macros are
425 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
426 overriding via -D on the command line possible.
427
428 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
429 to unsigned int is reported to make a quite noticeable speed difference in
430 a specific Windows environment. Testing on Linux did also appear to show
431 some benefit (and it is clearly not harmful). Also fixed the definition of
432 Xop which should be unsigned.
433
434 5. Related to (4), changing the definition of the intermediate variable cc
435 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
436 improvements.
437
438 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
439 unnecessary spaces.
440
441 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
442 and later.
443
444 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
445
446 9. Optimizing fast_forward_start_bits in JIT.
447
448 10. Adding support for callouts in JIT, and fixing some issues revealed
449 during this work. Namely:
450
451 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
452
453 (b) Minimum length was not checked before the matching is started.
454
455 11. The value of capture_last that is passed to callouts was incorrect in some
456 cases when there was a capture on one path that was subsequently abandoned
457 after a backtrack. Also, the capture_last value is now reset after a
458 recursion, since all captures are also reset in this case.
459
460 12. The interpreter no longer returns the "too many substrings" error in the
461 case when an overflowing capture is in a branch that is subsequently
462 abandoned after a backtrack.
463
464 13. In the pathological case when an offset vector of size 2 is used, pcretest
465 now prints out the matched string after a yield of 0 or 1.
466
467 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
468 Only supported by the JIT compiler at the moment.
469
470 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
471
472 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
473 offset of the starting point of the matching process, provided the offsets
474 vector is large enough.
475
476 17. The \A escape now records a lookbehind value of 1, though its execution
477 does not actually inspect the previous character. This is to ensure that,
478 in partial multi-segment matching, at least one character from the old
479 segment is retained when a new segment is processed. Otherwise, if there
480 are no lookbehinds in the pattern, \A might match incorrectly at the start
481 of a new segment.
482
483 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
484
485 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
486 gives some modest performance improvement in 8-bit mode.
487
488 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
489 be expressed in certain programming languages using Universal Character
490 Names.
491
492 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
493 which points out that "non characters" are not "characters that may not
494 appear in Unicode strings" but rather "characters that are reserved for
495 internal use and have only local meaning".
496
497 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
498 there was a conditional group that depended on an assertion, if the
499 assertion was false, the callout that immediately followed the alternation
500 in the condition was skipped when pcre_exec() was used for matching.
501
502 23. Allow an explicit callout to be inserted before an assertion that is the
503 condition for a conditional group, for compatibility with automatic
504 callouts, which always insert a callout at this point.
505
506 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
507 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
508
509 25. (*PRUNE) is now supported by the JIT compiler.
510
511 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
512
513 27. Fix the case where there are two or more SKIPs with arguments that may be
514 ignored.
515
516 28. (*SKIP) is now supported by the JIT compiler.
517
518 29. (*THEN) is now supported by the JIT compiler.
519
520 30. Update RunTest with additional test selector options.
521
522 31. The way PCRE handles backtracking verbs has been changed in two ways.
523
524 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
525 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
526 backtracking. In some cases this makes it more Perl-compatible, but Perl's
527 rather obscure rules do not always do the same thing.
528
529 (2) Previously, backtracking verbs were confined within assertions. This is
530 no longer the case for positive assertions, except for (*ACCEPT). Again,
531 this sometimes improves Perl compatibility, and sometimes does not.
532
533 32. A number of tests that were in test 2 because Perl did things differently
534 have been moved to test 1, because either Perl or PCRE has changed, and
535 these tests are now compatible.
536
537 32. Backtracking control verbs are now handled in the same way in JIT and
538 interpreter.
539
540 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
541 contained a forward subroutine reference caused a compile error.
542
543 34. Auto-detect and optimize limited repetitions in JIT.
544
545 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
546 blocking (*UTF) etc.
547
548 36. In the interpreter, maximizing pattern repetitions for characters and
549 character types now use tail recursion, which reduces stack usage.
550
551 37. The value of the max lookbehind was not correctly preserved if a compiled
552 and saved regex was reloaded on a host of different endianness.
553
554 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
555 of the compiled pattern block, expand the flags field from 16 to 32 bits
556 because it was almost full.
557
558 39. Try madvise first before posix_madvise.
559
560 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
561 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
562 original change locked out all matches of empty strings. This has been
563 changed so that one match of an empty string per line is recognized.
564 Subsequent searches on the same line (for colouring or for --only-matching,
565 for example) do not recognize empty strings.
566
567 41. Applied a user patch to fix a number of spelling mistakes in comments.
568
569 42. Data lines longer than 65536 caused pcretest to crash.
570
571 43. Clarified the data type for length and startoffset arguments for pcre_exec
572 and pcre_dfa_exec in the function-specific man pages, where they were
573 explicitly stated to be in bytes, never having been updated. I also added
574 some clarification to the pcreapi man page.
575
576 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
577 a segmentation fault.
578
579
580 Version 8.32 30-November-2012
581 -----------------------------
582
583 1. Improved JIT compiler optimizations for first character search and single
584 character iterators.
585
586 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
587 Patch by Daniel Richard G.
588
589 3. Single character iterator optimizations in the JIT compiler.
590
591 4. Improved JIT compiler optimizations for character ranges.
592
593 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
594 Reported by Giuseppe D'Angelo.
595
596 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
597 of a line, was being set incorrectly in cases where .* appeared inside
598 atomic brackets at the start of a pattern, or where there was a subsequent
599 *PRUNE or *SKIP.
600
601 7. Improved instruction cache flush for POWER/PowerPC.
602 Patch by Daniel Richard G.
603
604 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
605 grep:
606
607 (a) There is now no limit to the number of patterns to be matched.
608
609 (b) An error is given if a pattern is too long.
610
611 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
612 are now supported.
613
614 (d) --exclude-from and --include-from (multiple use) have been added.
615
616 (e) Exclusions and inclusions now apply to all files and directories, not
617 just to those obtained from scanning a directory recursively.
618
619 (f) Multiple uses of -f and --file-list are now supported.
620
621 (g) In a Windows environment, the default for -d has been changed from
622 "read" (the GNU grep default) to "skip", because otherwise the presence
623 of a directory in the file list provokes an error.
624
625 (h) The documentation has been revised and clarified in places.
626
627 9. Improve the matching speed of capturing brackets.
628
629 10. Changed the meaning of \X so that it now matches a Unicode extended
630 grapheme cluster.
631
632 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
633 out POSIX threads when JIT support is configured.
634
635 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
636
637 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
638 structure could go wrong in environments where size_t is not the same size
639 as int.
640
641 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
642 set.
643
644 15. The EBCDIC support had decayed; later updates to the code had included
645 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
646 general tidy up of EBCDIC-related issues, and the documentation was also
647 not quite right. There is now a test that can be run on ASCII systems to
648 check some of the EBCDIC-related things (but is it not a full test).
649
650 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
651 in a small tidy to the code.
652
653 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
654
655 18. If the --only-matching (-o) option in pcregrep is specified multiple
656 times, each one causes appropriate output. For example, -o1 -o2 outputs the
657 substrings matched by the 1st and 2nd capturing parentheses. A separating
658 string can be specified by --om-separator (default empty).
659
660 19. Improving the first n character searches.
661
662 20. Turn case lists for horizontal and vertical white space into macros so that
663 they are defined only once.
664
665 21. This set of changes together give more compatible Unicode case-folding
666 behaviour for characters that have more than one other case when UCP
667 support is available.
668
669 (a) The Unicode property table now has offsets into a new table of sets of
670 three or more characters that are case-equivalent. The MultiStage2.py
671 script that generates these tables (the pcre_ucd.c file) now scans
672 CaseFolding.txt instead of UnicodeData.txt for character case
673 information.
674
675 (b) The code for adding characters or ranges of characters to a character
676 class has been abstracted into a generalized function that also handles
677 case-independence. In UTF-mode with UCP support, this uses the new data
678 to handle characters with more than one other case.
679
680 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
681 whose other case is greater than 256 are now correctly matched
682 caselessly. Previously, the high codepoint matched the low one, but not
683 vice versa.
684
685 (d) The processing of \h, \H, \v, and \ in character classes now makes use
686 of the new class addition function, using character lists defined as
687 macros alongside the case definitions of 20 above.
688
689 (e) Caseless back references now work with characters that have more than
690 one other case.
691
692 (f) General caseless matching of characters with more than one other case
693 is supported.
694
695 22. Unicode character properties were updated from Unicode 6.2.0
696
697 23. Improved CMake support under Windows. Patch by Daniel Richard G.
698
699 24. Add support for 32-bit character strings, and UTF-32
700
701 25. Major JIT compiler update (code refactoring and bugfixing).
702 Experimental Sparc 32 support is added.
703
704 26. Applied a modified version of Daniel Richard G's patch to create
705 pcre.h.generic and config.h.generic by "make" instead of in the
706 PrepareRelease script.
707
708 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
709 pcre_compile.c when checking for a zero character.
710
711 28. Introducing a native interface for JIT. Through this interface, the compiled
712 machine code can be directly executed. The purpose of this interface is to
713 provide fast pattern matching, so several sanity checks are not performed.
714 However, feature tests are still performed. The new interface provides
715 1.4x speedup compared to the old one.
716
717 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
718 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
719 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
720
721 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
722 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
723 the "old" RFC 2279). Instead, it ended up passing a negative length to
724 pcre_exec().
725
726 31. Add support for GCC's visibility feature to hide internal functions.
727
728 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
729 "unknown -C option" after outputting 0 or 1.
730
731 33. There is now support for generating a code coverage report for the test
732 suite in environments where gcc is the compiler and lcov is installed. This
733 is mainly for the benefit of the developers.
734
735 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
736 unaddressable using valgrind annotations, allowing valgrind to detect
737 invalid memory accesses. This is mainly for the benefit of the developers.
738
739 25. (*UTF) can now be used to start a pattern in any of the three libraries.
740
741 26. Give configure error if --enable-cpp but no C++ compiler found.
742
743
744 Version 8.31 06-July-2012
745 -------------------------
746
747 1. Fixing a wrong JIT test case and some compiler warnings.
748
749 2. Removed a bashism from the RunTest script.
750
751 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
752 to unsigned type, result still unsigned" that was given by an MS compiler
753 on encountering the code "-sizeof(xxx)".
754
755 4. Partial matching support is added to the JIT compiler.
756
757 5. Fixed several bugs concerned with partial matching of items that consist
758 of more than one character:
759
760 (a) /^(..)\1/ did not partially match "aba" because checking references was
761 done on an "all or nothing" basis. This also applied to repeated
762 references.
763
764 (b) \R did not give a hard partial match if \r was found at the end of the
765 subject.
766
767 (c) \X did not give a hard partial match after matching one or more
768 characters at the end of the subject.
769
770 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
771 a partial match for the string "\r".
772
773 (e) When newline was set to CRLF, the metacharacter "." did not recognize
774 a partial match for a CR character at the end of the subject string.
775
776 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
777 running pcretest, the text "(JIT)" added to the output whenever JIT is
778 actually used to run the match.
779
780 7. Individual JIT compile options can be set in pcretest by following -s+[+]
781 or /S+[+] with a digit between 1 and 7.
782
783 8. OP_NOT now supports any UTF character not just single-byte ones.
784
785 9. (*MARK) control verb is now supported by the JIT compiler.
786
787 10. The command "./RunTest list" lists the available tests without actually
788 running any of them. (Because I keep forgetting what they all are.)
789
790 11. Add PCRE_INFO_MAXLOOKBEHIND.
791
792 12. Applied a (slightly modified) user-supplied patch that improves performance
793 when the heap is used for recursion (compiled with --disable-stack-for-
794 recursion). Instead of malloc and free for each heap frame each time a
795 logical recursion happens, frames are retained on a chain and re-used where
796 possible. This sometimes gives as much as 30% improvement.
797
798 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
799 call.
800
801 14. As documented, (*COMMIT) is now confined to within a positive assertion.
802
803 15. It is now possible to link pcretest with libedit as an alternative to
804 libreadline.
805
806 16. (*COMMIT) control verb is now supported by the JIT compiler.
807
808 17. The Unicode data tables have been updated to Unicode 6.1.0.
809
810 18. Added --file-list option to pcregrep.
811
812 19. Added binary file support to pcregrep, including the -a, --binary-files,
813 -I, and --text options.
814
815 20. The madvise function is renamed for posix_madvise for QNX compatibility
816 reasons. Fixed by Giuseppe D'Angelo.
817
818 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
819
820 22. Changed the option for creating symbolic links for 16-bit man pages from
821 -s to -sf so that re-installing does not cause issues.
822
823 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
824
825 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
826 matches in certain environments (the workspace was not being correctly
827 retained). Also added to pcre_dfa_exec() a simple plausibility check on
828 some of the workspace data at the beginning of a restart.
829
830 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
831 was not doing so when it should - probably a typo introduced by SVN 528
832 (change 8.10/14).
833
834 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
835 \w+ when the character tables indicated that \x{c4} was a word character.
836 There were several related cases, all because the tests for doing a table
837 lookup were testing for characters less than 127 instead of 255.
838
839 27. If a pattern contains capturing parentheses that are not used in a match,
840 their slots in the ovector are set to -1. For those that are higher than
841 any matched groups, this happens at the end of processing. In the case when
842 there were back references that the ovector was too small to contain
843 (causing temporary malloc'd memory to be used during matching), and the
844 highest capturing number was not used, memory off the end of the ovector
845 was incorrectly being set to -1. (It was using the size of the temporary
846 memory instead of the true size.)
847
848 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
849 ovector size, it uses memory at the end of the block that it has got.
850
851 29. Check for an overlong MARK name and give an error at compile time. The
852 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
853
854 30. JIT compiler update.
855
856 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
857 Rill for the patch.
858
859 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
860
861 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
862
863 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
864 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
865 was enabled.
866
867 35. Improve JIT code generation for greedy plus quantifier.
868
869 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
870 1 to "aa" instead of to an empty string. The bug affected repeated groups
871 that could potentially match an empty string.
872
873 37. Optimizing single character iterators in JIT.
874
875 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
876 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
877 codepoints that are too big for the mode are faulted, and in a UTF mode,
878 disallowed codepoints are also faulted.
879
880 39. If PCRE was compiled with UTF support, in three places in the DFA
881 matcher there was code that should only have been obeyed in UTF mode, but
882 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
883 processing when bytes with values greater than 127 were present. In 16-bit
884 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
885 both cases the values are those that cannot be the first data item in a UTF
886 character. The three items that might have provoked this were recursions,
887 possessively repeated groups, and atomic groups.
888
889 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
890 and pcregrep, because some OS require shared objects to be explicitly
891 passed to ld, causing the link step to fail if they are not.
892
893 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
894 mode, patterns that started with \h* or \R* might be incorrectly matched.
895
896
897 Version 8.30 04-February-2012
898 -----------------------------
899
900 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
901 name is defined in ctype.h.
902
903 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
904 only in quite long subpatterns.
905
906 3. Removed the function pcre_info(), which has been obsolete and deprecated
907 since it was replaced by pcre_fullinfo() in February 2000.
908
909 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
910 match a (*MARK), and the match failed at the start of the subject, a
911 reference to memory before the start of the subject could occur. This bug
912 was introduced by fix 17 of release 8.21.
913
914 5. A reference to an unset group with zero minimum repetition was giving
915 totally wrong answers (in non-JavaScript-compatibility mode). For example,
916 /(another)?(\1?)test/ matched against "hello world test". This bug was
917 introduced in release 8.13.
918
919 6. Add support for 16-bit character strings (a large amount of work involving
920 many changes and refactorings).
921
922 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
923 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
924 from a file.
925
926 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
927 rounding is not applied in this particular case).
928
929 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
930 if they appear, or are escaped, in patterns.
931
932 10. Get rid of a number of -Wunused-but-set-variable warnings.
933
934 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
935 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
936 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
937 also returns the mark "x". This bug applied to capturing parentheses,
938 non-capturing parentheses, and atomic parentheses. It also applied to some
939 assertions.
940
941 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
942 information out of configure.ac instead of relying on pcre.h.generic, which
943 is not stored in the repository.
944
945 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
946 -lreadline.
947
948 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
949
950 15. Applied Graycode's patch to put the top-level frame on the stack rather
951 than the heap when not using the stack for recursion. This gives a
952 performance improvement in many cases when recursion is not deep.
953
954 16. Experimental code added to "pcretest -C" to output the stack frame size.
955
956
957 Version 8.21 12-Dec-2011
958 ------------------------
959
960 1. Updating the JIT compiler.
961
962 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
963 are added as well.
964
965 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
966 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
967 calling _pcre_jit_exec. Some extra comments are added.
968
969 4. (*MARK) settings inside atomic groups that do not contain any capturing
970 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
971 was introduced by change 18 for 8.20.
972
973 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
974 ECMA-262 standard.
975
976 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
977 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
978 This bug was probably introduced by change 9 of 8.13.
979
980 7. While fixing 6 above, I noticed that a number of other items were being
981 incorrectly rejected as "not fixed length". This arose partly because newer
982 opcodes had not been added to the fixed-length checking code. I have (a)
983 corrected the bug and added tests for these items, and (b) arranged for an
984 error to occur if an unknown opcode is encountered while checking for fixed
985 length instead of just assuming "not fixed length". The items that were
986 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
987 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
988 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
989
990 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
991 being incorrectly compiled and would have given unpredicatble results.
992
993 9. A possessively repeated subpattern with minimum repeat count greater than
994 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
995 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
996 the first (A) could occur when it should not.
997
998 10. Add a cast and remove a redundant test from the code.
999
1000 11. JIT should use pcre_malloc/pcre_free for allocation.
1001
1002 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1003 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1004 is anything other than /usr, -L is still shown).
1005
1006 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1007
1008 14. Perl does not support \N without a following name in a [] class; PCRE now
1009 also gives an error.
1010
1011 15. If a forward reference was repeated with an upper limit of around 2000,
1012 it caused the error "internal error: overran compiling workspace". The
1013 maximum number of forward references (including repeats) was limited by the
1014 internal workspace, and dependent on the LINK_SIZE. The code has been
1015 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1016 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1017 of around 200,000 forward references. While doing this, I also speeded up
1018 the filling in of repeated forward references.
1019
1020 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1021 incorrectly expecting the subject to contain another "a" after the start.
1022
1023 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1024 in the match, the SKIP should be ignored. This was not happening; instead
1025 the SKIP was being treated as NOMATCH. For patterns such as
1026 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1027 tested.
1028
1029 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1030 now much more compatible with Perl, in particular in cases where the result
1031 is a non-match for a non-anchored pattern. For example, if
1032 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1033 "m", where previously it did not return a name. A side effect of this
1034 change is that for partial matches, the last encountered mark name is
1035 returned, as for non matches. A number of tests that were previously not
1036 Perl-compatible have been moved into the Perl-compatible test files. The
1037 refactoring has had the pleasing side effect of removing one argument from
1038 the match() function, thus reducing its stack requirements.
1039
1040 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1041 subsequent uses of /S (without +) incorrectly behaved like /S+.
1042
1043 21. Retrieve executable code size support for the JIT compiler and fixing
1044 some warnings.
1045
1046 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1047 not work when the shorter character appeared right at the end of the
1048 subject string.
1049
1050 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1051 systems.
1052
1053 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1054 output it when the /M option is used in pcretest.
1055
1056 25. The CheckMan script was not being included in the distribution. Also, added
1057 an explicit "perl" to run Perl scripts from the PrepareRelease script
1058 because this is reportedly needed in Windows.
1059
1060 26. If study data was being save in a file and studying had not found a set of
1061 "starts with" bytes for the pattern, the data written to the file (though
1062 never used) was taken from uninitialized memory and so caused valgrind to
1063 complain.
1064
1065 27. Updated RunTest.bat as provided by Sheri Pierce.
1066
1067 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1068
1069 29. Computation of memory usage for the table of capturing group names was
1070 giving an unnecessarily large value.
1071
1072
1073 Version 8.20 21-Oct-2011
1074 ------------------------
1075
1076 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1077 a POSIX class. After further experiments with Perl, which convinced me that
1078 Perl has bugs and confusions, a closing square bracket is no longer allowed
1079 in a POSIX name. This bug also affected patterns with classes that started
1080 with full stops.
1081
1082 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1083 captured substring, but while checking the failing first alternative,
1084 substring 1 is temporarily captured. If the output vector supplied to
1085 pcre_exec() was not big enough for this capture, the yield of the function
1086 was still zero ("insufficient space for captured substrings"). This cannot
1087 be totally fixed without adding another stack variable, which seems a lot
1088 of expense for a edge case. However, I have improved the situation in cases
1089 such as /(a)(b)x|abc/ matched against "abc", where the return code
1090 indicates that fewer than the maximum number of slots in the ovector have
1091 been set.
1092
1093 3. Related to (2) above: when there are more back references in a pattern than
1094 slots in the output vector, pcre_exec() uses temporary memory during
1095 matching, and copies in the captures as far as possible afterwards. It was
1096 using the entire output vector, but this conflicts with the specification
1097 that only 2/3 is used for passing back captured substrings. Now it uses
1098 only the first 2/3, for compatibility. This is, of course, another edge
1099 case.
1100
1101 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1102 main code base, and can be used by building with --enable-jit. When this is
1103 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1104 runtime --no-jit option is given.
1105
1106 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1107 ovector, the return from the function was zero, implying that there were
1108 other matches that did not fit. The correct "exactly full" value is now
1109 returned.
1110
1111 6. If a subpattern that was called recursively or as a subroutine contained
1112 (*PRUNE) or any other control that caused it to give a non-standard return,
1113 invalid errors such as "Error -26 (nested recursion at the same subject
1114 position)" or even infinite loops could occur.
1115
1116 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1117 computing the minimum length on reaching *ACCEPT, and so ended up with the
1118 wrong value of 1 rather than 0. Further investigation indicates that
1119 computing a minimum subject length in the presence of *ACCEPT is difficult
1120 (think back references, subroutine calls), and so I have changed the code
1121 so that no minimum is registered for a pattern that contains *ACCEPT.
1122
1123 8. If (*THEN) was present in the first (true) branch of a conditional group,
1124 it was not handled as intended. [But see 16 below.]
1125
1126 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1127 Sheri Pierce.
1128
1129 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1130 the first byte in a match must be "a".
1131
1132 11. Change 17 for 8.13 increased the recursion depth for patterns like
1133 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1134 pattern contains any instances of (*THEN). If it does not, the old
1135 optimizations are restored. It would be nice to do this on a per-group
1136 basis, but at the moment that is not feasible.
1137
1138 12. In some environments, the output of pcretest -C is CRLF terminated. This
1139 broke RunTest's code that checks for the link size. A single white space
1140 character after the value is now allowed for.
1141
1142 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1143 For "fr", it uses the Windows-specific input and output files.
1144
1145 14. If (*THEN) appeared in a group that was called recursively or as a
1146 subroutine, it did not work as intended. [But see next item.]
1147
1148 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1149 pattern fragments (but not containing any | characters). If A and B are
1150 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1151 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1152 D. In other words, Perl considers parentheses that do not contain any |
1153 characters to be part of a surrounding alternative, whereas PCRE was
1154 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1155 differently. PCRE now behaves in the same way as Perl, except in the case
1156 of subroutine/recursion calls such as (?1) which have in any case always
1157 been different (but PCRE had them first :-).
1158
1159 16. Related to 15 above: Perl does not treat the | in a conditional group as
1160 creating alternatives. Such a group is treated in the same way as an
1161 ordinary group without any | characters when processing (*THEN). PCRE has
1162 been changed to match Perl's behaviour.
1163
1164 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1165 RunGrepTest script failed.
1166
1167 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1168 inevitable for groups that contain captures, but it can lead to a lot of
1169 stack use in large patterns. The old behaviour has been restored for atomic
1170 groups that do not contain any capturing parentheses.
1171
1172 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1173 suppress the check for a minimum subject length at run time. (If it was
1174 given to pcre_exec() or pcre_dfa_exec() it did work.)
1175
1176 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1177 fail to work when decoding hex characters in data strings in EBCDIC
1178 environments.
1179
1180 21. It appears that in at least one Mac OS environment, the isxdigit() function
1181 is implemented as a macro that evaluates to its argument more than once,
1182 contravening the C 90 Standard (I haven't checked a later standard). There
1183 was an instance in pcretest which caused it to go wrong when processing
1184 \x{...} escapes in subject strings. The has been rewritten to avoid using
1185 things like p++ in the argument of isxdigit().
1186
1187
1188 Version 8.13 16-Aug-2011
1189 ------------------------
1190
1191 1. The Unicode data tables have been updated to Unicode 6.0.0.
1192
1193 2. Two minor typos in pcre_internal.h have been fixed.
1194
1195 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1196 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1197 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1198
1199 4. There were a number of related bugs in the code for matching backrefences
1200 caselessly in UTF-8 mode when codes for the characters concerned were
1201 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1202 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1203 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1204 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1205 2-byte code at the end of the subject (it thought there wasn't enough data
1206 left).
1207
1208 5. Comprehensive information about what went wrong is now returned by
1209 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1210 as the output vector has at least 2 elements. The offset of the start of
1211 the failing character and a reason code are placed in the vector.
1212
1213 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1214 now returned is for the first byte of the failing character, instead of the
1215 last byte inspected. This is an incompatible change, but I hope it is small
1216 enough not to be a problem. It makes the returned offset consistent with
1217 pcre_exec() and pcre_dfa_exec().
1218
1219 7. pcretest now gives a text phrase as well as the error number when
1220 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1221 failure, the offset and reason code are output.
1222
1223 8. When \R was used with a maximizing quantifier it failed to skip backwards
1224 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1225 back over a single character (\n). This seems wrong (because it treated the
1226 two characters as a single entity when going forwards), conflicts with the
1227 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1228 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1229 has been changed.
1230
1231 9. Some internal refactoring has changed the processing so that the handling
1232 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1233 time (the PCRE_DOTALL option was changed this way some time ago: version
1234 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1235 which was always a bit of a fudge. It also means that there is one less
1236 argument for the match() function, which reduces its stack requirements
1237 slightly. This change also fixes an incompatibility with Perl: the pattern
1238 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1239
1240 10. More internal refactoring has drastically reduced the number of recursive
1241 calls to match() for possessively repeated groups such as (abc)++ when
1242 using pcre_exec().
1243
1244 11. While implementing 10, a number of bugs in the handling of groups were
1245 discovered and fixed:
1246
1247 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1248 (a|)*(?1) gave a compile-time internal error.
1249 ((a|)+)+ did not notice that the outer group could match an empty string.
1250 (^a|^)+ was not marked as anchored.
1251 (.*a|.*)+ was not marked as matching at start or after a newline.
1252
1253 12. Yet more internal refactoring has removed another argument from the match()
1254 function. Special calls to this function are now indicated by setting a
1255 value in a variable in the "match data" data block.
1256
1257 13. Be more explicit in pcre_study() instead of relying on "default" for
1258 opcodes that mean there is no starting character; this means that when new
1259 ones are added and accidentally left out of pcre_study(), testing should
1260 pick them up.
1261
1262 14. The -s option of pcretest has been documented for ages as being an old
1263 synonym of -m (show memory usage). I have changed it to mean "force study
1264 for every regex", that is, assume /S for every regex. This is similar to -i
1265 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1266 using it. It makes it easier to run collections of tests with and without
1267 study enabled, and thereby test pcre_study() more easily. All the standard
1268 tests are now run with and without -s (but some patterns can be marked as
1269 "never study" - see 20 below).
1270
1271 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1272 restoration of the capturing data to the outer values was not happening
1273 correctly.
1274
1275 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1276 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1277 pattern had matched an empty string, and so incorrectly returned a no
1278 match.
1279
1280 17. There was optimizing code for the last branch of non-capturing parentheses,
1281 and also for the obeyed branch of a conditional subexpression, which used
1282 tail recursion to cut down on stack usage. Unfortunately, now that there is
1283 the possibility of (*THEN) occurring in these branches, tail recursion is
1284 no longer possible because the return has to be checked for (*THEN). These
1285 two optimizations have therefore been removed. [But see 8.20/11 above.]
1286
1287 18. If a pattern containing \R was studied, it was assumed that \R always
1288 matched two bytes, thus causing the minimum subject length to be
1289 incorrectly computed because \R can also match just one byte.
1290
1291 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1292 was incorrectly computed.
1293
1294 20. If /S is present twice on a test pattern in pcretest input, it now
1295 *disables* studying, thereby overriding the use of -s on the command line
1296 (see 14 above). This is necessary for one or two tests to keep the output
1297 identical in both cases.
1298
1299 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1300 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1301
1302 22. When an atomic group that contained a capturing parenthesis was
1303 successfully matched, but the branch in which it appeared failed, the
1304 capturing was not being forgotten if a higher numbered group was later
1305 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1306 group 1 to "a", when in fact it should be unset. This applied to multi-
1307 branched capturing and non-capturing groups, repeated or not, and also to
1308 positive assertions (capturing in negative assertions does not happen
1309 in PCRE) and also to nested atomic groups.
1310
1311 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1312 subject after a captured substring, to make it easier to tell which of a
1313 number of identical substrings has been captured.
1314
1315 24. The way atomic groups are processed by pcre_exec() has been changed so that
1316 if they are repeated, backtracking one repetition now resets captured
1317 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1318 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1319 "aaa". Previously, it would have been "a". As part of this code
1320 refactoring, the way recursive calls are handled has also been changed.
1321
1322 25. If an assertion condition captured any substrings, they were not passed
1323 back unless some other capturing happened later. For example, if
1324 (?(?=(a))a) was matched against "a", no capturing was returned.
1325
1326 26. When studying a pattern that contained subroutine calls or assertions,
1327 the code for finding the minimum length of a possible match was handling
1328 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1329 group 1 called group 2 while simultaneously a separate group 2 called group
1330 1). A stack overflow occurred in this case. I have fixed this by limiting
1331 the recursion depth to 10.
1332
1333 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1334 Fortmann. This supports explicit test numbers on the command line, and has
1335 argument validation and error reporting.
1336
1337 28. An instance of \X with an unlimited repeat could fail if at any point the
1338 first character it looked at was a mark character.
1339
1340 29. Some minor code refactoring concerning Unicode properties and scripts
1341 should reduce the stack requirement of match() slightly.
1342
1343 30. Added the '=' option to pcretest to check the setting of unused capturing
1344 slots at the end of the pattern, which are documented as being -1, but are
1345 not included in the return count.
1346
1347 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1348 compiled something random. Now it gives a compile-time error (as does
1349 Perl).
1350
1351 32. A *MARK encountered during the processing of a positive assertion is now
1352 recorded and passed back (compatible with Perl).
1353
1354 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1355 had alternative anchored branches, the search for a second match in a line
1356 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1357 matched the line "0102" twice. The same bug affected patterns that started
1358 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1359 twice.
1360
1361 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1362 does, and because of capturing effects, quantifying parenthesized
1363 assertions may at times be useful. Quantifiers are now allowed for
1364 parenthesized assertions.
1365
1366 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1367
1368 36. \g was being checked for fancy things in a character class, when it should
1369 just be a literal "g".
1370
1371 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1372 appearance of a nested POSIX class supersedes an apparent external class.
1373 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1374 unescaped square brackets may also appear as part of class names. For
1375 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1376 more like Perl. (But see 8.20/1 above.)
1377
1378 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1379 was because it thought it was \N{name}, which is not supported).
1380
1381 39. Add minix to OS list not supporting the -S option in pcretest.
1382
1383 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1384 cannot analyze patterns in sufficient detail to catch mutual recursions
1385 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1386 subgroup is called recursively as a subpattern for a second time at the
1387 same position in the subject string. In previous releases this might have
1388 been caught by the recursion limit, or it might have run out of stack.
1389
1390 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1391 happen only once. PCRE was, however incorrectly giving a compile time error
1392 "recursive call could loop indefinitely" because it cannot analyze the
1393 pattern in sufficient detail. The compile time test no longer happens when
1394 PCRE is compiling a conditional subpattern, but actual runaway loops are
1395 now caught at runtime (see 40 above).
1396
1397 42. It seems that Perl allows any characters other than a closing parenthesis
1398 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1399 has been changed to be the same.
1400
1401 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1402 as not to get warnings when autogen.sh is called. Also changed
1403 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1404
1405 44. To help people who use pcregrep to scan files containing exceedingly long
1406 lines, the following changes have been made:
1407
1408 (a) The default value of the buffer size parameter has been increased from
1409 8K to 20K. (The actual buffer used is three times this size.)
1410
1411 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1412 PCRE is built.
1413
1414 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1415 to be set at run time.
1416
1417 (d) Numerical values in pcregrep options can be followed by K or M, for
1418 example --buffer-size=50K.
1419
1420 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1421 given and the return code is set to 2.
1422
1423 45. Add a pointer to the latest mark to the callout data block.
1424
1425 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1426 partial match of an empty string instead of no match. This was specific to
1427 the use of ".".
1428
1429 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1430 complete match instead of a partial match. This bug was dependent on both
1431 the PCRE_UTF8 and PCRE_DOTALL options being set.
1432
1433 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1434 starting byte set, because \b was not being ignored.
1435
1436
1437 Version 8.12 15-Jan-2011
1438 ------------------------
1439
1440 1. Fixed some typos in the markup of the man pages, and wrote a script that
1441 checks for such things as part of the documentation building process.
1442
1443 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1444 --match-limit and --recursion-limit options (added for 8.11). In
1445 particular, this made one of the standard tests fail. (The integer value
1446 went into the wrong half of a long int.)
1447
1448 3. If the --colour option was given to pcregrep with -v (invert match), it
1449 did strange things, either producing crazy output, or crashing. It should,
1450 of course, ignore a request for colour when reporting lines that do not
1451 match.
1452
1453 4. Another pcregrep bug caused similar problems if --colour was specified with
1454 -M (multiline) and the pattern match finished with a line ending.
1455
1456 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1457 matched in multiline mode, the following line was shown as part of the
1458 match. This seems wrong, so I have changed it.
1459
1460 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1461 the check for further matches in the same line (so they could be coloured)
1462 to overrun the end of the current line. If another match was found, it was
1463 incorrectly shown (and then shown again when found in the next line).
1464
1465 7. If pcregrep was compiled under Windows, there was a reference to the
1466 function pcregrep_exit() before it was defined. I am assuming this was
1467 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1468 reported by a user. I've moved the definition above the reference.
1469
1470
1471 Version 8.11 10-Dec-2010
1472 ------------------------
1473
1474 1. (*THEN) was not working properly if there were untried alternatives prior
1475 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1476 backtracked to try for "b" instead of moving to the next alternative branch
1477 at the same level (in this case, to look for "c"). The Perl documentation
1478 is clear that when (*THEN) is backtracked onto, it goes to the "next
1479 alternative in the innermost enclosing group".
1480
1481 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1482 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1483 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1484 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1485 (*THEN).
1486
1487 3. If \s appeared in a character class, it removed the VT character from
1488 the class, even if it had been included by some previous item, for example
1489 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1490 of \s, but is part of the POSIX "space" class.)
1491
1492 4. A partial match never returns an empty string (because you can always
1493 match an empty string at the end of the subject); however the checking for
1494 an empty string was starting at the "start of match" point. This has been
1495 changed to the "earliest inspected character" point, because the returned
1496 data for a partial match starts at this character. This means that, for
1497 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1498 (previously it gave "no match").
1499
1500 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1501 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1502 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1503 has an implication that the given string is incomplete (because a partial
1504 match is preferred over a full match). For this reason, these items now
1505 give a partial match in this situation. [Aside: previously, the one case
1506 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1507 match rather than a full match, which was wrong by the old rules, but is
1508 now correct.]
1509
1510 6. There was a bug in the handling of #-introduced comments, recognized when
1511 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1512 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1513 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1514 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1515 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1516 places in pcre_compile().
1517
1518 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1519 comments when looking ahead for named forward references to subpatterns,
1520 the only newline sequence it recognized was NL. It now handles newlines
1521 according to the set newline convention.
1522
1523 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1524 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1525 cater for a lack of strerror(). These oversights have been fixed.
1526
1527 9. Added --match-limit and --recursion-limit to pcregrep.
1528
1529 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1530
1531 11. When the -o option was used, pcregrep was setting a return code of 1, even
1532 when matches were found, and --line-buffered was not being honoured.
1533
1534 12. Added an optional parentheses number to the -o and --only-matching options
1535 of pcregrep.
1536
1537 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1538 can match an empty string. The code to do it in pcretest and pcredemo
1539 needed fixing:
1540
1541 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1542 only one byte after an empty string match just before CRLF (this case
1543 just got forgotten; "any" and "anycrlf" were OK).
1544
1545 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1546 mode when an empty string match preceded an ASCII character followed by
1547 a non-ASCII character. (The code for advancing by one character rather
1548 than one byte was nonsense.)
1549
1550 (c) The pcredemo.c sample program did not have any code at all to handle
1551 the cases when CRLF is a valid newline sequence.
1552
1553 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1554 as a starting offset was within the subject string. There is now a new
1555 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1556 negative or greater than the length of the string. In order to test this,
1557 pcretest is extended to allow the setting of negative starting offsets.
1558
1559 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1560 starting offset points to the beginning of a UTF-8 character was
1561 unnecessarily clumsy. I tidied it up.
1562
1563 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1564 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1565
1566 17. Nobody had reported that the --include_dir option, which was added in
1567 release 7.7 should have been called --include-dir (hyphen, not underscore)
1568 for compatibility with GNU grep. I have changed it to --include-dir, but
1569 left --include_dir as an undocumented synonym, and the same for
1570 --exclude-dir, though that is not available in GNU grep, at least as of
1571 release 2.5.4.
1572
1573 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1574 characters from a string of bytes) have been redefined so as not to use
1575 loops, in order to improve performance in some environments. At the same
1576 time, I abstracted some of the common code into auxiliary macros to save
1577 repetition (this should not affect the compiled code).
1578
1579 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1580 compile-time error is now given if \c is not followed by an ASCII
1581 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1582 different, and any byte value is allowed.)
1583
1584 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1585 START_OPTIMIZE option, which is now allowed at compile time - but just
1586 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1587 to pcregrep and other applications that have no direct access to PCRE
1588 options. The new /Y option in pcretest sets this option when calling
1589 pcre_compile().
1590
1591 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1592 back references. Groups containing recursive back references were forced to
1593 be atomic by that change, but in the case of named groups, the amount of
1594 memory required was incorrectly computed, leading to "Failed: internal
1595 error: code overflow". This has been fixed.
1596
1597 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1598 pcretest.c, to avoid build problems in some Borland environments.
1599
1600
1601 Version 8.10 25-Jun-2010
1602 ------------------------
1603
1604 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1605 THEN.
1606
1607 2. (*ACCEPT) was not working when inside an atomic group.
1608
1609 3. Inside a character class, \B is treated as a literal by default, but
1610 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1611 causes the error). The code is unchanged, but I tidied the documentation.
1612
1613 4. Inside a character class, PCRE always treated \R and \X as literals,
1614 whereas Perl faults them if its -w option is set. I have changed PCRE so
1615 that it faults them when PCRE_EXTRA is set.
1616
1617 5. Added support for \N, which always matches any character other than
1618 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1619
1620 6. When compiling pcregrep with newer versions of gcc which may have
1621 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1622 declared with attribute warn_unused_result" were given. Just casting the
1623 result to (void) does not stop the warnings; a more elaborate fudge is
1624 needed. I've used a macro to implement this.
1625
1626 7. Minor change to pcretest.c to avoid a compiler warning.
1627
1628 8. Added four artifical Unicode properties to help with an option to make
1629 \s etc use properties (see next item). The new properties are: Xan
1630 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1631
1632 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1633 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1634 this option. Modified pcretest to add /W to test this facility. Added
1635 REG_UCP to make it available via the POSIX interface.
1636
1637 10. Added --line-buffered to pcregrep.
1638
1639 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1640 studied, and the match started with a letter with a code point greater than
1641 127 whose first byte was different to the first byte of the other case of
1642 the letter, the other case of this starting letter was not recognized
1643 (#976).
1644
1645 12. If a pattern that was studied started with a repeated Unicode property
1646 test, for example, \p{Nd}+, there was the theoretical possibility of
1647 setting up an incorrect bitmap of starting bytes, but fortunately it could
1648 not have actually happened in practice until change 8 above was made (it
1649 added property types that matched character-matching opcodes).
1650
1651 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1652 possible starting bytes for non-anchored patterns.
1653
1654 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1655 \R, and also a number of cases that involve Unicode properties, both
1656 explicit and implicit when PCRE_UCP is set.
1657
1658 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1659 input, it could crash or give wrong results if characters with values
1660 greater than 0xc0 were present in the subject string. (Detail: it assumed
1661 UTF-8 input when processing these items.)
1662
1663 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1664 size_t is 64-bit (#991).
1665
1666 17. Added a check for running out of memory when PCRE is compiled with
1667 --disable-stack-for-recursion (#990).
1668
1669 18. If the last data line in a file for pcretest does not have a newline on
1670 the end, a newline was missing in the output.
1671
1672 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1673 less than 128) in its various bitmaps. However, there is a facility for
1674 generating tables according to the current locale when PCRE is compiled. It
1675 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1676 characters, are recognized by isspace() and therefore were getting set in
1677 these tables, and indeed these tables seem to approximate to ISO 8859. This
1678 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1679 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1680 which of course cannot start UTF-8 characters. I have changed the code so
1681 that only real ASCII characters (less than 128) and the correct starting
1682 bytes for UTF-8 encodings are set for characters greater than 127 when in
1683 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1684 altogether.)
1685
1686 20. Added the /T option to pcretest so as to be able to run tests with non-
1687 standard character tables, thus making it possible to include the tests
1688 used for 19 above in the standard set of tests.
1689
1690 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1691 reference to a subpattern the other side of a comment that contains an
1692 opening parenthesis caused either an internal compiling error, or a
1693 reference to the wrong subpattern.
1694
1695
1696 Version 8.02 19-Mar-2010
1697 ------------------------
1698
1699 1. The Unicode data tables have been updated to Unicode 5.2.0.
1700
1701 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1702 configured.
1703
1704 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1705 original author of that file, following a query about its status.
1706
1707 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1708 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1709
1710 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1711 quantifier applied to a forward-referencing subroutine call, could compile
1712 incorrect code or give the error "internal error: previously-checked
1713 referenced subpattern not found".
1714
1715 6. Both MS Visual Studio and Symbian OS have problems with initializing
1716 variables to point to external functions. For these systems, therefore,
1717 pcre_malloc etc. are now initialized to local functions that call the
1718 relevant global functions.
1719
1720 7. There were two entries missing in the vectors called coptable and poptable
1721 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1722 I've fixed the data, and added a kludgy way of testing at compile time that
1723 the lengths are correct (equal to the number of opcodes).
1724
1725 8. Following on from 7, I added a similar kludge to check the length of the
1726 eint vector in pcreposix.c.
1727
1728 9. Error texts for pcre_compile() are held as one long string to avoid too
1729 much relocation at load time. To find a text, the string is searched,
1730 counting zeros. There was no check for running off the end of the string,
1731 which could happen if a new error number was added without updating the
1732 string.
1733
1734 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1735
1736 11. \K was not working if it appeared in an atomic group or in a group that
1737 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1738 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1739 the assertion is positive, but not if it is negative.
1740
1741 12. Change 11 fortuitously reduced the size of the stack frame used in the
1742 "match()" function of pcre_exec.c by one pointer. Forthcoming
1743 implementation of support for (*MARK) will need an extra pointer on the
1744 stack; I have reserved it now, so that the stack frame size does not
1745 decrease.
1746
1747 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1748 item in branch that calls a recursion is a subroutine call - as in the
1749 second branch in the above example - was incorrectly given the compile-
1750 time error "recursive call could loop indefinitely" because pcre_compile()
1751 was not correctly checking the subroutine for matching a non-empty string.
1752
1753 14. The checks for overrunning compiling workspace could trigger after an
1754 overrun had occurred. This is a "should never occur" error, but it can be
1755 triggered by pathological patterns such as hundreds of nested parentheses.
1756 The checks now trigger 100 bytes before the end of the workspace.
1757
1758 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1759
1760
1761 Version 8.01 19-Jan-2010
1762 ------------------------
1763
1764 1. If a pattern contained a conditional subpattern with only one branch (in
1765 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1766 computed the wrong minimum data length (which is of course zero for such
1767 subpatterns). This could cause incorrect "no match" results.
1768
1769 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1770 the pattern is reset in the first branch, pcre_compile() failed with
1771 "internal error: code overflow at offset...". This happened only when
1772 the reset was to the original external option setting. (An optimization
1773 abstracts leading options settings into an external setting, which was the
1774 cause of this.)
1775
1776 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1777 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1778 assertion pattern did not match (meaning that the assertion was true), it
1779 was incorrectly treated as false if the SKIP had been reached during the
1780 matching. This also applied to assertions used as conditions.
1781
1782 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1783 assertion subpattern, including such a pattern used as a condition,
1784 unpredictable results occurred, instead of the error return
1785 PCRE_ERROR_DFA_UITEM.
1786
1787 5. The C++ GlobalReplace function was not working like Perl for the special
1788 situation when an empty string is matched. It now does the fancy magic
1789 stuff that is necessary.
1790
1791 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1792 removed. (These were left over from very, very early versions of PCRE.)
1793
1794 7. Some cosmetic changes to the code to make life easier when compiling it
1795 as part of something else:
1796
1797 (a) Change DEBUG to PCRE_DEBUG.
1798
1799 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1800 called "current" as "current_branch", to prevent a collision with the
1801 Linux macro when compiled as a kernel module.
1802
1803 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1804 prevent a collision with the Linux macro when compiled as a kernel
1805 module.
1806
1807 8. In pcre_compile() there are some checks for integer overflows that used to
1808 cast potentially large values to (double). This has been changed to that
1809 when building, a check for int64_t is made, and if it is found, it is used
1810 instead, thus avoiding the use of floating point arithmetic. (There is no
1811 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1812 double.
1813
1814 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1815 2005 (difference between two addresses compared to an unsigned value).
1816
1817 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1818 custom one, because of the following reported problem in Windows:
1819
1820 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1821 under Win32.
1822 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1823 therefore missing the function definition.
1824 - The compiler thus generates a "C" signature for the test function.
1825 - The linker fails to find the "C" function.
1826 - PCRE fails to configure if asked to do so against libbz2.
1827
1828 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1829 messages were output:
1830
1831 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1832 rerunning libtoolize, to keep the correct libtool macros in-tree.
1833 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1834
1835 I have done both of these things.
1836
1837 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1838 most of the time, it *can* run out if it is given a pattern that contains a
1839 runaway infinite recursion. I updated the discussion in the pcrestack man
1840 page.
1841
1842 13. Now that we have gone to the x.xx style of version numbers, the minor
1843 version may start with zero. Using 08 or 09 is a bad idea because users
1844 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1845 interpreted as invalid octal numbers. I've updated the previous comment in
1846 configure.ac, and also added a check that gives an error if 08 or 09 are
1847 used.
1848
1849 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1850 causing partial matching to fail when the end of the subject matched \W
1851 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1852
1853 15. There were some discrepancies between the declarations in pcre_internal.h
1854 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1855 their definitions. The declarations used "const uschar *" and the
1856 definitions used USPTR. Even though USPTR is normally defined as "const
1857 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1858 reported that: "This difference in casting confuses some C++ compilers, for
1859 example, SunCC recognizes above declarations as different functions and
1860 generates broken code for hbpcre." I have changed the declarations to use
1861 USPTR.
1862
1863 16. GNU libtool is named differently on some systems. The autogen.sh script now
1864 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1865 (FreeBSD).
1866
1867 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1868 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1869 comment: "Figure out how to create a longlong from a string: strtoll and
1870 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1871 instance, but it only takes 2 args instead of 3!"
1872
1873 18. A subtle bug concerned with back references has been fixed by a change of
1874 specification, with a corresponding code fix. A pattern such as
1875 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1876 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1877 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1878 same bug. Such groups have to be quantified to be useful, or contained
1879 inside another quantified group. (If there's no repetition, the reference
1880 can never match.) The problem arises because, having left the group and
1881 moved on to the rest of the pattern, a later failure that backtracks into
1882 the group uses the captured value from the final iteration of the group
1883 rather than the correct earlier one. I have fixed this in PCRE by forcing
1884 any group that contains a reference to itself to be an atomic group; that
1885 is, there cannot be any backtracking into it once it has completed. This is
1886 similar to recursive and subroutine calls.
1887
1888
1889 Version 8.00 19-Oct-09
1890 ----------------------
1891
1892 1. The table for translating pcre_compile() error codes into POSIX error codes
1893 was out-of-date, and there was no check on the pcre_compile() error code
1894 being within the table. This could lead to an OK return being given in
1895 error.
1896
1897 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1898 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1899 in a Windows environment.
1900
1901 3. The pcregrep --count option prints the count for each file even when it is
1902 zero, as does GNU grep. However, pcregrep was also printing all files when
1903 --files-with-matches was added. Now, when both options are given, it prints
1904 counts only for those files that have at least one match. (GNU grep just
1905 prints the file name in this circumstance, but including the count seems
1906 more useful - otherwise, why use --count?) Also ensured that the
1907 combination -clh just lists non-zero counts, with no names.
1908
1909 4. The long form of the pcregrep -F option was incorrectly implemented as
1910 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1911 but it seems right to fix it, and I didn't think it was worth preserving
1912 the old behaviour.
1913
1914 5. The command line items --regex=pattern and --regexp=pattern were not
1915 recognized by pcregrep, which required --regex pattern or --regexp pattern
1916 (with a space rather than an '='). The man page documented the '=' forms,
1917 which are compatible with GNU grep; these now work.
1918
1919 6. No libpcreposix.pc file was created for pkg-config; there was just
1920 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1921
1922 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1923 when UCP support is not needed, by modifying the Python script that
1924 generates it from Unicode data files. This should not matter if the module
1925 is correctly used as a library, but I received one complaint about 50K of
1926 unwanted data. My guess is that the person linked everything into his
1927 program rather than using a library. Anyway, it does no harm.
1928
1929 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1930 was a minimum greater than 1 for a wide character in a possessive
1931 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1932 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1933 character. Chaos in the form of incorrect output or a compiling loop could
1934 result.
1935
1936 9. The restrictions on what a pattern can contain when partial matching is
1937 requested for pcre_exec() have been removed. All patterns can now be
1938 partially matched by this function. In addition, if there are at least two
1939 slots in the offset vector, the offset of the earliest inspected character
1940 for the match and the offset of the end of the subject are set in them when
1941 PCRE_ERROR_PARTIAL is returned.
1942
1943 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1944 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1945 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1946 and may be more useful for multi-segment matching.
1947
1948 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1949 used to be given if ever the end of the subject was reached; now it is
1950 given only if matching could not proceed because another character was
1951 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1952 string "Z", which now yields "no match" instead of "partial match". In the
1953 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1954 final character ended with (*FAIL).
1955
1956 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1957 if the pattern had a "must contain" character that was already found in the
1958 earlier partial match, unless partial matching was again requested. For
1959 example, with the pattern /dog.(body)?/, the "must contain" character is
1960 "g". If the first part-match was for the string "dog", restarting with
1961 "sbody" failed. This bug has been fixed.
1962
1963 13. The string returned by pcre_dfa_exec() after a partial match has been
1964 changed so that it starts at the first inspected character rather than the
1965 first character of the match. This makes a difference only if the pattern
1966 starts with a lookbehind assertion or \b or \B (\K is not supported by
1967 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1968 matching functions compatible, and I think it's the right thing to do.
1969
1970 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1971 so that the demonstration program is easily available in environments where
1972 PCRE has not been installed from source.
1973
1974 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1975 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1976 library.
1977
1978 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1979 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1980 is not the first non-POSIX option to be added. Clearly some people find
1981 these options useful.
1982
1983 17. If a caller to the POSIX matching function regexec() passes a non-zero
1984 value for nmatch with a NULL value for pmatch, the value of
1985 nmatch is forced to zero.
1986
1987 18. RunGrepTest did not have a test for the availability of the -u option of
1988 the diff command, as RunTest does. It now checks in the same way as
1989 RunTest, and also checks for the -b option.
1990
1991 19. If an odd number of negated classes containing just a single character
1992 interposed, within parentheses, between a forward reference to a named
1993 subpattern and the definition of the subpattern, compilation crashed with
1994 an internal error, complaining that it could not find the referenced
1995 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1996 [The bug was that it was starting one character too far in when skipping
1997 over the character class, thus treating the ] as data rather than
1998 terminating the class. This meant it could skip too much.]
1999
2000 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2001 /g option in pcretest when the pattern contains \K, which makes it possible
2002 to have an empty string match not at the start, even when the pattern is
2003 anchored. Updated pcretest and pcredemo to use this option.
2004
2005 21. If the maximum number of capturing subpatterns in a recursion was greater
2006 than the maximum at the outer level, the higher number was returned, but
2007 with unset values at the outer level. The correct (outer level) value is
2008 now given.
2009
2010 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2011 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2012 make it do so. The string so far is captured, making this feature
2013 compatible with Perl.
2014
2015 23. The tests have been re-organized, adding tests 11 and 12, to make it
2016 possible to check the Perl 5.10 features against Perl 5.10.
2017
2018 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2019 pattern matches a fixed length string. PCRE did not allow this; now it
2020 does. Neither allows recursion.
2021
2022 25. I finally figured out how to implement a request to provide the minimum
2023 length of subject string that was needed in order to match a given pattern.
2024 (It was back references and recursion that I had previously got hung up
2025 on.) This code has now been added to pcre_study(); it finds a lower bound
2026 to the length of subject needed. It is not necessarily the greatest lower
2027 bound, but using it to avoid searching strings that are too short does give
2028 some useful speed-ups. The value is available to calling programs via
2029 pcre_fullinfo().
2030
2031 26. While implementing 25, I discovered to my embarrassment that pcretest had
2032 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2033 study optimizations had never been tested with that matching function.
2034 Oops. What is worse, even when it was passed study data, there was a bug in
2035 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2036 were also very few tests of studied patterns with pcre_dfa_exec().
2037
2038 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2039 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2040 on the other side of the coin, they are no longer allowed to have different
2041 names, because these cannot be distinguished in PCRE, and this has caused
2042 confusion. (This is a difference from Perl.)
2043
2044 28. When duplicate subpattern names are present (necessarily with different
2045 numbers, as required by 27 above), and a test is made by name in a
2046 conditional pattern, either for a subpattern having been matched, or for
2047 recursion in such a pattern, all the associated numbered subpatterns are
2048 tested, and the overall condition is true if the condition is true for any
2049 one of them. This is the way Perl works, and is also more like the way
2050 testing by number works.
2051
2052
2053 Version 7.9 11-Apr-09
2054 ---------------------
2055
2056 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2057 (pcretest), all targets were linked against these libraries. This included
2058 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2059 libraries. This caused unwanted dependencies to be created. This problem
2060 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2061 pcretest is linked with readline.
2062
2063 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2064 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2065 moved outside it again, because FALSE and TRUE are already defined in AIX,
2066 but BOOL is not.
2067
2068 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2069 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2070
2071 4. The pcregrep documentation said spaces were inserted as well as colons (or
2072 hyphens) following file names and line numbers when outputting matching
2073 lines. This is not true; no spaces are inserted. I have also clarified the
2074 wording for the --colour (or --color) option.
2075
2076 5. In pcregrep, when --colour was used with -o, the list of matching strings
2077 was not coloured; this is different to GNU grep, so I have changed it to be
2078 the same.
2079
2080 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2081 each matching line was coloured. Now it goes on to look for further matches
2082 of any of the test patterns, which is the same behaviour as GNU grep.
2083
2084 7. A pattern that could match an empty string could cause pcregrep to loop; it
2085 doesn't make sense to accept an empty string match in pcregrep, so I have
2086 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2087 seems to be how GNU grep behaves. [But see later change 40 for release
2088 8.33.]
2089
2090 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2091 start or after a newline", because the conditional assertion was not being
2092 correctly handled. The rule now is that both the assertion and what follows
2093 in the first alternative must satisfy the test.
2094
2095 9. If auto-callout was enabled in a pattern with a conditional group whose
2096 condition was an assertion, PCRE could crash during matching, both with
2097 pcre_exec() and pcre_dfa_exec().
2098
2099 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2100 used for matching.
2101
2102 11. Unicode property support in character classes was not working for
2103 characters (bytes) greater than 127 when not in UTF-8 mode.
2104
2105 12. Added the -M command line option to pcretest.
2106
2107 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2108
2109 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2110
2111 16. Added comments and documentation about mis-use of no_arg in the C++
2112 wrapper.
2113
2114 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2115 from Martin Jerabek that uses macro names for all relevant character and
2116 string constants.
2117
2118 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2119 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2120 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2121 these, but not everybody uses configure.
2122
2123 19. A conditional group that had only one branch was not being correctly
2124 recognized as an item that could match an empty string. This meant that an
2125 enclosing group might also not be so recognized, causing infinite looping
2126 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2127 with the subject "ab", where knowledge that the repeated group can match
2128 nothing is needed in order to break the loop.
2129
2130 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2131 exec(), but without supplying a callout function, matching went wrong.
2132
2133 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2134 leak if the size of the offset vector was greater than 30. When the vector
2135 is smaller, the saved offsets during recursion go onto a local stack
2136 vector, but for larger vectors malloc() is used. It was failing to free
2137 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2138 error, in fact).
2139
2140 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2141 heapframe that is used only when UTF-8 support is enabled. This caused no
2142 problem, but was untidy.
2143
2144 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2145 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2146 included within another project.
2147
2148 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2149 slightly modified by me:
2150
2151 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2152 not building pcregrep.
2153
2154 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2155 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2156
2157 25. Forward references, both numeric and by name, in patterns that made use of
2158 duplicate group numbers, could behave incorrectly or give incorrect errors,
2159 because when scanning forward to find the reference group, PCRE was not
2160 taking into account the duplicate group numbers. A pattern such as
2161 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2162
2163 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2164 the feature of a custom pointer more persuasive (as requested by a user).
2165
2166 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2167 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2168 pre-defined.
2169
2170 28. Added support for (*UTF8) at the start of a pattern.
2171
2172 29. Arrange for flags added by the "release type" setting in CMake to be shown
2173 in the configuration summary.
2174
2175
2176 Version 7.8 05-Sep-08
2177 ---------------------
2178
2179 1. Replaced UCP searching code with optimized version as implemented for Ad
2180 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2181 stage table and inline lookup instead of a function, giving speed ups of 2
2182 to 5 times on some simple patterns that I tested. Permission was given to
2183 distribute the MultiStage2.py script that generates the tables (it's not in
2184 the tarball, but is in the Subversion repository).
2185
2186 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2187 scripts.
2188
2189 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2190 a group with a zero qualifier. The result of the study could be incorrect,
2191 or the function might crash, depending on the pattern.
2192
2193 4. Caseless matching was not working for non-ASCII characters in back
2194 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2195 It now works when Unicode Property Support is available.
2196
2197 5. In pcretest, an escape such as \x{de} in the data was always generating
2198 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2199 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2200 truncation.
2201
2202 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2203
2204 7. Added two (int) casts to pcregrep when printing the difference of two
2205 pointers, in case they are 64-bit values.
2206
2207 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2208 test 2 if it fails.
2209
2210 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2211 and a #define of that name to empty if it is not externally set. This is to
2212 allow users of MSVC to set it if necessary.
2213
2214 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2215 the convenience functions in the pcre_get.c source file.
2216
2217 11. An option change at the start of a pattern that had top-level alternatives
2218 could cause overwriting and/or a crash. This command provoked a crash in
2219 some environments:
2220
2221 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2222
2223 This potential security problem was recorded as CVE-2008-2371.
2224
2225 12. For a pattern where the match had to start at the beginning or immediately
2226 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2227 pcre_dfa_exec() could read past the end of the passed subject if there was
2228 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2229 pcretest so that it places the subject at the end of its malloc-ed buffer.
2230
2231 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2232 exec() might read past the end of the data buffer in UTF-8 mode.
2233
2234 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2235 the data contained the byte 0x85 as part of a UTF-8 character within its
2236 first line. This applied both to normal and DFA matching.
2237
2238 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2239 /^[^d]*?$/8 failed to match "abc".
2240
2241 16. Added a missing copyright notice to pcrecpp_internal.h.
2242
2243 17. Make it more clear in the documentation that values returned from
2244 pcre_exec() in ovector are byte offsets, not character counts.
2245
2246 18. Tidied a few places to stop certain compilers from issuing warnings.
2247
2248 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2249 supplied by Stefan Weber. I made a further small update for 7.8 because
2250 there is a change of source arrangements: the pcre_searchfuncs.c module is
2251 replaced by pcre_ucd.c.
2252
2253
2254 Version 7.7 07-May-08
2255 ---------------------
2256
2257 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2258 a string to a long long, pretend we don't even have a long long." This is
2259 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2260
2261 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2262 pre-7.6 versions, which defined a global no_arg variable instead of putting
2263 it in the RE class. (See also #8 below.)
2264
2265 3. Remove a line of dead code, identified by coverity and reported by Nuno
2266 Lopes.
2267
2268 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2269
2270 (1) The include/exclude patterns were being applied to the whole pathnames
2271 of files, instead of just to the final components.
2272
2273 (2) If there was more than one level of directory, the subdirectories were
2274 skipped unless they satisfied the include/exclude conditions. This is
2275 inconsistent with GNU grep (and could even be seen as contrary to the
2276 pcregrep specification - which I improved to make it absolutely clear).
2277 The action now is always to scan all levels of directory, and just
2278 apply the include/exclude patterns to regular files.
2279
2280 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2281 --exclude_dir in the tests to avoid scanning .svn directories.
2282
2283 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2284 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2285 doesn't support NULs in patterns.
2286
2287 7. Added some missing "const"s to declarations of static tables in
2288 pcre_compile.c and pcre_dfa_exec.c.
2289
2290 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2291 caused by fix #2 above. (Subsequently also a second patch to fix the
2292 first patch. And a third patch - this was a messy problem.)
2293
2294 9. Applied Craig's patch to remove the use of push_back().
2295
2296 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2297 matching function regexec().
2298
2299 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2300 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2301 references. PCRE supports relative numbers with this syntax (I don't think
2302 Oniguruma does).
2303
2304 12. Previously, a group with a zero repeat such as (...){0} was completely
2305 omitted from the compiled regex. However, this means that if the group
2306 was called as a subroutine from elsewhere in the pattern, things went wrong
2307 (an internal error was given). Such groups are now left in the compiled
2308 pattern, with a new opcode that causes them to be skipped at execution
2309 time.
2310
2311 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2312 to the way PCRE behaves:
2313
2314 (a) A lone ] character is dis-allowed (Perl treats it as data).
2315
2316 (b) A back reference to an unmatched subpattern matches an empty string
2317 (Perl fails the current match path).
2318
2319 (c) A data ] in a character class must be notated as \] because if the
2320 first data character in a class is ], it defines an empty class. (In
2321 Perl it is not possible to have an empty class.) The empty class []
2322 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2323 The negative empty class [^] matches any one character, independently
2324 of the DOTALL setting.
2325
2326 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2327 non-existent subpattern following a character class starting with ']' and
2328 containing () gave an internal compiling error instead of "reference to
2329 non-existent subpattern". Fortunately, when the pattern did exist, the
2330 compiled code was correct. (When scanning forwards to check for the
2331 existence of the subpattern, it was treating the data ']' as terminating
2332 the class, so got the count wrong. When actually compiling, the reference
2333 was subsequently set up correctly.)
2334
2335 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2336 it was being rejected as not supported by pcre_dfa_exec(), even though
2337 other assertions are supported. I have made pcre_dfa_exec() support
2338 (*FAIL).
2339
2340 16. The implementation of 13c above involved the invention of a new opcode,
2341 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2342 cannot be changed at match time, I realized I could make a small
2343 improvement to matching performance by compiling OP_ALLANY instead of
2344 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2345 on the OP_ANY path.
2346
2347 17. Compiling pcretest on Windows with readline support failed without the
2348 following two fixes: (1) Make the unistd.h include conditional on
2349 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2350
2351 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2352 ncurses library to be included for pcretest when ReadLine support is
2353 requested, but also to allow for it to be overridden. This patch came from
2354 Daniel Bergström.
2355
2356 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2357 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2358 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2359 spotting this.
2360
2361
2362 Version 7.6 28-Jan-08
2363 ---------------------
2364
2365 1. A character class containing a very large number of characters with
2366 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2367 overflow.
2368
2369 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2370 HAVE_LONG_LONG is not defined.
2371
2372 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2373 bring them up to date and include new features. This patch includes:
2374
2375 - Fixed PH's badly added libz and libbz2 support.
2376 - Fixed a problem with static linking.
2377 - Added pcredemo. [But later removed - see 7 below.]
2378 - Fixed dftables problem and added an option.
2379 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2380 HAVE_LONG_LONG.
2381 - Added readline support for pcretest.
2382 - Added an listing of the option settings after cmake has run.
2383
2384 4. A user submitted a patch to Makefile that makes it easy to create
2385 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2386 Makefile.am that cause it to include this special target, without
2387 affecting anything else. Note that the same mingw target plus all
2388 the other distribution libraries and programs are now supported
2389 when configuring with CMake (see 6 below) instead of with
2390 Configure/Make.
2391
2392 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2393 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2394 exported in the Windows port". It has not yet been confirmed that the patch
2395 solves the problem, but it does no harm.
2396
2397 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2398 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2399 with CMake, and also correct the comment about stack recursion.
2400
2401 7. Remove the automatic building of pcredemo from the ./configure system and
2402 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2403 of a program that users should build themselves after PCRE is installed, so
2404 building it automatically is not really right. What is more, it gave
2405 trouble in some build environments.
2406
2407 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2408
2409
2410 Version 7.5 10-Jan-08
2411 ---------------------
2412
2413 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2414 values in parens when parsing an RE using the C++ wrapper."
2415
2416 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2417 Characters greater than 255 were excluded from the class instead of being
2418 included.
2419
2420 3. The same bug as (2) above applied to negated POSIX classes such as
2421 [:^space:].
2422
2423 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2424 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2425 I have changed it.
2426
2427 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2428 first named subpattern) and a construct such as (?&a) would reference the
2429 first named subpattern whose name started with "a" (in other words, the
2430 length check was missing). Both these problems are fixed. "Subpattern name
2431 expected" is now given for (?&) (a zero-length name), and this patch also
2432 makes it give the same error for \k'' (previously it complained that that
2433 was a reference to a non-existent subpattern).
2434
2435 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2436 this is right because (?- can be followed by option settings as well as by
2437 digits. I have, however, made the messages clearer.
2438
2439 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2440 than the number used in the conditional) now cause a compile-time error.
2441 This is actually not compatible with Perl, which accepts such patterns, but
2442 treats the conditional as always being FALSE (as PCRE used to), but it
2443 seems to me that giving a diagnostic is better.
2444
2445 8. Change "alphameric" to the more common word "alphanumeric" in comments
2446 and messages.
2447
2448 9. Fix two occurrences of "backslash" in comments that should have been
2449 "backspace".
2450
2451 10. Remove two redundant lines of code that can never be obeyed (their function
2452 was moved elsewhere).
2453
2454 11. The program that makes PCRE's Unicode character property table had a bug
2455 which caused it to generate incorrect table entries for sequences of
2456 characters that have the same character type, but are in different scripts.
2457 It amalgamated them into a single range, with the script of the first of
2458 them. In other words, some characters were in the wrong script. There were
2459 thirteen such cases, affecting characters in the following ranges:
2460
2461 U+002b0 - U+002c1
2462 U+0060c - U+0060d
2463 U+0061e - U+00612
2464 U+0064b - U+0065e
2465 U+0074d - U+0076d
2466 U+01800 - U+01805
2467 U+01d00 - U+01d77
2468 U+01d9b - U+01dbf
2469 U+0200b - U+0200f
2470 U+030fc - U+030fe
2471 U+03260 - U+0327f
2472 U+0fb46 - U+0fbb1
2473 U+10450 - U+1049d
2474
2475 12. The -o option (show only the matching part of a line) for pcregrep was not
2476 compatible with GNU grep in that, if there was more than one match in a
2477 line, it showed only the first of them. It now behaves in the same way as
2478 GNU grep.
2479
2480 13. If the -o and -v options were combined for pcregrep, it printed a blank
2481 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2482 does the same. The return code can be used to tell if there were any
2483 non-matching lines.
2484
2485 14. Added --file-offsets and --line-offsets to pcregrep.
2486
2487 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2488 infinitely looping recursion. The bug was that positive lookaheads were not
2489 being skipped when checking for a possible empty match (negative lookaheads
2490 and both kinds of lookbehind were skipped).
2491
2492 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2493 inclusion of <windows.h> to before rather than after the definition of
2494 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2495
2496 17. Specifying a possessive quantifier with a specific limit for a Unicode
2497 character property caused pcre_compile() to compile bad code, which led at
2498 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2499 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2500 caused the error; without that there was no problem.
2501
2502 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2503
2504 19. Added --enable-pcretest-libreadline.
2505
2506 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2507 RE::GlobalReplace(). As a result, the number of replacements returned was
2508 double what it should be. I removed one of the increments, but Craig sent a
2509 later patch that removed the other one (the right fix) and added unit tests
2510 that check the return values (which was not done before).
2511
2512 21. Several CMake things:
2513
2514 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2515 the names libpcre and libpcreposix, not just pcre and pcreposix.
2516
2517 (2) The above change means that pcretest and pcregrep are now correctly
2518 linked with the newly-built libraries, not previously installed ones.
2519
2520 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2521
2522 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2523 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2524 UTF-8 newline character). The key issue is that the pattern starts .*;
2525 this means that the match must be either at the beginning, or after a
2526 newline. The bug was in the code for advancing after a failed match and
2527 checking that the new position followed a newline. It was not taking
2528 account of UTF-8 characters correctly.
2529
2530 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2531 character classes. PCRE was not treating the sequence [:...:] as a
2532 character class unless the ... were all letters. Perl, however, seems to
2533 allow any characters between [: and :], though of course it rejects as
2534 unknown any "names" that contain non-letters, because all the known class
2535 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2536 for example, whereas PCRE did not - it did not recognize a POSIX character
2537 class. This seemed a bit dangerous, so the code has been changed to be
2538 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2539 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2540 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2541 Perl does, and where it didn't before.
2542
2543 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2544 Windows environments %n is disabled by default.
2545
2546
2547 Version 7.4 21-Sep-07
2548 ---------------------
2549
2550 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2551 means that a class such as [\s] counted as "explicit reference to CR or
2552 LF". That isn't really right - the whole point of the change was to try to
2553 help when there was an actual mention of one of the two characters. So now
2554 the change happens only if \r or \n (or a literal CR or LF) character is
2555 encountered.
2556
2557 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2558 of both had grown to the point where there were only 3 bits left.
2559 Fortunately, there was spare space in the data structure, and so I have
2560 moved the internal flags into a new 16-bit field to free up more option
2561 bits.
2562
2563 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2564 but did not set the internal JCHANGED flag - either of these is enough to
2565 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2566 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2567 start sets both bits.
2568
2569 4. Added options (at build time, compile time, exec time) to change \R from
2570 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2571
2572 5. doc/pcresyntax.html was missing from the distribution.
2573
2574 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2575 compatibility, even though it is no longer used.
2576
2577 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2578 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2579 windows.h file is present (where different names are used). [This was
2580 reversed later after testing - see 16 below.]
2581
2582 8. Changed all #include <config.h> to #include "config.h". There were also
2583 some further <pcre.h> cases that I changed to "pcre.h".
2584
2585 9. When pcregrep was used with the --colour option, it missed the line ending
2586 sequence off the lines that it output.
2587
2588 10. It was pointed out to me that arrays of string pointers cause lots of
2589 relocations when a shared library is dynamically loaded. A technique of
2590 using a single long string with a table of offsets can drastically reduce
2591 these. I have refactored PCRE in four places to do this. The result is
2592 dramatic:
2593
2594 Originally: 290
2595 After changing UCP table: 187
2596 After changing error message table: 43
2597 After changing table of "verbs" 36
2598 After changing table of Posix names 22
2599
2600 Thanks to the folks working on Gregex for glib for this insight.
2601
2602 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2603 unicode-properties was also set.
2604
2605 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2606
2607 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2608 checked only for CRLF.
2609
2610 14. Added casts to pcretest.c to avoid compiler warnings.
2611
2612 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2613
2614 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2615 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2616 entirely. This removes changes made in 7 above.
2617
2618 17. The CMake files have been updated, and there is now more information about
2619 building with CMake in the NON-UNIX-USE document.
2620
2621
2622 Version 7.3 28-Aug-07
2623 ---------------------
2624
2625 1. In the rejigging of the build system that eventually resulted in 7.1, the
2626 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2627 brackets there is not right, since it causes compilers to look for an
2628 installed pcre.h, not the version that is in the source that is being
2629 compiled (which of course may be different). I have changed it back to:
2630
2631 #include "pcre.h"
2632
2633 I have a vague recollection that the change was concerned with compiling in
2634 different directories, but in the new build system, that is taken care of
2635 by the VPATH setting the Makefile.
2636
2637 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2638 when the subject happened to end in the byte 0x85 (e.g. if the last
2639 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2640 characters but of course it shouldn't be taken as a newline when it is part
2641 of another character. The bug was that, for an unlimited repeat of . in
2642 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2643 characters when looking for a newline.
2644
2645 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2646
2647 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2648 in debug output.
2649
2650 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2651 long printing in the pcrecpp unittest when running under MinGW.
2652
2653 6. ESC_K was left out of the EBCDIC table.
2654
2655 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2656 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2657 limit also applies to "virtual nesting" when a pattern is recursive, and in
2658 this case 1000 isn't so big. I have been able to remove this limit at the
2659 expense of backing off one optimization in certain circumstances. Normally,
2660 when pcre_exec() would call its internal match() function recursively and
2661 immediately return the result unconditionally, it uses a "tail recursion"
2662 feature to save stack. However, when a subpattern that can match an empty
2663 string has an unlimited repetition quantifier, it no longer makes this
2664 optimization. That gives it a stack frame in which to save the data for
2665 checking that an empty string has been matched. Previously this was taken
2666 from the 1000-entry workspace that had been reserved. So now there is no
2667 explicit limit, but more stack is used.
2668
2669 8. Applied Daniel's patches to solve problems with the import/export magic
2670 syntax that is required for Windows, and which was going wrong for the
2671 pcreposix and pcrecpp parts of the library. These were overlooked when this
2672 problem was solved for the main library.
2673
2674 9. There were some crude static tests to avoid integer overflow when computing
2675 the size of patterns that contain repeated groups with explicit upper
2676 limits. As the maximum quantifier is 65535, the maximum group length was
2677 set at 30,000 so that the product of these two numbers did not overflow a
2678 32-bit integer. However, it turns out that people want to use groups that
2679 are longer than 30,000 bytes (though not repeat them that many times).
2680 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2681 made it possible to implement the integer overflow checks in a much more
2682 dynamic way, which I have now done. The artificial limitation on group
2683 length has been removed - we now have only the limit on the total length of
2684 the compiled pattern, which depends on the LINK_SIZE setting.
2685
2686 10. Fixed a bug in the documentation for get/copy named substring when
2687 duplicate names are permitted. If none of the named substrings are set, the
2688 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2689 empty string.
2690
2691 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2692 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2693 because the ] is interpreted as the first data character and the
2694 terminating ] is not found. PCRE has been made compatible with Perl in this
2695 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2696 cause memory overwriting.
2697
2698 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2699 string has been matched (to stop an infinite loop). It was not recognizing
2700 a conditional subpattern that could match an empty string if that
2701 subpattern was within another subpattern. For example, it looped when
2702 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2703 condition was not nested. This bug has been fixed.
2704
2705 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2706 past the start of the subject in the presence of bytes with the top bit
2707 set, for example "\x8aBCD".
2708
2709 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2710 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2711
2712 14. Optimized (?!) to (*FAIL).
2713
2714 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2715 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2716 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2717 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2718 does: it's just the validity check that is more restrictive.
2719
2720 16. Inserted checks for integer overflows during escape sequence (backslash)
2721 processing, and also fixed erroneous offset values for syntax errors during
2722 backslash processing.
2723
2724 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2725 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2726
2727 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2728 caused an overrun.
2729
2730 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2731 something other than just ASCII characters) inside a group that had an
2732 unlimited repeat caused a loop at compile time (while checking to see
2733 whether the group could match an empty string).
2734
2735 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2736 [\P{Any}] did so. (Error in the code for printing property names.)
2737
2738 21. An orphan \E inside a character class could cause a crash.
2739
2740 22. A repeated capturing bracket such as (A)? could cause a wild memory
2741 reference during compilation.
2742
2743 23. There are several functions in pcre_compile() that scan along a compiled
2744 expression for various reasons (e.g. to see if it's fixed length for look
2745 behind). There were bugs in these functions when a repeated \p or \P was
2746 present in the pattern. These operators have additional parameters compared
2747 with \d, etc, and these were not being taken into account when moving along
2748 the compiled data. Specifically:
2749
2750 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2751 length.
2752
2753 (b) An item such as \pL+ within a repeated group could cause crashes or
2754 loops.
2755
2756 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2757 "reference to non-existent subpattern" error.
2758
2759 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2760
2761 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2762 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2763
2764 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2765
2766 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2767 character were causing crashes (broken optimization).
2768
2769 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2770 \p or \P) caused a compile-time loop.
2771
2772 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2773 break. For example, the unstudied pattern [\r\n]A does not match the string
2774 "\r\nA" because change 7.0/46 below moves the current point on by two
2775 characters after failing to match at the start. However, the pattern \nA
2776 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2777 the same is true. There doesn't seem any very clean way out of this, but
2778 what I have chosen to do makes the common cases work: PCRE now takes note
2779 of whether there can be an explicit match for \r or \n anywhere in the
2780 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2781 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2782 pattern has explicit CR or LF references.
2783
2784 29. Added (*CR) etc for changing newline setting at start of pattern.
2785
2786
2787 Version 7.2 19-Jun-07
2788 ---------------------
2789
2790 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2791 which is apparently normally available under Windows.
2792
2793 2. Re-jig the pcregrep tests with different newline settings in an attempt
2794 to make them independent of the local environment's newline setting.
2795
2796 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2797
2798 4. Some of the "internals" tests were previously cut out when the link size
2799 was not 2, because the output contained actual offsets. The recent new
2800 "Z" feature of pcretest means that these can be cut out, making the tests
2801 usable with all link sizes.
2802
2803 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2804 stack recursion. This gives a massive performance boost under BSD, but just
2805 a small improvement under Linux. However, it saves one field in the frame
2806 in all cases.
2807
2808 6. Added more features from the forthcoming Perl 5.10:
2809
2810 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2811 recursion call. It refers to the nth most recently opened parentheses.
2812
2813 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2814 to be opened parentheses.
2815
2816 (c) Conditions that refer to capturing parentheses can be specified
2817 relatively, for example, (?(-2)... or (?(+3)...
2818
2819 (d) \K resets the start of the current match so that everything before
2820 is not part of it.
2821
2822 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2823
2824 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2825 reference syntax.
2826
2827 (g) (?| introduces a group in which the numbering of parentheses in each
2828 alternative starts with the same number.
2829
2830 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2831
2832 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2833 PCRE_INFO_JCHANGED.
2834
2835 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2836 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2837 for detecting groups that can match an empty string.
2838
2839 9. A pattern with a very large number of alternatives (more than several
2840 hundred) was running out of internal workspace during the pre-compile
2841 phase, where pcre_compile() figures out how much memory will be needed. A
2842 bit of new cunning has reduced the workspace needed for groups with
2843 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2844 workspace instead of running out of the 4096 that are available.
2845
2846 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2847
2848 11. Applied patch from Google to remove an optimization that didn't quite work.
2849 The report of the bug said:
2850
2851 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2852 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2853 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2854
2855 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2856 it matched the wrong number of bytes.
2857
2858
2859 Version 7.1 24-Apr-07
2860 ---------------------
2861
2862 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2863 that is more "standard", making use of automake and other Autotools. There
2864 is some re-arrangement of the files and adjustment of comments consequent
2865 on this.
2866
2867 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2868 for recursive directory scanning broke on some systems because the files
2869 are not scanned in any specific order and on different systems the order
2870 was different. A call to "sort" has been inserted into RunGrepTest for the
2871 approprate test as a short-term fix. In the longer term there may be an
2872 alternative.
2873
2874 3. I had an email from Eric Raymond about problems translating some of PCRE's
2875 man pages to HTML (despite the fact that I distribute HTML pages, some
2876 people do their own conversions for various reasons). The problems
2877 concerned the use of low-level troff macros .br and .in. I have therefore
2878 removed all such uses from the man pages (some were redundant, some could
2879 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2880 HTML has been updated to handle .nf/.fi and to complain if it encounters
2881 .br or .in.
2882
2883 4. Updated comments in configure.ac that get placed in config.h.in and also
2884 arranged for config.h to be included in the distribution, with the name
2885 config.h.generic, for the benefit of those who have to compile without
2886 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2887
2888 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2889 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2890 makevp.bat for the current PCRE, using the additional files
2891 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2892
2893 6. A Windows user reported a minor discrepancy with test 2, which turned out
2894 to be caused by a trailing space on an input line that had got lost in his
2895 copy. The trailing space was an accident, so I've just removed it.
2896
2897 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2898 that is needed.
2899
2900 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2901 as "const" (a) because they are and (b) because it helps the PHP
2902 maintainers who have recently made a script to detect big data structures
2903 in the php code that should be moved to the .rodata section. I remembered
2904 to update Builducptable as well, so it won't revert if ucptable.h is ever
2905 re-created.
2906
2907 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2908 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2909 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2910 support is not required. This saves 1.5-2K of code, which is important in
2911 some applications.
2912
2913 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2914 so as not to refer to the tables, even though these functions will never be
2915 called when UTF-8 support is disabled. Otherwise there are problems with a
2916 shared library.
2917
2918 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2919
2920 (a) It was defining its arguments as char * instead of void *.
2921
2922 (b) It was assuming that all moves were upwards in memory; this was true
2923 a long time ago when I wrote it, but is no longer the case.
2924
2925 The emulated memove() is provided for those environments that have neither
2926 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2927 is clearly not the case, as these two bugs were recently reported.
2928
2929 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2930 and Detrail to create the HTML documentation, the .txt form of the man
2931 pages, and it removes trailing spaces from listed files. It also creates
2932 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2933 case, it wraps all the #defines with #ifndefs. This script should be run
2934 before "make dist".
2935
2936 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2937 with Unicode property support.
2938
2939 (a) For a maximizing quantifier, if the two different cases of the
2940 character were of different lengths in their UTF-8 codings (there are
2941 some cases like this - I found 11), and the matching function had to
2942 back up over a mixture of the two cases, it incorrectly assumed they
2943 were both the same length.
2944
2945 (b) When PCRE was configured to use the heap rather than the stack for
2946 recursion during matching, it was not correctly preserving the data for
2947 the other case of a UTF-8 character when checking ahead for a match
2948 while processing a minimizing repeat. If the check also involved
2949 matching a wide character, but failed, corruption could cause an
2950 erroneous result when trying to check for a repeat of the original
2951 character.
2952
2953 13. Some tidying changes to the testing mechanism:
2954
2955 (a) The RunTest script now detects the internal link size and whether there
2956 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2957 values substituted by "configure". (The RunGrepTest script already did
2958 this for UTF-8.) The configure.ac script no longer substitutes the
2959 relevant variables.
2960
2961 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2962 with length and offset values. This means that the output is different
2963 for different internal link sizes. Test 2 is skipped for link sizes
2964 other than 2 because of this, bypassing the problem. Unfortunately,
2965 there was also a test in test 3 (the locale tests) that used /B and
2966 failed for link sizes other than 2. Rather than cut the whole test out,
2967 I have added a new /Z option to pcretest that replaces the length and
2968 offset values with spaces. This is now used to make test 3 independent
2969 of link size. (Test 2 will be tidied up later.)
2970
2971 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2972 segmentation fault instead of returning the appropriate error message.
2973
2974 15. In multiline mode when the newline sequence was set to "any", the pattern
2975 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2976 This doesn't seem right; it now treats the CRLF combination as the line
2977 ending, and so does not match in that case. It's only a pattern such as ^$
2978 that would hit this one: something like ^ABC$ would have failed after \r
2979 and then tried again after \r\n.
2980
2981 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2982 in an attempt to make files that differ only in their line terminators
2983 compare equal. This works on Linux.
2984
2985 17. Under certain error circumstances pcregrep might try to free random memory
2986 as it exited. This is now fixed, thanks to valgrind.
2987
2988 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2989 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2990 was because its rules for how to advance for /g after matching an empty
2991 string at the end of a line did not allow for this case. They now check for
2992 it specially.
2993
2994 20. pcretest is supposed to handle patterns and data of any length, by
2995 extending its buffers when necessary. It was getting this wrong when the
2996 buffer for a data line had to be extended.
2997
2998 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2999 CRLF as a newline sequence.
3000
3001 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3002 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3003 I have nevertheless tidied it up.
3004
3005 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3006
3007 24. Added a man page for pcre-config.
3008
3009
3010 Version 7.0 19-Dec-06
3011 ---------------------
3012
3013 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3014 moving to gcc 4.1.1.
3015
3016 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3017 sys/time.h, which is documented as needed for this function. It doesn't
3018 seem to matter on Linux, but it showed up on some releases of OS X.
3019
3020 3. It seems that there are systems where bytes whose values are greater than
3021 127 match isprint() in the "C" locale. The "C" locale should be the
3022 default when a C program starts up. In most systems, only ASCII printing
3023 characters match isprint(). This difference caused the output from pcretest
3024 to vary, making some of the tests fail. I have changed pcretest so that:
3025
3026 (a) When it is outputting text in the compiled version of a pattern, bytes
3027 other than 32-126 are always shown as hex escapes.
3028
3029 (b) When it is outputting text that is a matched part of a subject string,
3030 it does the same, unless a different locale has been set for the match
3031 (using the /L modifier). In this case, it uses isprint() to decide.
3032
3033 4. Fixed a major bug that caused incorrect computation of the amount of memory
3034 required for a compiled pattern when options that changed within the
3035 pattern affected the logic of the preliminary scan that determines the
3036 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3037 that the computed length was too small. The symptoms of this bug were
3038 either the PCRE error "internal error: code overflow" from pcre_compile(),
3039 or a glibc crash with a message such as "pcretest: free(): invalid next
3040 size (fast)". Examples of patterns that provoked this bug (shown in
3041 pcretest format) are:
3042
3043 /(?-x: )/x
3044 /(?x)(?-x: \s*#\s*)/
3045 /((?i)[\x{c0}])/8
3046 /(?i:[\x{c0}])/8
3047
3048 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3049 is now done differently.
3050
3051 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3052 wrapper classes; (b) implement a new function in the C++ scanner that is
3053 more efficient than the old way of doing things because it avoids levels of
3054 recursion in the regex matching; (c) add a paragraph to the documentation
3055 for the FullMatch() function.
3056
3057 6. The escape sequence \n was being treated as whatever was defined as
3058 "newline". Not only was this contrary to the documentation, which states
3059 that \n is character 10 (hex 0A), but it also went horribly wrong when
3060 "newline" was defined as CRLF. This has been fixed.
3061
3062 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3063 was being set to -1 for the "end of line" case (supposedly a value that no
3064 character can have). Though this value is never used (the check for end of
3065 line is "zero bytes in current character"), it caused compiler complaints.
3066 I've changed it to 0xffffffff.
3067
3068 8. In pcre_version.c, the version string was being built by a sequence of
3069 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3070 string (as it is for production releases) called a macro with an empty
3071 argument. The C standard says the result of this is undefined. The gcc
3072 compiler treats it as an empty string (which was what was wanted) but it is
3073 reported that Visual C gives an error. The source has been hacked around to
3074 avoid this problem.
3075
3076 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3077 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3078 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3079 of them did).
3080
3081 10. Originally, pcretest opened its input and output without "b"; then I was
3082 told that "b" was needed in some environments, so it was added for release
3083 5.0 to both the input and output. (It makes no difference on Unix-like
3084 systems.) Later I was told that it is wrong for the input on Windows. I've
3085 now abstracted the modes into two macros, to make it easier to fiddle with
3086 them, and removed "b" from the input mode under Windows.
3087
3088 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3089
3090 12. Added -help and --help to pcretest as an official way of being reminded
3091 of the options.
3092
3093 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3094 and pcrecpp.cc because they annoy compilers at high warning levels.
3095
3096 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3097
3098 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3099 scripts are not C programs :-) and which was not noticed because it works
3100 on Linux.
3101
3102 16. pcretest is supposed to handle any length of pattern and data line (as one
3103 line or as a continued sequence of lines) by extending its input buffer if
3104 necessary. This feature was broken for very long pattern lines, leading to
3105 a string of junk being passed to pcre_compile() if the pattern was longer
3106 than about 50K.
3107
3108 17. I have done a major re-factoring of the way pcre_compile() computes the
3109 amount of memory needed for a compiled pattern. Previously, there was code
3110 that made a preliminary scan of the pattern in order to do this. That was
3111 OK when PCRE was new, but as the facilities have expanded, it has become
3112 harder and harder to keep it in step with the real compile phase, and there
3113 have been a number of bugs (see for example, 4 above). I have now found a
3114 cunning way of running the real compile function in a "fake" mode that
3115 enables it to compute how much memory it would need, while actually only
3116 ever using a few hundred bytes of working memory and without too many
3117 tests of the mode. This should make future maintenance and development
3118 easier. A side effect of this work is that the limit of 200 on the nesting
3119 depth of parentheses has been removed (though this was never a serious
3120 limitation, I suspect). However, there is a downside: pcre_compile() now
3121 runs more slowly than before (30% or more, depending on the pattern). I
3122 hope this isn't a big issue. There is no effect on runtime performance.
3123
3124 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3125 newline (only possible for the last line of a file) and it was a
3126 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3127
3128 19. Added additional timing features to pcretest. (1) The -tm option now times
3129 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3130 separate command line item, by a number that specifies the number of
3131 repeats to use when timing. The default is 50000; this gives better
3132 precision, but takes uncomfortably long for very large patterns.
3133
3134 20. Extended pcre_study() to be more clever in cases where a branch of a
3135 subpattern has no definite first character. For example, (a*|b*)[cd] would
3136 previously give no result from pcre_study(). Now it recognizes that the
3137 first character must be a, b, c, or d.
3138
3139 21. There was an incorrect error "recursive call could loop indefinitely" if
3140 a subpattern (or the entire pattern) that was being tested for matching an
3141 empty string contained only one non-empty item after a nested subpattern.
3142 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3143 incorrectly, because the \d was being skipped in the check.
3144
3145 22. The pcretest program now has a new pattern option /B and a command line
3146 option -b, which is equivalent to adding /B to every pattern. This causes
3147 it to show the compiled bytecode, without the additional information that
3148 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3149 is the same as /B/I).
3150
3151 23. A new optimization is now able automatically to treat some sequences such
3152 as a*b as a*+b. More specifically, if something simple (such as a character
3153 or a simple class like \d) has an unlimited quantifier, and is followed by
3154 something that cannot possibly match the quantified thing, the quantifier
3155 is automatically "possessified".
3156
3157 24. A recursive reference to a subpattern whose number was greater than 39
3158 went wrong under certain circumstances in UTF-8 mode. This bug could also
3159 have affected the operation of pcre_study().
3160
3161 25. Realized that a little bit of performance could be had by replacing
3162 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3163
3164 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3165
3166 27. Possessive quantifiers such as a++ were previously implemented by turning
3167 them into atomic groups such as ($>a+). Now they have their own opcodes,
3168 which improves performance. This includes the automatically created ones
3169 from 23 above.
3170
3171 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3172 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3173 the first matched character to be a colon. This applied both to named and
3174 numbered groups.
3175
3176 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3177
3178 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3179 building PCRE on Windows easier, so I have included it in the distribution.
3180
3181 31. There is now a check in pcretest against a ridiculously large number being
3182 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3183 loop, the loop is abandoned.
3184
3185 32. Forward references to subpatterns in conditions such as (?(2)...) where
3186 subpattern 2 is defined later cause pcre_compile() to search forwards in
3187 the pattern for the relevant set of parentheses. This search went wrong
3188 when there were unescaped parentheses in a character class, parentheses
3189 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3190
3191 33. "Subroutine" calls and backreferences were previously restricted to
3192 referencing subpatterns earlier in the regex. This restriction has now
3193 been removed.
3194
3195 34. Added a number of extra features that are going to be in Perl 5.10. On the
3196 whole, these are just syntactic alternatives for features that PCRE had
3197 previously implemented using the Python syntax or my own invention. The
3198 other formats are all retained for compatibility.
3199
3200 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3201 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3202 also .NET compatible.
3203
3204 (b) A recursion or subroutine call to a named group can now be defined as
3205 (?&name) as well as (?P>name).
3206
3207 (c) A backreference to a named group can now be defined as \k<name> or
3208 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3209 5.10, are also .NET compatible.
3210
3211 (d) A conditional reference to a named group can now use the syntax
3212 (?(<name>) or (?('name') as well as (?(name).
3213
3214 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3215 groups (named and numbered) that are never evaluated inline, but can be
3216 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3217 is always false. There may be only one alternative in such a group.
3218
3219 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3220 as the simple (?(R). The condition is true only if the most recent
3221 recursion is that of the given number or name. It does not search out
3222 through the entire recursion stack.
3223
3224 (g) The escape \gN or \g{N} has been added, where N is a positive or
3225 negative number, specifying an absolute or relative reference.
3226
3227 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3228 some "unreachable code" warnings.
3229
3230 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3231 things, this adds five new scripts.
3232
3233 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3234 There were also incompatibilities regarding the handling of \Q..\E inside
3235 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3236 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3237
3238 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3239 matches an empty string, and forcibly breaks the loop. There were bugs in
3240 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3241 against aaaa the result was just "a" rather than "aaaa", for example. Two
3242 separate and independent bugs (that affected different cases) have been
3243 fixed.
3244
3245 39. Refactored the code to abolish the use of different opcodes for small
3246 capturing bracket numbers. This is a tidy that I avoided doing when I
3247 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3248 The new approach is not only tidier, it makes it possible to reduce the
3249 memory needed to fix the previous bug (38).
3250
3251 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3252 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3253 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3254 mode.
3255
3256 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3257 report.
3258
3259 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3260 copy construction and assignment in the C++ wrapper.
3261
3262 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3263 couple of bugs that could have given wrong results in the "--newline=crlf"
3264 case.
3265
3266 44. Added a number of casts and did some reorganization of signed/unsigned int
3267 variables following suggestions from Dair Grant. Also renamed the variable
3268 "this" as "item" because it is a C++ keyword.
3269
3270 45. Arranged for dftables to add
3271
3272 #include "pcre_internal.h"
3273
3274 to pcre_chartables.c because without it, gcc 4.x may remove the array
3275 definition from the final binary if PCRE is built into a static library and
3276 dead code stripping is activated.
3277
3278 46. For an unanchored pattern, if a match attempt fails at the start of a
3279 newline sequence, and the newline setting is CRLF or ANY, and the next two
3280 characters are CRLF, advance by two characters instead of one.
3281
3282
3283 Version 6.7 04-Jul-06
3284 ---------------------
3285
3286 1. In order to handle tests when input lines are enormously long, pcretest has
3287 been re-factored so that it automatically extends its buffers when
3288 necessary. The code is crude, but this _is_ just a test program. The
3289 default size has been increased from 32K to 50K.
3290
3291 2. The code in pcre_study() was using the value of the re argument before
3292 testing it for NULL. (Of course, in any sensible call of the function, it
3293 won't be NULL.)
3294
3295 3. The memmove() emulation function in pcre_internal.h, which is used on
3296 systems that lack both memmove() and bcopy() - that is, hardly ever -
3297 was missing a "static" storage class specifier.
3298
3299 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3300 containing an extended class (one that cannot be represented by a bitmap
3301 because it contains high-valued characters or Unicode property items, e.g.
3302 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3303 pattern, but PCRE should not loop if you do not (it no longer does).
3304 [Detail: two cases were found: (a) a repeated subpattern containing an
3305 extended class; (b) a recursive reference to a subpattern that followed a
3306 previous extended class. It wasn't skipping over the extended class
3307 correctly when UTF-8 mode was not set.]
3308
3309 5. A negated single-character class was not being recognized as fixed-length
3310 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3311 compile error "lookbehind assertion is not fixed length".
3312
3313 6. The RunPerlTest auxiliary script was showing an unexpected difference
3314 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3315 write a Perl script that can interpret lines of an input file either as
3316 byte characters or as UTF-8, which is what "perltest" was being required to
3317 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3318 can't do is switch easily at run time between having the "use utf8;" pragma
3319 or not. In the end, I fudged it by using the RunPerlTest script to insert
3320 "use utf8;" explicitly for the UTF-8 tests.
3321
3322 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3323 the end of the subject string, contrary to the documentation and to what
3324 Perl does. This was true of both matching functions. Now it matches only at
3325 the start of the subject and immediately after *internal* newlines.
3326
3327 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3328 a pointer to an int instead of a pointer to an unsigned long int. This
3329 caused problems on 64-bit systems.
3330
3331 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3332 instance of the 'standard' template library not being so standard".
3333
3334 10. There was no check on the number of named subpatterns nor the maximum
3335 length of a subpattern name. The product of these values is used to compute
3336 the size of the memory block for a compiled pattern. By supplying a very
3337 long subpattern name and a large number of named subpatterns, the size
3338 computation could be caused to overflow. This is now prevented by limiting
3339 the length of names to 32 characters, and the number of named subpatterns
3340 to 10,000.
3341
3342 11. Subpatterns that are repeated with specific counts have to be replicated in
3343 the compiled pattern. The size of memory for this was computed from the
3344 length of the subpattern and the repeat count. The latter is limited to
3345 65535, but there was no limit on the former, meaning that integer overflow
3346 could in principle occur. The compiled length of a repeated subpattern is
3347 now limited to 30,000 bytes in order to prevent this.
3348
3349 12. Added the optional facility to have named substrings with the same name.
3350
3351 13. Added the ability to use a named substring as a condition, using the
3352 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3353 are numbers (not recommended). Forward references are permitted.
3354
3355 14. Added forward references in named backreferences (if you see what I mean).
3356
3357 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3358 pattern could run off the end of the subject. For example, the pattern
3359 "(?s)(.{1,5})"8 did this with the subject "ab".
3360
3361 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3362 PCRE_CASELESS was set when matching characters that were quantified with ?
3363 or *.
3364
3365 17. A character class other than a single negated character that had a minimum
3366 but no maximum quantifier - for example [ab]{6,} - was not handled
3367 correctly by pce_dfa_exec(). It would match only one character.
3368
3369 18. A valid (though odd) pattern that looked like a POSIX character
3370 class but used an invalid character after [ (for example [[,abc,]]) caused
3371 pcre_compile() to give the error "Failed: internal error: code overflow" or
3372 in some cases to crash with a glibc free() error. This could even happen if
3373 the pattern terminated after [[ but there just happened to be a sequence of
3374 letters, a binary zero, and a closing ] in the memory that followed.
3375
3376 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3377 over the years. Originally (before any Unicode support), just the bottom 8
3378 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3379 output from "man perlunicode" includes this:
3380
3381 The regular expression compiler produces polymorphic opcodes. That
3382 is, the pattern adapts to the data and automatically switches to
3383 the Unicode character scheme when presented with Unicode data--or
3384 instead uses a traditional byte scheme when presented with byte
3385 data.
3386
3387 Sadly, a wide octal escape does not cause a switch, and in a string with
3388 no other multibyte characters, these octal escapes are treated as before.
3389 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3390 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3391 Unicode string.
3392
3393 I have not perpetrated such confusion in PCRE. Up till now, it took just
3394 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3395 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3396 translate to the appropriate multibyte character.
3397
3398 29. Applied some refactoring to reduce the number of warnings from Microsoft
3399 and Borland compilers. This has included removing the fudge introduced
3400 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3401 a warning about an unused variable.
3402
3403 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3404 characters since release 4.0, because Perl (from release 5.004) does not.
3405 [Or at least, is documented not to: some releases seem to be in conflict
3406 with the documentation.] However, when a pattern was studied with
3407 pcre_study() and all its branches started with \s, PCRE still included VT
3408 as a possible starting character. Of course, this did no harm; it just
3409 caused an unnecessary match attempt.
3410
3411 22. Removed a now-redundant internal flag bit that recorded the fact that case
3412 dependency changed within the pattern. This was once needed for "required
3413 byte" processing, but is no longer used. This recovers a now-scarce options
3414 bit. Also moved the least significant internal flag bit to the most-
3415 significant bit of the word, which was not previously used (hangover from
3416 the days when it was an int rather than a uint) to free up another bit for
3417 the future.
3418
3419 23. Added support for CRLF line endings as well as CR and LF. As well as the
3420 default being selectable at build time, it can now be changed at runtime
3421 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3422 specify that it is scanning data with non-default line endings.
3423
3424 24. Changed the definition of CXXLINK to make it agree with the definition of
3425 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3426
3427 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3428 recursions. This makes a big different to stack usage for some patterns.
3429
3430 26. If a subpattern containing a named recursion or subroutine reference such
3431 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3432 the space required for the compiled pattern went wrong and gave too small a
3433 value. Depending on the environment, this could lead to "Failed: internal
3434 error: code overflow at offset 49" or "glibc detected double free or
3435 corruption" errors.
3436
3437 27. Applied patches from Google (a) to support the new newline modes and (b) to
3438 advance over multibyte UTF-8 characters in GlobalReplace.
3439
3440 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3441 difference for some implementation of PCRE in some Windows version.
3442
3443 29. Added some extra testing facilities to pcretest:
3444
3445 \q<number> in a data line sets the "match limit" value
3446 \Q<number> in a data line sets the "match recursion limt" value
3447 -S <number> sets the stack size, where <number> is in megabytes
3448
3449 The -S option isn't available for Windows.
3450
3451
3452 Version 6.6 06-Feb-06
3453 ---------------------
3454
3455 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3456 in pcreposix.h. I have copied the definition from pcre.h.
3457
3458 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3459 because pcre.h is no longer a built file.
3460
3461 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3462 not normally included in the compiled code.
3463
3464
3465 Version 6.5 01-Feb-06
3466 ---------------------
3467
3468 1. When using the partial match feature with pcre_dfa_exec(), it was not
3469 anchoring the second and subsequent partial matches at the new starting
3470 point. This could lead to incorrect results. For example, with the pattern
3471 /1234/, partially matching against "123" and then "a4" gave a match.
3472
3473 2. Changes to pcregrep:
3474
3475 (a) All non-match returns from pcre_exec() were being treated as failures
3476 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3477 error message is output. Some extra information is given for the
3478 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3479 probably the only errors that are likely to be caused by users (by
3480 specifying a regex that has nested indefinite repeats, for instance).
3481 If there are more than 20 of these errors, pcregrep is abandoned.
3482
3483 (b) A binary zero was treated as data while matching, but terminated the
3484 output line if it was written out. This has been fixed: binary zeroes
3485 are now no different to any other data bytes.
3486
3487 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3488 used to set a locale for matching. The --locale=xxxx long option has
3489 been added (no short equivalent) to specify a locale explicitly on the
3490 pcregrep command, overriding the environment variables.
3491
3492 (d) When -B was used with -n, some line numbers in the output were one less
3493 than they should have been.
3494
3495 (e) Added the -o (--only-matching) option.
3496
3497 (f) If -A or -C was used with -c (count only), some lines of context were
3498 accidentally printed for the final match.
3499
3500 (g) Added the -H (--with-filename) option.
3501
3502 (h) The combination of options -rh failed to suppress file names for files
3503 that were found from directory arguments.
3504
3505 (i) Added the -D (--devices) and -d (--directories) options.
3506
3507 (j) Added the -F (--fixed-strings) option.
3508
3509 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3510
3511 (l) Added the --colo(u)r option.
3512
3513 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3514 is not present by default.
3515
3516 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3517 items such as (?R) or (?1), when the recursion could match a number of
3518 alternatives. If it matched one of the alternatives, but subsequently,
3519 outside the recursion, there was a failure, the code tried to back up into
3520 the recursion. However, because of the way PCRE is implemented, this is not
3521 possible, and the result was an incorrect result from the match.
3522
3523 In order to prevent this happening, the specification of recursion has
3524 been changed so that all such subpatterns are automatically treated as
3525 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3526
3527 4. I had overlooked the fact that, in some locales, there are characters for
3528 which isalpha() is true but neither isupper() nor islower() are true. In
3529 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3530 and ordfeminine) are like this. This affected the treatment of \w and \W
3531 when they appeared in character classes, but not when they appeared outside
3532 a character class. The bit map for "word" characters is now created
3533 separately from the results of isalnum() instead of just taking it from the
3534 upper, lower, and digit maps. (Plus the underscore character, of course.)
3535
3536 5. The above bug also affected the handling of POSIX character classes such as
3537 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3538 permanent tables. Instead, the bit maps for such a class were previously
3539 created as the appropriate unions of the upper, lower, and digit bitmaps.
3540 Now they are created by subtraction from the [[:word:]] class, which has
3541 its own bitmap.
3542
3543 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3544 It is created by subtracting the vertical space characters (\x09, \x0a,
3545 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3546 subtraction was done in the overall bitmap for a character class, meaning
3547 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3548 be recognized. This bug has been fixed.
3549
3550 7. Patches from the folks at Google:
3551
3552 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3553 real life, but is still worth protecting against".
3554
3555 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3556 regular expressions".
3557
3558 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3559 have it.
3560
3561 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3562 "configure" and the latter not, in order to fix a problem somebody had
3563 with compiling the Arg class on HP-UX.
3564
3565 (e) Improve the error-handling of the C++ wrapper a little bit.
3566
3567 (f) New tests for checking recursion limiting.
3568
3569 8. The pcre_memmove() function, which is used only if the environment does not
3570 have a standard memmove() function (and is therefore rarely compiled),
3571 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3572 returning a result (though PCRE never actually uses the result).
3573
3574 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3575 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3576 returned instead of calling malloc() with an overflowing number that would
3577 most likely cause subsequent chaos.
3578
3579 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3580
3581 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3582 with this option is matched, the nmatch and pmatch options of regexec() are
3583 ignored.
3584
3585 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3586 provided in case anyone wants to the the POSIX interface with UTF-8
3587 strings.
3588
3589 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3590 C++ linking (needed for some HP-UX environments).
3591
3592 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3593 (unused parameter) and in the pcre_printint() function (omitted "default"
3594 switch label when the default is to do nothing).
3595
3596 15. Added some code to make it possible, when PCRE is compiled as a C++
3597 library, to replace subject pointers for pcre_exec() with a smart pointer
3598 class, thus making it possible to process discontinuous strings.
3599
3600 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3601 much the same function. They were added by different people who were trying
3602 to make PCRE easy to compile on non-Unix systems. It has been suggested
3603 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3604 for compiling on Windows systems. I have therefore replaced it with
3605 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3606 defaults to "extern" for C or "extern C" for C++, which works fine on
3607 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3608 SCOPE with something explicit in config.h. In addition:
3609
3610 (a) pcreposix.h still had just "extern" instead of either of these macros;
3611 I have replaced it with PCRE_DATA_SCOPE.
3612
3613 (b) Functions such as _pcre_xclass(), which are internal to the library,
3614 but external in the C sense, all had PCRE_EXPORT in their definitions.
3615 This is apparently wrong for the Windows case, so I have removed it.
3616 (It makes no difference on Unix-like systems.)
3617
3618 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3619 of recursive calls to match(). This is different to MATCH_LIMIT because
3620 that limits the total number of calls to match(), not all of which increase
3621 the depth of recursion. Limiting the recursion depth limits the amount of
3622 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3623 when PCRE is compiled, and changed at run time. A patch from Google adds
3624 this functionality to the C++ interface.
3625
3626 18. Changes to the handling of Unicode character properties:
3627
3628 (a) Updated the table to Unicode 4.1.0.
3629
3630 (b) Recognize characters that are not in the table as "Cn" (undefined).
3631
3632 (c) I revised the way the table is implemented to a much improved format
3633 which includes recognition of ranges. It now supports the ranges that
3634 are defined in UnicodeData.txt, and it also amalgamates other
3635 characters into ranges. This has reduced the number of entries in the
3636 table from around 16,000 to around 3,000, thus reducing its size
3637 considerably. I realized I did not need to use a tree structure after
3638 all - a binary chop search is just as efficient. Having reduced the
3639 number of entries, I extended their size from 6 bytes to 8 bytes to
3640 allow for more data.
3641
3642 (d) Added support for Unicode script names via properties such as \p{Han}.
3643
3644 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3645 matching that character.
3646
3647 20. When matching a repeated Unicode property with a minimum greater than zero,
3648 (for example \pL{2,}), PCRE could look past the end of the subject if it
3649 reached it while seeking the minimum number of characters. This could
3650 happen only if some of the characters were more than one byte long, because
3651 there is a check for at least the minimum number of bytes.
3652
3653 21. Refactored the implementation of \p and \P so as to be more general, to
3654 allow for more different types of property in future. This has changed the
3655 compiled form incompatibly. Anybody with saved compiled patterns that use
3656 \p or \P will have to recompile them.
3657
3658 22. Added "Any" and "L&" to the supported property types.
3659
3660 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3661 but give a compile time error if the value is greater than 0xff.
3662
3663 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3664 accidentally not being installed or uninstalled.
3665
3666 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3667 made were to insert the current release number. This seemed silly, because
3668 it made things harder for people building PCRE on systems that don't run
3669 "configure". I have turned pcre.h into a distributed file, no longer built
3670 by "configure", with the version identification directly included. There is
3671 no longer a pcre.h.in file.
3672
3673 However, this change necessitated a change to the pcre-config script as
3674 well. It is built from pcre-config.in, and one of the substitutions was the
3675 release number. I have updated configure.ac so that ./configure now finds
3676 the release number by grepping pcre.h.
3677
3678 26. Added the ability to run the tests under valgrind.
3679
3680
3681 Version 6.4 05-Sep-05
3682 ---------------------
3683
3684 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3685 "--" to be printed when multiple files were scanned, even when none of the
3686 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3687 consider it to be a bug, and have restored the previous behaviour.
3688
3689 2. A couple of code tidies to get rid of compiler warnings.
3690
3691 3. The pcretest program used to cheat by referring to symbols in the library
3692 whose names begin with _pcre_. These are internal symbols that are not
3693 really supposed to be visible externally, and in some environments it is
3694 possible to suppress them. The cheating is now confined to including
3695 certain files from the library's source, which is a bit cleaner.
3696
3697 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3698 file's purpose clearer.
3699
3700 5. Reorganized pcre_ucp_findchar().
3701
3702
3703 Version 6.3 15-Aug-05
3704 ---------------------
3705
3706 1. The file libpcre.pc.in did not have general read permission in the tarball.
3707
3708 2. There were some problems when building without C++ support:
3709
3710 (a) If C++ support was not built, "make install" and "make test" still
3711 tried to test it.
3712
3713 (b) There were problems when the value of CXX was explicitly set. Some
3714 changes have been made to try to fix these, and ...
3715
3716 (c) --disable-cpp can now be used to explicitly disable C++ support.
3717
3718 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3719 backslash in a target when C++ was disabled. This confuses some
3720 versions of "make", apparently. Using an intermediate variable solves
3721 this. (Same for CPP_LOBJ.)
3722
3723 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3724 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3725 necessary on certain architectures.
3726
3727 4. Added a setting of -export-symbols-regex to the link command to remove
3728 those symbols that are exported in the C sense, but actually are local
3729 within the library, and not documented. Their names all begin with
3730 "_pcre_". This is not a perfect job, because (a) we have to except some
3731 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3732 available (and never for static libraries). I have made a note to try to
3733 find a way round (a) in the future.
3734
3735
3736 Version 6.2 01-Aug-05
3737 ---------------------
3738
3739 1. There was no test for integer overflow of quantifier values. A construction
3740 such as {1111111111111111} would give undefined results. What is worse, if
3741 a minimum quantifier for a parenthesized subpattern overflowed and became
3742 negative, the calculation of the memory size went wrong. This could have
3743 led to memory overwriting.
3744
3745 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3746
3747 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3748 operating environments where this matters.
3749
3750 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3751 PCRE options from within the C++ wrapper.
3752
3753 5. Named capturing subpatterns were not being correctly counted when a pattern
3754 was compiled. This caused two problems: (a) If there were more than 100
3755 such subpatterns, the calculation of the memory needed for the whole
3756 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3757 back references of the form \12, where the number was greater than 9, were
3758 not recognized as back references, even though there were sufficient
3759 previous subpatterns.
3760
3761 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3762 versions of gcc, e.g. 2.95.4.
3763
3764
3765 Version 6.1 21-Jun-05
3766 ---------------------
3767
3768 1. There was one reference to the variable "posix" in pcretest.c that was not
3769 surrounded by "#if !defined NOPOSIX".
3770
3771 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3772 the cross-check on the old pcre_info() function, for the benefit of the
3773 cut-down version of PCRE that is currently imported into Exim.
3774
3775 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3776 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3777 patterns that start (?i) but I don't think that matters. The use of (?i) is
3778 just an example; this all applies to the other options as well.
3779
3780 4. Since libtool seems to echo the compile commands it is issuing, the output
3781 from "make" can be reduced a bit by putting "@" in front of each libtool
3782 compile command.
3783
3784 5. Patch from the folks at Google for configure.in to be a bit more thorough
3785 in checking for a suitable C++ installation before trying to compile the
3786 C++ stuff. This should fix a reported problem when a compiler was present,
3787 but no suitable headers.
3788
3789 6. The man pages all had just "PCRE" as their title. I have changed them to
3790 be the relevant file name. I have also arranged that these names are
3791 retained in the file doc/pcre.txt, which is a concatenation in text format
3792 of all the man pages except the little individual ones for each function.
3793
3794 7. The NON-UNIX-USE file had not been updated for the different set of source
3795 files that come with release 6. I also added a few comments about the C++
3796 wrapper.
3797
3798
3799 Version 6.0 07-Jun-05
3800 ---------------------
3801
3802 1. Some minor internal re-organization to help with my DFA experiments.
3803
3804 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3805 didn't matter for the library itself when fully configured, but did matter
3806 when compiling without UCP support, or within Exim, where the ucp files are
3807 not imported.
3808
3809 3. Refactoring of the library code to split up the various functions into
3810 different source modules. The addition of the new DFA matching code (see
3811 below) to a single monolithic source would have made it really too
3812 unwieldy, quite apart from causing all the code to be include in a
3813 statically linked application, when only some functions are used. This is
3814 relevant even without the DFA addition now that patterns can be compiled in
3815 one application and matched in another.
3816
3817 The downside of splitting up is that there have to be some external
3818 functions and data tables that are used internally in different modules of
3819 the library but which are not part of the API. These have all had their
3820 names changed to start with "_pcre_" so that they are unlikely to clash
3821 with other external names.
3822
3823 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3824 a different (DFA) algorithm. Although it is slower than the original
3825 function, it does have some advantages for certain types of matching
3826 problem.
3827
3828 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3829 including restarting after a partial match.
3830
3831 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3832 defined when compiling for Windows was sent to me. I have put it into the
3833 code, though I have no means of testing or verifying it.
3834
3835 7. Added the pcre_refcount() auxiliary function.
3836
3837 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3838 match before or at the first newline in the subject string. In pcretest,
3839 the /f option on a pattern can be used to set this.
3840
3841 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3842 would behave wrongly. This has been present in PCRE since release 4.0.
3843
3844 10. A number of changes to the pcregrep command:
3845
3846 (a) Refactored how -x works; insert ^(...)$ instead of setting
3847 PCRE_ANCHORED and checking the length, in preparation for adding
3848 something similar for -w.
3849
3850 (b) Added the -w (match as a word) option.
3851
3852 (c) Refactored the way lines are read and buffered so as to have more
3853 than one at a time available.
3854
3855 (d) Implemented a pcregrep test script.
3856
3857 (e) Added the -M (multiline match) option. This allows patterns to match
3858 over several lines of the subject. The buffering ensures that at least
3859 8K, or the rest of the document (whichever is the shorter) is available
3860 for matching (and similarly the previous 8K for lookbehind assertions).
3861
3862 (f) Changed the --help output so that it now says
3863
3864 -w, --word-regex(p)
3865
3866 instead of two lines, one with "regex" and the other with "regexp"
3867 because that confused at least one person since the short forms are the
3868 same. (This required a bit of code, as the output is generated
3869 automatically from a table. It wasn't just a text change.)
3870
3871 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3872 option but starts with a hyphen. Could be a pattern or a path name
3873 starting with a hyphen, for instance.
3874
3875 (h) "-" can be given as a file name to represent stdin.
3876
3877 (i) When file names are being printed, "(standard input)" is used for
3878 the standard input, for compatibility with GNU grep. Previously
3879 "<stdin>" was used.
3880
3881 (j) The option --label=xxx can be used to supply a name to be used for
3882 stdin when file names are being printed. There is no short form.
3883
3884 (k) Re-factored the options decoding logic because we are going to add
3885 two more options that take data. Such options can now be given in four
3886 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3887
3888 (l) Added the -A, -B, and -C options for requesting that lines of context
3889 around matches be printed.
3890
3891 (m) Added the -L option to print the names of files that do not contain
3892 any matching lines, that is, the complement of -l.
3893
3894 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3895 continue to scan other files.
3896
3897 (o) The -s option was incorrectly implemented. For compatibility with other
3898 greps, it now suppresses the error message for a non-existent or non-
3899 accessible file (but not the return code). There is a new option called
3900 -q that suppresses the output of matching lines, which was what -s was
3901 previously doing.
3902
3903 (p) Added --include and --exclude options to specify files for inclusion
3904 and exclusion when recursing.
3905
3906 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3907 Hopefully, it now does.
3908
3909 12. Missing cast in pcre_study().
3910
3911 13. Added an "uninstall" target to the makefile.
3912
3913 14. Replaced "extern" in the function prototypes in Makefile.in with
3914 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3915 world, but is set differently for Windows.
3916
3917 15. Added a second compiling function called pcre_compile2(). The only
3918 difference is that it has an extra argument, which is a pointer to an
3919 integer error code. When there is a compile-time failure, this is set
3920 non-zero, in addition to the error test pointer being set to point to an
3921 error message. The new argument may be NULL if no error number is required
3922 (but then you may as well call pcre_compile(), which is now just a
3923 wrapper). This facility is provided because some applications need a
3924 numeric error indication, but it has also enabled me to tidy up the way
3925 compile-time errors are handled in the POSIX wrapper.
3926
3927 16. Added VPATH=.libs to the makefile; this should help when building with one
3928 prefix path and installing with another. (Or so I'm told by someone who
3929 knows more about this stuff than I do.)
3930
3931 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3932 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3933 match everything, including newlines. This is not POSIX-compatible, but
3934 somebody wanted the feature. From pcretest it can be activated by using
3935 both the P and the s flags.
3936
3937 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3938
3939 19. libpcre.pc was being incorrectly installed as executable.
3940
3941 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3942 it now also looks for '\r' so that it will work unmodified on Windows.
3943
3944 21. Added Google's contributed C++ wrapper to the distribution.
3945
3946 22. Added some untidy missing memory free() calls in pcretest, to keep
3947 Electric Fence happy when testing.
3948
3949
3950
3951 Version 5.0 13-Sep-04
3952 ---------------------
3953
3954 1. Internal change: literal characters are no longer packed up into items
3955 containing multiple characters in a single byte-string. Each character
3956 is now matched using a separate opcode. However, there may be more than one
3957 byte in the character in UTF-8 mode.
3958
3959 2. The pcre_callout_block structure has two new fields: pattern_position and
3960 next_item_length. These contain the offset in the pattern to the next match
3961 item, and its length, respectively.
3962
3963 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3964 insertion of callouts before each pattern item. Added the /C option to
3965 pcretest to make use of this.
3966
3967 4. On the advice of a Windows user, the lines
3968
3969 #if defined(_WIN32) || defined(WIN32)
3970 _setmode( _fileno( stdout ), 0x8000 );
3971 #endif /* defined(_WIN32) || defined(WIN32) */
3972
3973 have been added to the source of pcretest. This apparently does useful
3974 magic in relation to line terminators.
3975
3976 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3977 for the benefit of those environments where the "b" makes a difference.
3978
3979 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3980 to know about it. I have put a hack into configure.in that adds in code
3981 to set GCC=yes if CC=icc. This seems to end up at a point in the
3982 generated configure script that is early enough to affect the setting of
3983 compiler options, which is what is needed, but I have no means of testing
3984 whether it really works. (The user who reported this had patched the
3985 generated configure script, which of course I cannot do.)
3986
3987 LATER: After change 22 below (new libtool files), the configure script
3988 seems to know about icc (and also ecc). Therefore, I have commented out
3989 this hack in configure.in.
3990
3991 7. Added support for pkg-config (2 patches were sent in).
3992
3993 8. Negated POSIX character classes that used a combination of internal tables
3994 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3995 [[:^ascii]]. Typically, they would match almost any characters. The other
3996 POSIX classes were not broken in this way.
3997
3998 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3999 to find the match, as PCRE was deluded into thinking that the match had to
4000 start at the start point or following a newline. The same bug applied to
4001 patterns with negative forward assertions or any backward assertions
4002 preceding ".*" at the start, unless the pattern required a fixed first
4003 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4004
4005 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4006 starting at the last subject character, bytes beyond the end of the subject
4007 string were read.
4008
4009 11. Renamed the variable "class" as "classbits" to make life easier for C++
4010 users. (Previously there was a macro definition, but it apparently wasn't
4011 enough.)
4012
4013 12. Added the new field "tables" to the extra data so that tables can be passed
4014 in at exec time, or the internal tables can be re-selected. This allows
4015 a compiled regex to be saved and re-used at a later time by a different
4016 program that might have everything at different addresses.
4017
4018 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4019 -R library as well as a -L library.
4020
4021 14. The debugging options of pcretest (-d on the command line or D on a
4022 pattern) showed incorrect output for anything following an extended class
4023 that contained multibyte characters and which was followed by a quantifier.
4024
4025 15. Added optional support for general category Unicode character properties
4026 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4027 support. It adds about 90K to the size of the library. The meanings of the
4028 inbuilt class escapes such as \d and \s have NOT been changed.
4029
4030 16. Updated pcredemo.c to include calls to free() to release the memory for the
4031 compiled pattern.
4032
4033 17. The generated file chartables.c was being created in the source directory
4034 instead of in the building directory. This caused the build to fail if the
4035 source directory was different from the building directory, and was
4036 read-only.
4037
4038 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4039 file. No doubt somebody will tell me if they don't make sense... Also added
4040 Dan Mooney's comments about building on OpenVMS.
4041
4042 19. Added support for partial matching via the PCRE_PARTIAL option for
4043 pcre_exec() and the \P data escape in pcretest.
4044
4045 20. Extended pcretest with 3 new pattern features:
4046
4047 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4048 write the compiled pattern to the file whose name is "rest-of-line".
4049 This is a straight binary dump of the data, with the saved pointer to
4050 the character tables forced to be NULL. The study data, if any, is
4051 written too. After writing, pcretest reads a new pattern.
4052
4053 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4054 compiled pattern from the given file. There must not be any
4055 occurrences of "<" in the file name (pretty unlikely); if there are,
4056 pcretest will instead treat the initial "<" as a pattern delimiter.
4057 After reading in the pattern, pcretest goes on to read data lines as
4058 usual.
4059
4060 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4061 and 16-bit fields in a compiled pattern, to simulate a pattern that
4062 was compiled on a host of opposite endianness.
4063
4064 21. The pcre-exec() function can now cope with patterns that were compiled on
4065 hosts of opposite endianness, with this restriction:
4066
4067 As for any compiled expression that is saved and used later, the tables
4068 pointer field cannot be preserved; the extra_data field in the arguments
4069 to pcre_exec() should be used to pass in a tables address if a value
4070 other than the default internal tables were used at compile time.
4071
4072 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4073 now diagnosed as an error. Previously, most of the time, a negative number
4074 would have been treated as zero, but if in addition "ovector" was passed as
4075 NULL, a crash could occur.
4076
4077 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4078 new versions from the libtool 1.5 distribution (the last one is a copy of
4079 a file called libtool.m4). This seems to have fixed the need to patch
4080 "configure" to support Darwin 1.3 (which I used to do). However, I still
4081 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4082 workstation).
4083
4084 24. Changed the PCRE licence to be the more standard "BSD" licence.
4085
4086
4087 Version 4.5 01-Dec-03
4088 ---------------------
4089
4090 1. There has been some re-arrangement of the code for the match() function so
4091 that it can be compiled in a version that does not call itself recursively.
4092 Instead, it keeps those local variables that need separate instances for
4093 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4094 needs to "recurse". Keeping track of where control must go is done by means
4095 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4096 hide most of the details from the main code, and operates only if
4097 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4098 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4099 operating.
4100
4101 To make it easier for callers to provide specially tailored get/free
4102 functions for this usage, two new functions, pcre_stack_malloc, and
4103 pcre_stack_free, are used. They are always called in strict stacking order,
4104 and the size of block requested is always the same.
4105
4106 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4107 PCRE has been compiled to use the stack or the heap for recursion. The
4108 -C option of pcretest uses this to show which version is compiled.
4109
4110 A new data escape \S, is added to pcretest; it causes the amounts of store
4111 obtained and freed by both kinds of malloc/free at match time to be added
4112 to the output.
4113
4114 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4115 what's available on my current Linux desktop machine.
4116
4117 3. When matching a UTF-8 string, the test for a valid string at the start has
4118 been extended. If start_offset is not zero, PCRE now checks that it points
4119 to a byte that is the start of a UTF-8 character. If not, it returns
4120 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4121 this is necessary because there may be backward assertions in the pattern.
4122 When matching the same subject several times, it may save resources to use
4123 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4124
4125 4. The code for checking the validity of UTF-8 strings has been tightened so
4126 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4127 containing "overlong sequences".
4128
4129 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4130 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4131 should have been "&", but it just so happened that all the cases this let
4132 through by mistake were picked up later in the function.
4133
4134 6. I had used a variable called "isblank" - this is a C99 function, causing
4135 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4136
4137 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4138 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4139 so the version is shown; (c) stop "make test" echoing "./RunTest".
4140
4141 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4142
4143 9. The prototype for memmove() for systems that don't have it was using
4144 size_t, but the inclusion of the header that defines size_t was later. I've
4145 moved the #includes for the C headers earlier to avoid this.
4146
4147 10. Added some adjustments to the code to make it easier to compiler on certain
4148 special systems:
4149
4150 (a) Some "const" qualifiers were missing.
4151 (b) Added the macro EXPORT before all exported functions; by default this
4152 is defined to be empty.
4153 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4154 that it reads its output file name as an argument instead of writing
4155 to the standard output and assuming this can be redirected.
4156
4157 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4158 class containing characters with values greater than 255, PCRE compilation
4159 went into a loop.
4160
4161 12. A recursive reference to a subpattern that was within another subpattern
4162 that had a minimum quantifier of zero caused PCRE to crash. For example,
4163 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4164 recursion. If the recursively-called subpattern itself had a zero repeat,
4165 that was OK.
4166
4167 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4168 buffer into which it was copied (for escape processing) was still set at
4169 1024, so long lines caused crashes.
4170
4171 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4172 "internal error: code overflow...". This applied to any character class
4173 that was followed by a possessive quantifier.
4174
4175 15. Modified the Makefile to add libpcre.la as a prerequisite for
4176 libpcreposix.la because I was told this is needed for a parallel build to
4177 work.
4178
4179 16. If a pattern that contained .* following optional items at the start was
4180 studied, the wrong optimizing data was generated, leading to matching
4181 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4182 matching string must start with a or b or c. The correct conclusion for
4183 this pattern is that a match can start with any character.
4184
4185
4186 Version 4.4 13-Aug-03
4187 ---------------------
4188
4189 1. In UTF-8 mode, a character class containing characters with values between
4190 127 and 255 was not handled correctly if the compiled pattern was studied.
4191 In fixing this, I have also improved the studying algorithm for such
4192 classes (slightly).
4193
4194 2. Three internal functions had redundant arguments passed to them. Removal
4195 might give a very teeny performance improvement.
4196
4197 3. Documentation bug: the value of the capture_top field in a callout is *one
4198 more than* the number of the hightest numbered captured substring.
4199
4200 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4201 in incorrectly linking with a previously installed version. They now link
4202 explicitly with libpcre.la.
4203
4204 5. configure.in no longer needs to recognize Cygwin specially.
4205
4206 6. A problem in pcre.in for Windows platforms is fixed.
4207
4208 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4209 pcretest, it used to include the size of the study block as part of its
4210 output. Unfortunately, the structure contains a field that has a different
4211 size on different hardware architectures. This meant that the tests that
4212 showed this size failed. As the block is currently always of a fixed size,
4213 this information isn't actually particularly useful in pcretest output, so
4214 I have just removed it.
4215
4216 8. Three pre-processor statements accidentally did not start in column 1.
4217 Sadly, there are *still* compilers around that complain, even though
4218 standard C has not required this for well over a decade. Sigh.
4219
4220 9. In pcretest, the code for checking callouts passed small integers in the
4221 callout_data field, which is a void * field. However, some picky compilers
4222 complained about the casts involved for this on 64-bit systems. Now
4223 pcretest passes the address of the small integer instead, which should get
4224 rid of the warnings.
4225
4226 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4227 both compile and run time, and gives an error if an invalid UTF-8 sequence
4228 is found. There is a option for disabling this check in cases where the
4229 string is known to be correct and/or the maximum performance is wanted.
4230
4231 11. In response to a bug report, I changed one line in Makefile.in from
4232
4233 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4234 to
4235 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4236
4237 to look similar to other lines, but I have no way of telling whether this
4238 is the right thing to do, as I do not use Windows. No doubt I'll get told
4239 if it's wrong...
4240
4241
4242 Version 4.3 21-May-03
4243 ---------------------
4244
4245 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4246 Makefile.
4247
4248 2. Some refactoring to improve the quality of the code:
4249
4250 (i) The utf8_table... variables are now declared "const".
4251
4252 (ii) The code for \cx, which used the "case flipping" table to upper case
4253 lower case letters, now just substracts 32. This is ASCII-specific,
4254 but the whole concept of \cx is ASCII-specific, so it seems
4255 reasonable.
4256
4257 (iii) PCRE was using its character types table to recognize decimal and
4258 hexadecimal digits in the pattern. This is silly, because it handles
4259 only 0-9, a-f, and A-F, but the character types table is locale-
4260 specific, which means strange things might happen. A private
4261 table is now used for this - though it costs 256 bytes, a table is
4262 much faster than multiple explicit tests. Of course, the standard
4263 character types table is still used for matching digits in subject
4264 strings against \d.
4265
4266 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4267 ending in _t are). So I've renamed it as ESC_tee.
4268
4269 3. The first argument for regexec() in the POSIX wrapper should have been
4270 defined as "const".
4271
4272 4. Changed pcretest to use malloc() for its buffers so that they can be
4273 Electric Fenced for debugging.
4274
4275 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4276 to read one or more bytes before the start of the subject string. Often this
4277 had no effect on PCRE's behaviour, but in some circumstances it could
4278 provoke a segmentation fault.
4279
4280 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4281 to try to read one or more bytes before the start of the subject string.
4282
4283 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4284 UTF-8 support could misbehave in various ways if the subject string
4285 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4286 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4287 back over UTF-8 characters.)
4288
4289
4290 Version 4.2 14-Apr-03
4291 ---------------------
4292
4293 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4294
4295 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4296 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4297 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4298 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4299 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4300 and BUILD_EXEEXT
4301 Note: automatic setting of the BUILD variables is not yet working
4302 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4303 compile-time but not at link-time
4304 [LINK]: use for linking executables only
4305 make different versions for Windows and non-Windows
4306 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4307 libraries
4308 [LINK_FOR_BUILD]: new variable
4309 [OBJEXT]: use throughout
4310 [EXEEXT]: use throughout
4311 <winshared>: new target
4312 <wininstall>: new target
4313 <dftables.o>: use native compiler
4314 <dftables>: use native linker
4315 <install>: handle Windows platform correctly
4316 <clean>: ditto
4317 <check>: ditto
4318 copy DLL to top builddir before testing
4319
4320 As part of these changes, -no-undefined was removed again. This was reported
4321 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4322 in any case.
4323
4324 3. Some tidies to get rid of compiler warnings:
4325
4326 . In the match_data structure, match_limit was an unsigned long int, whereas
4327 match_call_count was an int. I've made them both unsigned long ints.
4328
4329 . In pcretest the fact that a const uschar * doesn't automatically cast to
4330 a void * provoked a warning.
4331
4332 . Turning on some more compiler warnings threw up some "shadow" variables
4333 and a few more missing casts.
4334
4335 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4336 option, a class that contained a single character with a value between 128
4337 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4338
4339 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4340 option, a class that contained several characters, but with at least one
4341 whose value was between 128 and 255 caused PCRE to crash.
4342
4343
4344 Version 4.1 12-Mar-03
4345 ---------------------
4346
4347 1. Compiling with gcc -pedantic found a couple of places where casts were
4348 needed, and a string in dftables.c that was longer than standard compilers are
4349 required to support.
4350
4351 2. Compiling with Sun's compiler found a few more places where the code could
4352 be tidied up in order to avoid warnings.
4353
4354 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4355 first of these names is deprecated in the latest Autoconf in favour of the name
4356 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4357 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4358 analogy I have changed it to CFLAGS_FOR_BUILD.
4359
4360 4. Added -no-undefined to the linking command in the Makefile, because this is
4361 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4362 linking step for the pcreposix library.
4363
4364 5. PCRE was failing to diagnose the case of two named groups with the same
4365 name.
4366
4367 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4368 literal character that is needed in the subject for a match, and scans along to
4369 ensure that it is present before embarking on the full matching process. This
4370 saves time in cases of nested unlimited repeats that are never going to match.
4371 Problem: the scan can take a lot of time if the subject is very long (e.g.
4372 megabytes), thus penalizing straightforward matches. It is now done only if the
4373 amount of subject to be scanned is less than 1000 bytes.
4374
4375 7. A lesser problem with the same optimization is that it was recording the
4376 first character of an anchored pattern as "needed", thus provoking a search
4377 right along the subject, even when the first match of the pattern was going to
4378 fail. The "needed" character is now not set for anchored patterns, unless it
4379 follows something in the pattern that is of non-fixed length. Thus, it still
4380 fulfils its original purpose of finding quick non-matches in cases of nested
4381 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4382
4383
4384 Version 4.0 17-Feb-03
4385 ---------------------
4386
4387 1. If a comment in an extended regex that started immediately after a meta-item
4388 extended to the end of string, PCRE compiled incorrect data. This could lead to
4389 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4390
4391 2. Moved to autoconf 2.53 and libtool 1.4.2.
4392
4393 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4394 the special perltest8 script is no longer needed - all the tests can be run
4395 from a single perltest script.
4396
4397 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4398 by \s. It has now been removed in PCRE. This means it isn't recognized as
4399 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4400 class [:space:] *does* include VT, thereby creating a mess.
4401
4402 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4403 space and tab.
4404
4405 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4406 its new features into the main test script, reducing the number of scripts.
4407
4408 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4409 were backward compatible, and made the (?i) apply to the whole pattern, as if
4410 /i were given. Now it behaves more logically, and applies the option setting
4411 only to what follows. PCRE has been changed to follow suit. However, if it
4412 finds options settings right at the start of the pattern, it extracts them into
4413 the global options, as before. Thus, they show up in the info data.
4414
4415 8. Added support for the \Q...\E escape sequence. Characters in between are
4416 treated as literals. This is slightly different from Perl in that $ and @ are
4417 also handled as literals inside the quotes. In Perl, they will cause variable
4418 interpolation. Note the following examples:
4419
4420 Pattern PCRE matches Perl matches
4421
4422 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4423 \Qabc\$xyz\E abc\$xyz abc\$xyz
4424 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4425
4426 For compatibility with Perl, \Q...\E sequences are recognized inside character
4427 classes as well as outside them.
4428
4429 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4430 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4431 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4432 signed/unsigned warnings.
4433
4434 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4435 option for pcretest, so I've replaced it by a simple function that does just
4436 that job.
4437
4438 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4439 "pcregrep -".
4440
4441 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4442 Java package. This provides some syntactic sugar for simple cases of what my
4443 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4444 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4445 item, you can use this simplified notation. Note that only makes sense with
4446 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4447 greediness, whatever the setting of the PCRE_UNGREEDY option.
4448
4449 13. A change of greediness default within a pattern was not taking effect at
4450 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4451 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4452 was abstracted outside.
4453
4454 14. PCRE now supports the \G assertion. It is true when the current matching
4455 position is at the start point of the match. This differs from \A when the
4456 starting offset is non-zero. Used with the /g option of pcretest (or similar
4457 code), it works in the same way as it does for Perl's /g option. If all
4458 alternatives of a regex begin with \G, the expression is anchored to the start
4459 match position, and the "anchored" flag is set in the compiled expression.
4460
4461 15. Some bugs concerning the handling of certain option changes within patterns
4462 have been fixed. These applied to options other than (?ims). For example,
4463 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4464 been the other way round. Some of this was related to change 7 above.
4465
4466 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4467 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4468 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4469 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4470
4471 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4472 mode. Unlike ".", it always matches newline, whatever the setting of
4473 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4474 assertions. Perl allows it, but it doesn't (in general) work because it can't
4475 calculate the length of the lookbehind. At least, that's the case for Perl
4476 5.8.0 - I've been told they are going to document that it doesn't work in
4477 future.
4478
4479 18. Added an error diagnosis for escapes that PCRE does not support: these are
4480 \L, \l, \N, \P, \p, \U, \u, and \X.
4481
4482 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4483 reading past the end of the pattern in cases such as /[abcd/.
4484
4485 20. PCRE was getting more memory than necessary for patterns with classes that
4486 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4487
4488 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4489 compiling PCRE for use with Virtual Pascal.
4490
4491 22. Small fix to the Makefile to make it work properly if the build is done
4492 outside the source tree.
4493
4494 23. Added a new extension: a condition to go with recursion. If a conditional
4495 subpattern starts with (?(R) the "true" branch is used if recursion has
4496 happened, whereas the "false" branch is used only at the top level.
4497
4498 24. When there was a very long string of literal characters (over 255 bytes
4499 without UTF support, over 250 bytes with UTF support), the computation of how
4500 much memory was required could be incorrect, leading to segfaults or other
4501 strange effects.
4502
4503 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4504 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4505 there was a subsequent back reference to those brackets. This meant that, for
4506 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4507 possible to check for precisely this case. All we can do is abandon the
4508 optimization if .* occurs inside capturing brackets when there are any back
4509 references whatsoever. (See below for a better fix that came later.)
4510
4511 26. The handling of the optimization for finding the first character of a
4512 non-anchored pattern, and for finding a character that is required later in the
4513 match were failing in some cases. This didn't break the matching; it just
4514 failed to optimize when it could. The way this is done has been re-implemented.
4515
4516 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4517
4518 28. Added a new feature that provides some of the functionality that Perl
4519 provides with (?{...}). The facility is termed a "callout". The way it is done
4520 in PCRE is for the caller to provide an optional function, by setting
4521 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4522 global variable. By default it is unset, which disables all calling out. To get
4523 the function called, the regex must include (?C) at appropriate points. This
4524 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4525 This provides a means of identifying different callout points. When PCRE
4526 reaches such a point in the regex, if pcre_callout has been set, the external
4527 function is called. It is provided with data in a structure called
4528 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4529 matching continues; if it returns a non-zero value, the match at the current
4530 point fails. However, backtracking will occur if possible. [This was changed
4531 later and other features added - see item 49 below.]
4532
4533 29. pcretest is upgraded to test the callout functionality. It provides a
4534 callout function that displays information. By default, it shows the start of
4535 the match and the current position in the text. There are some new data escapes
4536 to vary what happens:
4537
4538 \C+ in addition, show current contents of captured substrings
4539 \C- do not supply a callout function
4540 \C!n return 1 when callout number n is reached
4541 \C!n!m return 1 when callout number n is reached for the mth time
4542
4543 30. If pcregrep was called with the -l option and just a single file name, it
4544 output "<stdin>" if a match was found, instead of the file name.
4545
4546 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4547 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4548 pcre_exec(). This saves a malloc/free per call. The default value of
4549 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4550 when configuring.
4551
4552 32. The default maximum size of a compiled pattern is 64K. There have been a
4553 few cases of people hitting this limit. The code now uses macros to handle the
4554 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4555 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4556 configuring. Tests 2 and 5 work only with 2-byte links because they output
4557 debugging information about compiled patterns.
4558
4559 33. Internal code re-arrangements:
4560
4561 (a) Moved the debugging function for printing out a compiled regex into
4562 its own source file (printint.c) and used #include to pull it into
4563 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4564 separate copies.
4565
4566 (b) Defined the list of op-code names for debugging as a macro in
4567 internal.h so that it is next to the definition of the opcodes.
4568
4569 (c) Defined a table of op-code lengths for simpler skipping along compiled
4570 code. This is again a macro in internal.h so that it is next to the
4571 definition of the opcodes.
4572
4573 34. Added support for recursive calls to individual subpatterns, along the
4574 lines of Robin Houston's patch (but implemented somewhat differently).
4575
4576 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4577 allow it to read and process whole directories in Win32. This code was
4578 contributed by Lionel Fourquaux; it has not been tested by me.
4579
4580 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4581 used to name a group. Names consist of alphanumerics and underscores, and must
4582 be unique. Back references use the syntax (?P=name) and recursive calls use
4583 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4584 numbers. The function pcre_fullinfo() can be used after compilation to extract
4585 a name/number map. There are three relevant calls:
4586
4587 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4588 PCRE_INFO_NAMECOUNT yields the number of entries
4589 PCRE_INFO_NAMETABLE yields a pointer to the map.
4590
4591 The map is a vector of fixed-size entries. The size of each entry depends on
4592 the length of the longest name used. The first two bytes of each entry are the
4593 group number, most significant byte first. There follows the corresponding
4594 name, zero terminated. The names are in alphabetical order.
4595
4596 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4597 case instead of 255. Making it the same both with and without UTF-8 support
4598 means that the same test output works with both.
4599
4600 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4601 calling malloc() with a zero argument.
4602
4603 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4604 optimization. I've improved things by keeping a bitmap of backreferences with
4605 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4606 fact referenced, the optimization can be applied. It is unlikely that a
4607 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4608 the match to follow \n) will appear inside brackets with a number greater than
4609 31, but if it does, any back reference > 31 suppresses the optimization.
4610
4611 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4612 of disabling numbered capturing parentheses. Any opening parenthesis that is
4613 not followed by ? behaves as if it were followed by ?: but named parentheses
4614 can still be used for capturing (and they will acquire numbers in the usual
4615 way).
4616
4617 41. Redesigned the return codes from the match() function into yes/no/error so
4618 that errors can be passed back from deep inside the nested calls. A malloc
4619 failure while inside a recursive subpattern call now causes the
4620 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4621
4622 42. It is now possible to set a limit on the number of times the match()
4623 function is called in a call to pcre_exec(). This facility makes it possible to
4624 limit the amount of recursion and backtracking, though not in a directly
4625 obvious way, because the match() function is used in a number of different
4626 circumstances. The count starts from zero for each position in the subject
4627 string (for non-anchored patterns). The default limit is, for compatibility, a
4628 large number, namely 10 000 000. You can change this in two ways:
4629
4630 (a) When configuring PCRE before making, you can use --with-match-limit=n
4631 to set a default value for the compiled library.
4632
4633 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4634 a different value is set. See 45 below.
4635
4636 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4637
4638 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4639 of things that can be changed at compile time. The first argument specifies
4640 what is wanted and the second points to where the information is to be placed.
4641 The current list of available information is:
4642
4643 PCRE_CONFIG_UTF8
4644
4645 The output is an integer that is set to one if UTF-8 support is available;
4646 otherwise it is set to zero.
4647
4648 PCRE_CONFIG_NEWLINE
4649
4650 The output is an integer that it set to the value of the code that is used for
4651 newline. It is either LF (10) or CR (13).
4652
4653 PCRE_CONFIG_LINK_SIZE
4654
4655 The output is an integer that contains the number of bytes used for internal
4656 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4657
4658 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4659
4660 The output is an integer that contains the threshold above which the POSIX
4661 interface uses malloc() for output vectors. See item 31 above.
4662
4663 PCRE_CONFIG_MATCH_LIMIT
4664
4665 The output is an unsigned integer that contains the default limit of the number
4666 of match() calls in a pcre_exec() execution. See 42 above.
4667
4668 44. pcretest has been upgraded by the addition of the -C option. This causes it
4669 to extract all the available output from the new pcre_config() function, and to
4670 output it. The program then exits immediately.
4671
4672 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4673 order to support additional features. One way would have been to define
4674 pcre_exec2() (for example) with extra arguments, but this would not have been
4675 extensible, and would also have required all calls to the original function to
4676 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4677 is used for passing in "extra" data from pcre_study().
4678
4679 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4680 contains the following fields:
4681
4682 flags a bitmap indicating which of the following fields are set
4683 study_data opaque data from pcre_study()
4684 match_limit a way of specifying a limit on match() calls for a specific
4685 call to pcre_exec()
4686 callout_data data for callouts (see 49 below)
4687
4688 The flag bits are also defined in pcre.h, and are
4689
4690 PCRE_EXTRA_STUDY_DATA
4691 PCRE_EXTRA_MATCH_LIMIT
4692 PCRE_EXTRA_CALLOUT_DATA
4693
4694 The pcre_study() function now returns one of these new pcre_extra blocks, with
4695 the actual study data pointed to by the study_data field, and the
4696 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4697 before. That is, this change is entirely upwards-compatible and requires no
4698 change to existing code.
4699
4700 If you want to pass in additional data to pcre_exec(), you can either place it
4701 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4702 block.
4703
4704 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4705 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4706 times with different match limits, until it finds the minimum value needed for
4707 pcre_exec() to complete. The value is then output. This can be instructive; for
4708 most simple matches the number is quite small, but for pathological cases it
4709 gets very large very quickly.
4710
4711 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4712 returns the size of the data block pointed to by the study_data field in a
4713 pcre_extra block, that is, the value that was passed as the argument to
4714 pcre_malloc() when PCRE was getting memory in which to place the information
4715 created by pcre_study(). The fourth argument should point to a size_t variable.
4716 pcretest has been extended so that this information is shown after a successful
4717 pcre_study() call when information about the compiled regex is being displayed.
4718
4719 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4720 because what follows is always an absolute path. (Later: it turns out that this
4721 is more than cosmetic for MinGW, because it doesn't like empty path
4722 components.)
4723
4724 49. Some changes have been made to the callout feature (see 28 above):
4725
4726 (i) A callout function now has three choices for what it returns:
4727
4728 0 => success, carry on matching
4729 > 0 => failure at this point, but backtrack if possible
4730 < 0 => serious error, return this value from pcre_exec()
4731
4732 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4733 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4734 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4735 use by callout functions. It will never be used by PCRE itself.
4736
4737 (ii) The pcre_extra structure (see 45 above) has a void * field called
4738 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4739 pcre_callout_block structure has a field of the same name. The contents of
4740 the field passed in the pcre_extra structure are passed to the callout
4741 function in the corresponding field in the callout block. This makes it
4742 easier to use the same callout-containing regex from multiple threads. For
4743 testing, the pcretest program has a new data escape
4744
4745 \C*n pass the number n (may be negative) as callout_data
4746
4747 If the callout function in pcretest receives a non-zero value as
4748 callout_data, it returns that value.
4749
4750 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4751 there were some redundant $(CFLAGS) in commands that are now specified as
4752 $(LINK), which already includes $(CFLAGS).
4753
4754 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4755 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4756 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4757 one-byte characters throughout. Note that case-insensitive matching applies
4758 only to characters whose values are less than 256. PCRE doesn't support the
4759 notion of cases for higher-valued characters.
4760
4761 (i) A character class whose characters are all within 0-255 is handled as
4762 a bit map, and the map is inverted for negative classes. Previously, a
4763 character > 255 always failed to match such a class; however it should
4764 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4765
4766 (ii) A negated character class with a single character < 255 is coded as
4767 "not this character" (OP_NOT). This wasn't working properly when the test
4768 character was multibyte, either singly or repeated.
4769
4770 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4771 mode, for example: \x{100}{2,3}.
4772
4773 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4774 singly or repeated) now correctly test multibyte characters. However,
4775 PCRE doesn't recognize any characters with values greater than 255 as
4776 digits, spaces, or word characters. Such characters always match \D, \S,
4777 and \W, and never match \d, \s, or \w.
4778
4779 (v) Classes may now contain characters and character ranges with values
4780 greater than 255. For example: [ab\x{100}-\x{400}].
4781
4782 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4783 PCRE in UTF-8 mode.
4784
4785 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4786 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4787 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4788 value.)
4789
4790 53. The single man page has become too large. I have therefore split it up into
4791 a number of separate man pages. These also give rise to individual HTML pages;
4792 these are now put in a separate directory, and there is an index.html page that
4793 lists them all. Some hyperlinking between the pages has been installed.
4794
4795 54. Added convenience functions for handling named capturing parentheses.
4796
4797 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4798 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4799 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4800 are faulted.
4801
4802 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4803 calling configure. These values are used when compiling the dftables.c program
4804 which is run to generate the source of the default character tables. They
4805 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4806 you will need to set these values.
4807
4808 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4809
4810
4811 Version 3.9 02-Jan-02
4812 ---------------------
4813
4814 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4815
4816 2. If --disable-static was given, the building process failed when trying to
4817 build pcretest and pcregrep. (For some reason it was using libtool to compile
4818 them, which is not right, as they aren't part of the library.)
4819
4820
4821 Version 3.8 18-Dec-01
4822 ---------------------
4823
4824 1. The experimental UTF-8 code was completely screwed up. It was packing the
4825 bytes in the wrong order. How dumb can you get?
4826
4827
4828 Version 3.7 29-Oct-01
4829 ---------------------
4830
4831 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4832 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4833 this didn't happen under Solaris 8, where I normally test things.
4834
4835 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4836 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4837 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4838 this fix an hour or so after the initial 3.7 release.)
4839
4840
4841 Version 3.6 23-Oct-01
4842 ---------------------
4843
4844 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4845 offsets passed as NULL with zero offset count.
4846
4847 2. The config.guess and config.sub files had not been updated when I moved to
4848 the latest autoconf.
4849
4850
4851 Version 3.5 15-Aug-01
4852 ---------------------
4853
4854 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4855 had been forgotten.
4856
4857 2. By using declared but undefined structures, we can avoid using "void"
4858 definitions in pcre.h while keeping the internal definitions of the structures
4859 private.
4860
4861 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4862 user point of view, this means that both static and shared libraries are built
4863 by default, but this can be individually controlled. More of the work of
4864 handling this static/shared cases is now inside libtool instead of PCRE's make
4865 file.
4866
4867 4. The pcretest utility is now installed along with pcregrep because it is
4868 useful for users (to test regexs) and by doing this, it automatically gets
4869 relinked by libtool. The documentation has been turned into a man page, so
4870 there are now .1, .txt, and .html versions in /doc.
4871
4872 5. Upgrades to pcregrep:
4873 (i) Added long-form option names like gnu grep.
4874 (ii) Added --help to list all options with an explanatory phrase.
4875 (iii) Added -r, --recursive to recurse into sub-directories.
4876 (iv) Added -f, --file to read patterns from a file.
4877
4878 6. pcre_exec() was referring to its "code" argument before testing that
4879 argument for NULL (and giving an error if it was NULL).
4880
4881 7. Upgraded Makefile.in to allow for compiling in a different directory from
4882 the source directory.
4883
4884 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4885 options bits, the pointer it was passed was to an int instead of to an unsigned
4886 long int. This mattered only on 64-bit systems.
4887
4888 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4889 generated) instead of pcre.in, which it its source. Also made the same change
4890 in several of the .c files.
4891
4892 10. A new release of gcc defines printf() as a macro, which broke pcretest
4893 because it had an ifdef in the middle of a string argument for printf(). Fixed
4894 by using separate calls to printf().
4895
4896 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4897 script, to force use of CR or LF instead of \n in the source. On non-Unix
4898 systems, the value can be set in config.h.
4899
4900 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4901 absolute limit. Changed the text of the error message to make this clear, and
4902 likewise updated the man page.
4903
4904 13. The limit of 99 on the number of capturing subpatterns has been removed.
4905 The new limit is 65535, which I hope will not be a "real" limit.
4906
4907
4908 Version 3.4 22-Aug-00
4909 ---------------------
4910
4911 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4912
4913 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4914
4915
4916 Version 3.3 01-Aug-00
4917 ---------------------
4918
4919 1. If an octal character was given, but the value was greater than \377, it
4920 was not getting masked to the least significant bits, as documented. This could
4921 lead to crashes in some systems.
4922
4923 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4924 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4925
4926 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4927 These just pass their arguments on to (pcre_free)(), but they are provided
4928 because some uses of PCRE bind it to non-C systems that can call its functions,
4929 but cannot call free() or pcre_free() directly.
4930
4931 4. Add "make test" as a synonym for "make check". Corrected some comments in
4932 the Makefile.
4933
4934 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4935 Makefile.
4936
4937 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4938 command called pgrep for grepping around the active processes.
4939
4940 7. Added the beginnings of support for UTF-8 character strings.
4941
4942 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4943 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4944 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4945 out for the ar command.)
4946
4947
4948 Version 3.2 12-May-00
4949 ---------------------
4950
4951 This is purely a bug fixing release.
4952
4953 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4954 of ZA. This was just one example of several cases that could provoke this bug,
4955 which was introduced by change 9 of version 2.00. The code for breaking
4956 infinite loops after an iteration that matches an empty string was't working
4957 correctly.
4958
4959 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4960 when matched against abbab (for example). After matching an empty string, it
4961 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4962 caused it to match further down the string than it should.
4963
4964 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4965 was there because it doesn't seem to be needed, and it causes trouble on some
4966 systems, as it is not a Standard C header. It has been removed.
4967
4968 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4969 were reported on the Macintosh. The changes were from
4970
4971 while ((c = *(++ptr)) != 0 && c != '\n');
4972 to
4973 while ((c = *(++ptr)) != 0 && c != '\n') ;
4974
4975 Totally extraordinary, but if that's what it takes...
4976
4977 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4978 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4979 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4980 assumes the way PCRE uses memmove() (always moving upwards).
4981
4982 6. PCRE is being used in one environment where strchr() is not available. There
4983 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4984 faster code anyway.
4985
4986
4987 Version 3.1 09-Feb-00
4988 ---------------------
4989
4990 The only change in this release is the fixing of some bugs in Makefile.in for
4991 the "install" target:
4992
4993 (1) It was failing to install pcreposix.h.
4994
4995 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4996
4997
4998 Version 3.0 01-Feb-00
4999 ---------------------
5000
5001 1. Add support for the /+ modifier to perltest (to output $` like it does in
5002 pcretest).
5003
5004 2. Add support for the /g modifier to perltest.
5005
5006 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5007 matches null strings.
5008
5009 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5010 pattern. Perl treats empty patterns specially - it reuses the most recent
5011 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5012 effect.
5013
5014 5. The POSIX interface was broken in that it was just handing over the POSIX
5015 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5016 required a bigger vector, with some working space on the end. This means that
5017 the POSIX wrapper now has to get and free some memory, and copy the results.
5018
5019 6. Added some simple autoconf support, placing the test data and the
5020 documentation in separate directories, re-organizing some of the
5021 information files, and making it build pcre-config (a GNU standard). Also added
5022 libtool support for building PCRE as a shared library, which is now the
5023 default.
5024
5025 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5026 09 are not valid octal constants. Single digits will be used for minor values
5027 less than 10.
5028
5029 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5030 existing programs that set these in the POSIX interface can use PCRE without
5031 modification.
5032
5033 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5034 return all that pcre_info() returns, plus additional data. The pcre_info()
5035 function is retained for compatibility, but is considered to be obsolete.
5036
5037 10. Added experimental recursion feature (?R) to handle one common case that
5038 Perl 5.6 will be able to do with (?p{...}).
5039
5040 11. Added support for POSIX character classes like [:alpha:], which Perl is
5041 adopting.
5042
5043
5044 Version 2.08 31-Aug-99
5045 ----------------------
5046
5047 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5048 trying to match at the startoffset position, but instead was moving forward to
5049 the next newline as if a previous match had failed.
5050
5051 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5052 and could get into a loop if a null string was matched other than at the start
5053 of the subject.
5054
5055 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5056 be distinguished at compile time, and for completeness also added PCRE_DATE.
5057
5058 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5059 in GnuWin32 environments.
5060
5061
5062 Version 2.07 29-Jul-99
5063 ----------------------
5064
5065 1. The documentation is now supplied in plain text form and HTML as well as in
5066 the form of man page sources.
5067
5068 2. C++ compilers don't like assigning (void *) values to other pointer types.
5069 In particular this affects malloc(). Although there is no problem in Standard
5070 C, I've put in casts to keep C++ compilers happy.
5071
5072 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5073 should be (const char *).
5074
5075 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5076 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5077 However, I haven't made this a standard facility. The documentation doesn't
5078 mention it, and the Makefile doesn't support it.
5079
5080 5. The Makefile now contains an "install" target, with editable destinations at
5081 the top of the file. The pcretest program is not installed.
5082
5083 6. pgrep -V now gives the PCRE version number and date.
5084
5085 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5086 causing the entire string to be ignored, instead of just the last character.
5087
5088 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5089 non-matching string, it can take a very, very long time, even for strings of
5090 quite modest length, because of the nested recursion. PCRE now does better in
5091 some of these cases. It does this by remembering the last required literal
5092 character in the pattern, and pre-searching the subject to ensure it is present
5093 before running the real match. In other words, it applies a heuristic to detect
5094 some types of certain failure quickly, and in the above example, if presented
5095 with a string that has no trailing " it gives "no match" very quickly.
5096
5097 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5098 other alternatives are tried instead.
5099
5100
5101 Version 2.06 09-Jun-99
5102 ----------------------
5103
5104 1. Change pcretest's output for amount of store used to show just the code
5105 space, because the remainder (the data block) varies in size between 32-bit and
5106 64-bit systems.
5107
5108 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5109 start matching at. This allows lookbehinds to work when searching for multiple
5110 occurrences in a string.
5111
5112 3. Added additional options to pcretest for testing multiple occurrences:
5113
5114 /+ outputs the rest of the string that follows a match
5115 /g loops for multiple occurrences, using the new startoffset argument
5116 /G loops for multiple occurrences by passing an incremented pointer
5117
5118 4. PCRE wasn't doing the "first character" optimization for patterns starting
5119 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5120 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5121 the letter 'x'. On long subject strings, this gives a significant speed-up.
5122
5123
5124 Version 2.05 21-Apr-99
5125 ----------------------
5126
5127 1. Changed the type of magic_number from int to long int so that it works
5128 properly on 16-bit systems.
5129
5130 2. Fixed a bug which caused patterns starting with .* not to work correctly
5131 when the subject string contained newline characters. PCRE was assuming
5132 anchoring for such patterns in all cases, which is not correct because .* will
5133 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5134 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5135 must be retried after every newline in the subject.
5136
5137
5138 Version 2.04 18-Feb-99
5139 ----------------------
5140
5141 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5142 computation of the store needed to hold the pattern was incorrect (too large).
5143 If such patterns were nested a few deep, this could multiply and become a real
5144 problem.
5145
5146 2. Added /M option to pcretest to show the memory requirement of a specific
5147 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5148
5149 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5150 compiled in such a way that the backtracking after subsequent failure was
5151 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5152 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5153
5154
5155 Version 2.03 02-Feb-99
5156 ----------------------
5157
5158 1. Fixed typo and small mistake in man page.
5159
5160 2. Added 4th condition (GPL supersedes if conflict) and created separate
5161 LICENCE file containing the conditions.
5162
5163 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5164 Perl, that is the internal \ allows the delimiter to be included in the
5165 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5166 the final delimiter, add \ to the end of the pattern (to test the error).
5167
5168 4. Added the convenience functions for extracting substrings after a successful
5169 match. Updated pcretest to make it able to test these functions.
5170
5171
5172 Version 2.02 14-Jan-99
5173 ----------------------
5174
5175 1. Initialized the working variables associated with each extraction so that
5176 their saving and restoring doesn't refer to uninitialized store.
5177
5178 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5179 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5180 fix the problem.
5181
5182 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5183 calls, and wasn't printing the correct value for compiling calls. Increased the
5184 default value of LOOPREPEAT, and the number of significant figures in the
5185 times.
5186
5187 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5188
5189 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5190 a building problem on Windows NT with a FAT file system.
5191
5192
5193 Version 2.01 21-Oct-98
5194 ----------------------
5195
5196 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5197 to character tables built by pcre_maketables() in the current locale. If NULL
5198 is passed, the default tables are used.
5199
5200
5201 Version 2.00 24-Sep-98
5202 ----------------------
5203
5204 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5205 it any more.
5206
5207 2. Allow quantification of (?>) groups, and make it work correctly.
5208
5209 3. The first character computation wasn't working for (?>) groups.
5210
5211 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5212 end of the subject) and add 5.005's \z, which really does match only at the
5213 very end of the subject.
5214
5215 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5216
5217 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5218 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5219 localized options. All options to pcre_study() were also removed.
5220
5221 7. Add other new features from 5.005:
5222
5223 $(?<= positive lookbehind
5224 $(?<! negative lookbehind
5225 (?imsx-imsx) added the unsetting capability
5226 such a setting is global if at outer level; local otherwise
5227 (?imsx-imsx:) non-capturing groups with option setting
5228 (?(cond)re|re) conditional pattern matching
5229
5230 A backreference to itself in a repeated group matches the previous
5231 captured string.
5232
5233 8. General tidying up of studying (both automatic and via "study")
5234 consequential on the addition of new assertions.
5235
5236 9. As in 5.005, unlimited repeated groups that could match an empty substring
5237 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5238 runtime if any iteration does actually match an empty substring.
5239
5240 10. Include the RunTest script in the distribution.
5241
5242 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5243 discrepancies, some of which were old and were also with respect to 5.004. They
5244 have now been fixed.
5245
5246
5247 Version 1.09 28-Apr-98
5248 ----------------------
5249
5250 1. A negated single character class followed by a quantifier with a minimum
5251 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5252 program crashes, or just wrong answers. This did not apply to negated classes
5253 containing more than one character, or to minima other than one.
5254
5255
5256 Version 1.08 27-Mar-98
5257 ----------------------
5258
5259 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5260
5261 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5262 latter must appear before anything that relies on it in the pattern.
5263
5264
5265 Version 1.07 16-Feb-98
5266 ----------------------
5267
5268 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5269 repeat of a potentially empty string).
5270
5271
5272 Version 1.06 23-Jan-98
5273 ----------------------
5274
5275 1. Added Markus Oberhumer's little patches for C++.
5276
5277 2. Literal strings longer than 255 characters were broken.
5278
5279
5280 Version 1.05 23-Dec-97
5281 ----------------------
5282
5283 1. Negated character classes containing more than one character were failing if
5284 PCRE_CASELESS was set at run time.
5285
5286
5287 Version 1.04 19-Dec-97
5288 ----------------------
5289
5290 1. Corrected the man page, where some "const" qualifiers had been omitted.
5291
5292 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5293 input syntax.
5294
5295 3. Fixed memory leak which occurred when a regex with back references was
5296 matched with an offsets vector that wasn't big enough. The temporary memory
5297 that is used in this case wasn't being freed if the match failed.
5298
5299 4. Tidied pcretest to ensure it frees memory that it gets.
5300
5301 5. Temporary memory was being obtained in the case where the passed offsets
5302 vector was exactly big enough.
5303
5304 6. Corrected definition of offsetof() from change 5 below.
5305
5306 7. I had screwed up change 6 below and broken the rules for the use of
5307 setjmp(). Now fixed.
5308
5309
5310 Version 1.03 18-Dec-97
5311 ----------------------
5312
5313 1. A erroneous regex with a missing opening parenthesis was correctly
5314 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5315 on some systems.
5316
5317 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5318 it was reported that one broken compiler failed on the former because "code" is
5319 also an independent variable.
5320
5321 3. The erroneous regex a[]b caused an array overrun reference.
5322
5323 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5324 fail on data ending with that character. (It was going on too far, and checking
5325 the next character, typically a binary zero.) This was specific to the
5326 optimized code for single-character negative classes.
5327
5328 5. Added a contributed patch from the TIN world which does the following:
5329
5330 + Add an undef for memmove, in case the the system defines a macro for it.
5331
5332 + Add a definition of offsetof(), in case there isn't one. (I don't know
5333 the reason behind this - offsetof() is part of the ANSI standard - but
5334 it does no harm).
5335
5336 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5337 most of the places where whitespace preceded '#'. I have given up and
5338 allowed the remaining 2 cases to be at the margin.
5339
5340 + Rename some variables in pcre to eliminate shadowing. This seems very
5341 pedantic, but does no harm, of course.
5342
5343 6. Moved the call to setjmp() into its own function, to get rid of warnings
5344 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5345
5346 7. Constructs such as \d{8,} were compiling into the equivalent of
5347 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5348 outcome, but in this particular case used more store than had been allocated,
5349 which caused the bug to be discovered because it threw up an internal error.
5350
5351 8. The debugging code in both pcre and pcretest for outputting the compiled
5352 form of a regex was going wrong in the case of back references followed by
5353 curly-bracketed repeats.
5354
5355
5356 Version 1.02 12-Dec-97
5357 ----------------------
5358
5359 1. Typos in pcre.3 and comments in the source fixed.
5360
5361 2. Applied a contributed patch to get rid of places where it used to remove
5362 'const' from variables, and fixed some signed/unsigned and uninitialized
5363 variable warnings.
5364
5365 3. Added the "runtest" target to Makefile.
5366
5367 4. Set default compiler flag to -O2 rather than just -O.
5368
5369
5370 Version 1.01 19-Nov-97
5371 ----------------------
5372
5373 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5374 like /([ab]*)*/, that is, for classes with more than one character in them.
5375
5376 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5377 as /((?>a*))*/ (a PCRE_EXTRA facility).
5378
5379
5380 Version 1.00 18-Nov-97
5381 ----------------------
5382
5383 1. Added compile-time macros to support systems such as SunOS4 which don't have
5384 memmove() or strerror() but have other things that can be used instead.
5385
5386 2. Arranged that "make clean" removes the executables.
5387
5388
5389 Version 0.99 27-Oct-97
5390 ----------------------
5391
5392 1. Fixed bug in code for optimizing classes with only one character. It was
5393 initializing a 32-byte map regardless, which could cause it to run off the end
5394 of the memory it had got.
5395
5396 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5397
5398
5399 Version 0.98 22-Oct-97
5400 ----------------------
5401
5402 1. Fixed bug in code for handling temporary memory usage when there are more
5403 back references than supplied space in the ovector. This could cause segfaults.
5404
5405
5406 Version 0.97 21-Oct-97
5407 ----------------------
5408
5409 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5410
5411 2. Optimized negated single characters not to use a bit map.
5412
5413 3. Brought error texts together as macro definitions; clarified some of them;
5414 fixed one that was wrong - it said "range out of order" when it meant "invalid
5415 escape sequence".
5416
5417 4. Changed some char * arguments to const char *.
5418
5419 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5420
5421 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5422 pcretest.
5423
5424
5425 Version 0.96 16-Oct-97
5426 ----------------------
5427
5428 1. Added a simple "pgrep" utility to the distribution.
5429
5430 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5431 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5432 where "ddd" means "one or more decimal digits".
5433
5434 3. Fixed serious bug. If a pattern had a back reference, but the call to
5435 pcre_exec() didn't supply a large enough ovector to record the related
5436 identifying subpattern, the match always failed. PCRE now remembers the number
5437 of the largest back reference, and gets some temporary memory in which to save
5438 the offsets during matching if necessary, in order to ensure that
5439 backreferences always work.
5440
5441 4. Increased the compatibility with Perl in a number of ways:
5442
5443 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5444 to request this handling. The option can be set at compile or exec time.
5445
5446 (b) $ matches before a terminating newline by default; an option
5447 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5448 mode). The option can be set at compile or exec time.
5449
5450 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5451 the same as Perl's. If the decimal number it represents is less than 10
5452 or there aren't that many previous left capturing parentheses, an octal
5453 escape is read. Inside a character class, it's always an octal escape,
5454 even if it is a single digit.
5455
5456 (d) An escaped but undefined alphabetic character is taken as a literal,
5457 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5458 escapes.
5459
5460 (e) {0} is now permitted. (The previous item is removed from the compiled
5461 pattern).
5462
5463 5. Changed all the names of code files so that the basic parts are no longer
5464 than 10 characters, and abolished the teeny "globals.c" file.
5465
5466 6. Changed the handling of character classes; they are now done with a 32-byte
5467 bit map always.
5468
5469 7. Added the -d and /D options to pcretest to make it possible to look at the
5470 internals of compilation without having to recompile pcre.
5471
5472
5473 Version 0.95 23-Sep-97
5474 ----------------------
5475
5476 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5477 \x20 at the start of a run of normal characters. These were being treated as
5478 real characters, instead of the source characters being re-checked.
5479
5480
5481 Version 0.94 18-Sep-97
5482 ----------------------
5483
5484 1. The functions are now thread-safe, with the caveat that the global variables
5485 containing pointers to malloc() and free() or alternative functions are the
5486 same for all threads.
5487
5488 2. Get pcre_study() to generate a bitmap of initial characters for non-
5489 anchored patterns when this is possible, and use it if passed to pcre_exec().
5490
5491
5492 Version 0.93 15-Sep-97
5493 ----------------------
5494
5495 1. /(b)|(:+)/ was computing an incorrect first character.
5496
5497 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5498 but not actually doing anything yet.
5499
5500 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5501 as Perl does (e.g. [-az] or [az-]).
5502
5503 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5504 all possible positions.
5505
5506 5. Split up into different modules to avoid including unneeded functions in a
5507 compiled binary. However, compile and exec are still in one module. The "study"
5508 function is split off.
5509
5510 6. The character tables are now in a separate module whose source is generated
5511 by an auxiliary program - but can then be edited by hand if required. There are
5512 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5513 toupper() in the code.
5514
5515 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5516 make them global. Abolish the function for setting them, as the caller can now
5517 set them directly.
5518
5519
5520 Version 0.92 11-Sep-97
5521 ----------------------
5522
5523 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5524 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5525
5526 2. Caseless matching was not working in character classes if the characters in
5527 the pattern were in upper case.
5528
5529 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5530
5531 4. Make PCRE_ANCHORED public and accept as a compile option.
5532
5533 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5534 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5535 pass them.
5536
5537 6. Give an error if bad option bits passed at compile or run time.
5538
5539 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5540 pcretest to cause it to pass that flag.
5541
5542 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5543 options, and the first character, if set.
5544
5545 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5546
5547
5548 Version 0.91 10-Sep-97
5549 ----------------------
5550
5551 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5552 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5553
5554 2. PCRE was looping on encountering an indefinitely repeated back reference to
5555 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5556 Perl does - treats the match as successful.
5557
5558 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5