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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1485 - (show annotations)
Wed Jun 18 16:31:32 2014 UTC (5 years, 4 months ago) by ph10
File size: 266946 byte(s)
Error occurred while calculating annotation data.
Fix bad starting data when char with more than one other case follows 
circumflex in multiline UTF mode.
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
69 Version 8.35 04-April-2014
70 --------------------------
71
72 1. A new flag is set, when property checks are present in an XCLASS.
73 When this flag is not set, PCRE can perform certain optimizations
74 such as studying these XCLASS-es.
75
76 2. The auto-possessification of character sets were improved: a normal
77 and an extended character set can be compared now. Furthermore
78 the JIT compiler optimizes more character set checks.
79
80 3. Got rid of some compiler warnings for potentially uninitialized variables
81 that show up only when compiled with -O2.
82
83 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
84 of a match later then the end of the match. The pcretest program was not
85 handling the case sensibly - it was outputting from the start to the next
86 binary zero. It now reports this situation in a message, and outputs the
87 text from the end to the start.
88
89 5. Fast forward search is improved in JIT. Instead of the first three
90 characters, any three characters with fixed position can be searched.
91 Search order: first, last, middle.
92
93 6. Improve character range checks in JIT. Characters are read by an inprecise
94 function now, which returns with an unknown value if the character code is
95 above a certain treshold (e.g: 256). The only limitation is that the value
96 must be bigger than the treshold as well. This function is useful, when
97 the characters above the treshold are handled in the same way.
98
99 7. The macros whose names start with RAWUCHAR are placeholders for a future
100 mode in which only the bottom 21 bits of 32-bit data items are used. To
101 make this more memorable for those maintaining the code, the names have
102 been changed to start with UCHAR21, and an extensive comment has been added
103 to their definition.
104
105 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
106 to the export list in Makefile.am (they were accidentally omitted from the
107 8.34 tarball).
108
109 9. The informational output from pcretest used the phrase "starting byte set"
110 which is inappropriate for the 16-bit and 32-bit libraries. As the output
111 for "first char" and "need char" really means "non-UTF-char", I've changed
112 "byte" to "char", and slightly reworded the output. The documentation about
113 these values has also been (I hope) clarified.
114
115 10. Another JIT related optimization: use table jumps for selecting the correct
116 backtracking path, when more than four alternatives are present inside a
117 bracket.
118
119 11. Empty match is not possible, when the minimum length is greater than zero,
120 and there is no \K in the pattern. JIT should avoid empty match checks in
121 such cases.
122
123 12. In a caseless character class with UCP support, when a character with more
124 than one alternative case was not the first character of a range, not all
125 the alternative cases were added to the class. For example, s and \x{17f}
126 are both alternative cases for S: the class [RST] was handled correctly,
127 but [R-T] was not.
128
129 13. The configure.ac file always checked for pthread support when JIT was
130 enabled. This is not used in Windows, so I have put this test inside a
131 check for the presence of windows.h (which was already tested for).
132
133 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
134 The algorithm provides a way to skip certain starting offsets, and usually
135 faster than linear prefix searches.
136
137 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
138 as for 'fr_FR' and 'french'. For some reason, however, it then used the
139 Windows-specific input and output files, which have 'french' screwed in.
140 So this could never have worked. One of the problems with locales is that
141 they aren't always the same. I have now updated RunTest so that it checks
142 the output of the locale test (test 3) against three different output
143 files, and it allows the test to pass if any one of them matches. With luck
144 this should make the test pass on some versions of Solaris where it was
145 failing. Because of the uncertainty, the script did not used to stop if
146 test 3 failed; it now does. If further versions of a French locale ever
147 come to light, they can now easily be added.
148
149 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
150 there was a message during ./configure, but it did not stop. This now
151 provokes an error. The invalid example in README has been corrected.
152 If a value less than the minimum is given, the minimum value has always
153 been used, but now a warning is given.
154
155 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
156 was a bug in the test system, which is now fixed. Also, the list of various
157 configurations that are tested for each release did not have one with both
158 16/32 bits and --enable-bar-anycrlf. It now does.
159
160 18. pcretest was missing "-C bsr" for displaying the \R default setting.
161
162 19. Little endian PowerPC systems are supported now by the JIT compiler.
163
164 20. The fast forward newline mechanism could enter to an infinite loop on
165 certain invalid UTF-8 input. Although we don't support these cases
166 this issue can be fixed by a performance optimization.
167
168 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
169 not take account if existing stack usage. There is now a new global
170 variable called pcre_stack_guard that can be set to point to an external
171 function to check stack availability. It is called at the start of
172 processing every parenthesized group.
173
174 22. A typo in the code meant that in ungreedy mode the max/min qualifier
175 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
176 not match "ab".
177
178 23. When UTF was disabled, the JIT program reported some incorrect compile
179 errors. These messages are silenced now.
180
181 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
182 compiler.
183
184 25. Change all the temporary files used in RunGrepTest to be different to those
185 used by RunTest so that the tests can be run simultaneously, for example by
186 "make -j check".
187
188
189 Version 8.34 15-December-2013
190 -----------------------------
191
192 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
193 executable memory. Patch inspired by Carsten Klein.
194
195 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
196 this macro is never tested and has no effect, because the work to support
197 coverage involves only compiling and linking options and special targets in
198 the Makefile. The comment in config.h implied that defining the macro would
199 enable coverage support, which is totally false. There was also support for
200 setting this macro in the CMake files (my fault, I just copied it from
201 configure). SUPPORT_GCOV has now been removed.
202
203 3. Make a small performance improvement in strlen16() and strlen32() in
204 pcretest.
205
206 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
207 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
208 these cases). There was also one in pcretest.c.
209
210 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
211
212 6. In UTF mode, the code for checking whether a group could match an empty
213 string (which is used for indefinitely repeated groups to allow for
214 breaking an infinite loop) was broken when the group contained a repeated
215 negated single-character class with a character that occupied more than one
216 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
217 UTF-8 mode). The effect was undefined: the group might or might not be
218 deemed as matching an empty string, or the program might have crashed.
219
220 7. The code for checking whether a group could match an empty string was not
221 recognizing that \h, \H, \v, \V, and \R must match a character.
222
223 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
224 an empty string. If it can, pcretest shows this in its information output.
225
226 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
227 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
228 matched by pcre_exec() without using JIT:
229
230 (a) If the rest of the pattern did not match after a maximal run of
231 grapheme clusters, the code for backing up to try with fewer of them
232 did not always back up over a full grapheme when characters that do not
233 have the modifier quality were involved, e.g. Hangul syllables.
234
235 (b) If the match point in a subject started with modifier character, and
236 there was no match, the code could incorrectly back up beyond the match
237 point, and potentially beyond the first character in the subject,
238 leading to a segfault or an incorrect match result.
239
240 10. A conditional group with an assertion condition could lead to PCRE
241 recording an incorrect first data item for a match if no other first data
242 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
243 first data item, and therefore matched "ca" after "c" instead of at the
244 start.
245
246 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
247 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
248
249 12. The source of pcregrep now includes z/OS-specific code so that it can be
250 compiled for z/OS as part of the special z/OS distribution.
251
252 13. Added the -T and -TM options to pcretest.
253
254 14. The code in pcre_compile.c for creating the table of named capturing groups
255 has been refactored. Instead of creating the table dynamically during the
256 actual compiling pass, the information is remembered during the pre-compile
257 pass (on the stack unless there are more than 20 named groups, in which
258 case malloc() is used) and the whole table is created before the actual
259 compile happens. This has simplified the code (it is now nearly 150 lines
260 shorter) and prepared the way for better handling of references to groups
261 with duplicate names.
262
263 15. A back reference to a named subpattern when there is more than one of the
264 same name now checks them in the order in which they appear in the pattern.
265 The first one that is set is used for the reference. Previously only the
266 first one was inspected. This change makes PCRE more compatible with Perl.
267
268 16. Unicode character properties were updated from Unicode 6.3.0.
269
270 17. The compile-time code for auto-possessification has been refactored, based
271 on a patch by Zoltan Herczeg. It now happens after instead of during
272 compilation. The code is cleaner, and more cases are handled. The option
273 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
274 options in pcretest are provided to set it. It can also be set by
275 (*NO_AUTO_POSSESS) at the start of a pattern.
276
277 18. The character VT has been added to the default ("C" locale) set of
278 characters that match \s and are generally treated as white space,
279 following this same change in Perl 5.18. There is now no difference between
280 "Perl space" and "POSIX space". Whether VT is treated as white space in
281 other locales depends on the locale.
282
283 19. The code for checking named groups as conditions, either for being set or
284 for being recursed, has been refactored (this is related to 14 and 15
285 above). Processing unduplicated named groups should now be as fast at
286 numerical groups, and processing duplicated groups should be faster than
287 before.
288
289 20. Two patches to the CMake build system, by Alexander Barkov:
290
291 (1) Replace the "source" command by "." in CMakeLists.txt because
292 "source" is a bash-ism.
293
294 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
295 without these the CMake build does not work on Solaris.
296
297 21. Perl has changed its handling of \8 and \9. If there is no previously
298 encountered capturing group of those numbers, they are treated as the
299 literal characters 8 and 9 instead of a binary zero followed by the
300 literals. PCRE now does the same.
301
302 22. Following Perl, added \o{} to specify codepoints in octal, making it
303 possible to specify values greater than 0777 and also making them
304 unambiguous.
305
306 23. Perl now gives an error for missing closing braces after \x{... instead of
307 treating the string as literal. PCRE now does the same.
308
309 24. RunTest used to grumble if an inappropriate test was selected explicitly,
310 but just skip it when running all tests. This make it awkward to run ranges
311 of tests when one of them was inappropriate. Now it just skips any
312 inappropriate tests, as it always did when running all tests.
313
314 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
315 character types such as \d or \w, too many callouts were inserted, and the
316 data that they returned was rubbish.
317
318 26. In UCP mode, \s was not matching two of the characters that Perl matches,
319 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
320 were matched by \h. The code has now been refactored so that the lists of
321 the horizontal and vertical whitespace characters used for \h and \v (which
322 are defined only in one place) are now also used for \s.
323
324 27. Add JIT support for the 64 bit TileGX architecture.
325 Patch by Jiong Wang (Tilera Corporation).
326
327 28. Possessive quantifiers for classes (both explicit and automatically
328 generated) now use special opcodes instead of wrapping in ONCE brackets.
329
330 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
331 (because it's meaningless), this was not happening when PCRE_CASELESS was
332 set. Not wrong, but inefficient.
333
334 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
335 properties for \w, \d, etc) is present in a test regex. Otherwise if the
336 test contains no characters greater than 255, Perl doesn't realise it
337 should be using Unicode semantics.
338
339 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
340 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
341 does in Unicode mode.
342
343 32. Added the "forbid" facility to pcretest so that putting tests into the
344 wrong test files can sometimes be quickly detected.
345
346 33. There is now a limit (default 250) on the depth of nesting of parentheses.
347 This limit is imposed to control the amount of system stack used at compile
348 time. It can be changed at build time by --with-parens-nest-limit=xxx or
349 the equivalent in CMake.
350
351 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
352 errors. Perl warns for these when in warning mode, but PCRE has no facility
353 for giving warnings.
354
355 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
356 However, this was not working for (?!) because it is optimized to (*FAIL),
357 for which PCRE does not allow quantifiers. The optimization is now disabled
358 when a quantifier follows (?!). I can't see any use for this, but it makes
359 things uniform.
360
361 36. Perl no longer allows group names to start with digits, so I have made this
362 change also in PCRE. It simplifies the code a bit.
363
364 37. In extended mode, Perl ignores spaces before a + that indicates a
365 possessive quantifier. PCRE allowed a space before the quantifier, but not
366 before the possessive +. It now does.
367
368 38. The use of \K (reset reported match start) within a repeated possessive
369 group such as (a\Kb)*+ was not working.
370
371 40. Document that the same character tables must be used at compile time and
372 run time, and that the facility to pass tables to pcre_exec() and
373 pcre_dfa_exec() is for use only with saved/restored patterns.
374
375 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
376 features for Builds with MSVC:
377
378 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
379 builds). The .rc files can be used to set FileDescription and many other
380 attributes.
381
382 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
383 This allows higher-level build scripts which want .pdb files to avoid
384 hard-coding the exact files needed."
385
386 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
387 mean "start of word" and "end of word", respectively, as a transition aid.
388
389 43. A minimizing repeat of a class containing codepoints greater than 255 in
390 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
391 compiled to use the heap for recursion.
392
393 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
394 is configured.
395
396
397 Version 8.33 28-May-2013
398 ------------------------
399
400 1. Added 'U' to some constants that are compared to unsigned integers, to
401 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
402 variables that are added to signed variables, to ensure the result is
403 signed and can be negated.
404
405 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
406 CMake config files.
407
408 3. Revise the creation of config.h.generic so that all boolean macros are
409 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
410 overriding via -D on the command line possible.
411
412 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
413 to unsigned int is reported to make a quite noticeable speed difference in
414 a specific Windows environment. Testing on Linux did also appear to show
415 some benefit (and it is clearly not harmful). Also fixed the definition of
416 Xop which should be unsigned.
417
418 5. Related to (4), changing the definition of the intermediate variable cc
419 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
420 improvements.
421
422 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
423 unnecessary spaces.
424
425 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
426 and later.
427
428 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
429
430 9. Optimizing fast_forward_start_bits in JIT.
431
432 10. Adding support for callouts in JIT, and fixing some issues revealed
433 during this work. Namely:
434
435 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
436
437 (b) Minimum length was not checked before the matching is started.
438
439 11. The value of capture_last that is passed to callouts was incorrect in some
440 cases when there was a capture on one path that was subsequently abandoned
441 after a backtrack. Also, the capture_last value is now reset after a
442 recursion, since all captures are also reset in this case.
443
444 12. The interpreter no longer returns the "too many substrings" error in the
445 case when an overflowing capture is in a branch that is subsequently
446 abandoned after a backtrack.
447
448 13. In the pathological case when an offset vector of size 2 is used, pcretest
449 now prints out the matched string after a yield of 0 or 1.
450
451 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
452 Only supported by the JIT compiler at the moment.
453
454 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
455
456 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
457 offset of the starting point of the matching process, provided the offsets
458 vector is large enough.
459
460 17. The \A escape now records a lookbehind value of 1, though its execution
461 does not actually inspect the previous character. This is to ensure that,
462 in partial multi-segment matching, at least one character from the old
463 segment is retained when a new segment is processed. Otherwise, if there
464 are no lookbehinds in the pattern, \A might match incorrectly at the start
465 of a new segment.
466
467 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
468
469 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
470 gives some modest performance improvement in 8-bit mode.
471
472 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
473 be expressed in certain programming languages using Universal Character
474 Names.
475
476 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
477 which points out that "non characters" are not "characters that may not
478 appear in Unicode strings" but rather "characters that are reserved for
479 internal use and have only local meaning".
480
481 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
482 there was a conditional group that depended on an assertion, if the
483 assertion was false, the callout that immediately followed the alternation
484 in the condition was skipped when pcre_exec() was used for matching.
485
486 23. Allow an explicit callout to be inserted before an assertion that is the
487 condition for a conditional group, for compatibility with automatic
488 callouts, which always insert a callout at this point.
489
490 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
491 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
492
493 25. (*PRUNE) is now supported by the JIT compiler.
494
495 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
496
497 27. Fix the case where there are two or more SKIPs with arguments that may be
498 ignored.
499
500 28. (*SKIP) is now supported by the JIT compiler.
501
502 29. (*THEN) is now supported by the JIT compiler.
503
504 30. Update RunTest with additional test selector options.
505
506 31. The way PCRE handles backtracking verbs has been changed in two ways.
507
508 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
509 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
510 backtracking. In some cases this makes it more Perl-compatible, but Perl's
511 rather obscure rules do not always do the same thing.
512
513 (2) Previously, backtracking verbs were confined within assertions. This is
514 no longer the case for positive assertions, except for (*ACCEPT). Again,
515 this sometimes improves Perl compatibility, and sometimes does not.
516
517 32. A number of tests that were in test 2 because Perl did things differently
518 have been moved to test 1, because either Perl or PCRE has changed, and
519 these tests are now compatible.
520
521 32. Backtracking control verbs are now handled in the same way in JIT and
522 interpreter.
523
524 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
525 contained a forward subroutine reference caused a compile error.
526
527 34. Auto-detect and optimize limited repetitions in JIT.
528
529 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
530 blocking (*UTF) etc.
531
532 36. In the interpreter, maximizing pattern repetitions for characters and
533 character types now use tail recursion, which reduces stack usage.
534
535 37. The value of the max lookbehind was not correctly preserved if a compiled
536 and saved regex was reloaded on a host of different endianness.
537
538 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
539 of the compiled pattern block, expand the flags field from 16 to 32 bits
540 because it was almost full.
541
542 39. Try madvise first before posix_madvise.
543
544 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
545 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
546 original change locked out all matches of empty strings. This has been
547 changed so that one match of an empty string per line is recognized.
548 Subsequent searches on the same line (for colouring or for --only-matching,
549 for example) do not recognize empty strings.
550
551 41. Applied a user patch to fix a number of spelling mistakes in comments.
552
553 42. Data lines longer than 65536 caused pcretest to crash.
554
555 43. Clarified the data type for length and startoffset arguments for pcre_exec
556 and pcre_dfa_exec in the function-specific man pages, where they were
557 explicitly stated to be in bytes, never having been updated. I also added
558 some clarification to the pcreapi man page.
559
560 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
561 a segmentation fault.
562
563
564 Version 8.32 30-November-2012
565 -----------------------------
566
567 1. Improved JIT compiler optimizations for first character search and single
568 character iterators.
569
570 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
571 Patch by Daniel Richard G.
572
573 3. Single character iterator optimizations in the JIT compiler.
574
575 4. Improved JIT compiler optimizations for character ranges.
576
577 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
578 Reported by Giuseppe D'Angelo.
579
580 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
581 of a line, was being set incorrectly in cases where .* appeared inside
582 atomic brackets at the start of a pattern, or where there was a subsequent
583 *PRUNE or *SKIP.
584
585 7. Improved instruction cache flush for POWER/PowerPC.
586 Patch by Daniel Richard G.
587
588 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
589 grep:
590
591 (a) There is now no limit to the number of patterns to be matched.
592
593 (b) An error is given if a pattern is too long.
594
595 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
596 are now supported.
597
598 (d) --exclude-from and --include-from (multiple use) have been added.
599
600 (e) Exclusions and inclusions now apply to all files and directories, not
601 just to those obtained from scanning a directory recursively.
602
603 (f) Multiple uses of -f and --file-list are now supported.
604
605 (g) In a Windows environment, the default for -d has been changed from
606 "read" (the GNU grep default) to "skip", because otherwise the presence
607 of a directory in the file list provokes an error.
608
609 (h) The documentation has been revised and clarified in places.
610
611 9. Improve the matching speed of capturing brackets.
612
613 10. Changed the meaning of \X so that it now matches a Unicode extended
614 grapheme cluster.
615
616 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
617 out POSIX threads when JIT support is configured.
618
619 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
620
621 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
622 structure could go wrong in environments where size_t is not the same size
623 as int.
624
625 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
626 set.
627
628 15. The EBCDIC support had decayed; later updates to the code had included
629 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
630 general tidy up of EBCDIC-related issues, and the documentation was also
631 not quite right. There is now a test that can be run on ASCII systems to
632 check some of the EBCDIC-related things (but is it not a full test).
633
634 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
635 in a small tidy to the code.
636
637 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
638
639 18. If the --only-matching (-o) option in pcregrep is specified multiple
640 times, each one causes appropriate output. For example, -o1 -o2 outputs the
641 substrings matched by the 1st and 2nd capturing parentheses. A separating
642 string can be specified by --om-separator (default empty).
643
644 19. Improving the first n character searches.
645
646 20. Turn case lists for horizontal and vertical white space into macros so that
647 they are defined only once.
648
649 21. This set of changes together give more compatible Unicode case-folding
650 behaviour for characters that have more than one other case when UCP
651 support is available.
652
653 (a) The Unicode property table now has offsets into a new table of sets of
654 three or more characters that are case-equivalent. The MultiStage2.py
655 script that generates these tables (the pcre_ucd.c file) now scans
656 CaseFolding.txt instead of UnicodeData.txt for character case
657 information.
658
659 (b) The code for adding characters or ranges of characters to a character
660 class has been abstracted into a generalized function that also handles
661 case-independence. In UTF-mode with UCP support, this uses the new data
662 to handle characters with more than one other case.
663
664 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
665 whose other case is greater than 256 are now correctly matched
666 caselessly. Previously, the high codepoint matched the low one, but not
667 vice versa.
668
669 (d) The processing of \h, \H, \v, and \ in character classes now makes use
670 of the new class addition function, using character lists defined as
671 macros alongside the case definitions of 20 above.
672
673 (e) Caseless back references now work with characters that have more than
674 one other case.
675
676 (f) General caseless matching of characters with more than one other case
677 is supported.
678
679 22. Unicode character properties were updated from Unicode 6.2.0
680
681 23. Improved CMake support under Windows. Patch by Daniel Richard G.
682
683 24. Add support for 32-bit character strings, and UTF-32
684
685 25. Major JIT compiler update (code refactoring and bugfixing).
686 Experimental Sparc 32 support is added.
687
688 26. Applied a modified version of Daniel Richard G's patch to create
689 pcre.h.generic and config.h.generic by "make" instead of in the
690 PrepareRelease script.
691
692 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
693 pcre_compile.c when checking for a zero character.
694
695 28. Introducing a native interface for JIT. Through this interface, the compiled
696 machine code can be directly executed. The purpose of this interface is to
697 provide fast pattern matching, so several sanity checks are not performed.
698 However, feature tests are still performed. The new interface provides
699 1.4x speedup compared to the old one.
700
701 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
702 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
703 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
704
705 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
706 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
707 the "old" RFC 2279). Instead, it ended up passing a negative length to
708 pcre_exec().
709
710 31. Add support for GCC's visibility feature to hide internal functions.
711
712 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
713 "unknown -C option" after outputting 0 or 1.
714
715 33. There is now support for generating a code coverage report for the test
716 suite in environments where gcc is the compiler and lcov is installed. This
717 is mainly for the benefit of the developers.
718
719 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
720 unaddressable using valgrind annotations, allowing valgrind to detect
721 invalid memory accesses. This is mainly for the benefit of the developers.
722
723 25. (*UTF) can now be used to start a pattern in any of the three libraries.
724
725 26. Give configure error if --enable-cpp but no C++ compiler found.
726
727
728 Version 8.31 06-July-2012
729 -------------------------
730
731 1. Fixing a wrong JIT test case and some compiler warnings.
732
733 2. Removed a bashism from the RunTest script.
734
735 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
736 to unsigned type, result still unsigned" that was given by an MS compiler
737 on encountering the code "-sizeof(xxx)".
738
739 4. Partial matching support is added to the JIT compiler.
740
741 5. Fixed several bugs concerned with partial matching of items that consist
742 of more than one character:
743
744 (a) /^(..)\1/ did not partially match "aba" because checking references was
745 done on an "all or nothing" basis. This also applied to repeated
746 references.
747
748 (b) \R did not give a hard partial match if \r was found at the end of the
749 subject.
750
751 (c) \X did not give a hard partial match after matching one or more
752 characters at the end of the subject.
753
754 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
755 a partial match for the string "\r".
756
757 (e) When newline was set to CRLF, the metacharacter "." did not recognize
758 a partial match for a CR character at the end of the subject string.
759
760 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
761 running pcretest, the text "(JIT)" added to the output whenever JIT is
762 actually used to run the match.
763
764 7. Individual JIT compile options can be set in pcretest by following -s+[+]
765 or /S+[+] with a digit between 1 and 7.
766
767 8. OP_NOT now supports any UTF character not just single-byte ones.
768
769 9. (*MARK) control verb is now supported by the JIT compiler.
770
771 10. The command "./RunTest list" lists the available tests without actually
772 running any of them. (Because I keep forgetting what they all are.)
773
774 11. Add PCRE_INFO_MAXLOOKBEHIND.
775
776 12. Applied a (slightly modified) user-supplied patch that improves performance
777 when the heap is used for recursion (compiled with --disable-stack-for-
778 recursion). Instead of malloc and free for each heap frame each time a
779 logical recursion happens, frames are retained on a chain and re-used where
780 possible. This sometimes gives as much as 30% improvement.
781
782 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
783 call.
784
785 14. As documented, (*COMMIT) is now confined to within a positive assertion.
786
787 15. It is now possible to link pcretest with libedit as an alternative to
788 libreadline.
789
790 16. (*COMMIT) control verb is now supported by the JIT compiler.
791
792 17. The Unicode data tables have been updated to Unicode 6.1.0.
793
794 18. Added --file-list option to pcregrep.
795
796 19. Added binary file support to pcregrep, including the -a, --binary-files,
797 -I, and --text options.
798
799 20. The madvise function is renamed for posix_madvise for QNX compatibility
800 reasons. Fixed by Giuseppe D'Angelo.
801
802 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
803
804 22. Changed the option for creating symbolic links for 16-bit man pages from
805 -s to -sf so that re-installing does not cause issues.
806
807 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
808
809 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
810 matches in certain environments (the workspace was not being correctly
811 retained). Also added to pcre_dfa_exec() a simple plausibility check on
812 some of the workspace data at the beginning of a restart.
813
814 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
815 was not doing so when it should - probably a typo introduced by SVN 528
816 (change 8.10/14).
817
818 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
819 \w+ when the character tables indicated that \x{c4} was a word character.
820 There were several related cases, all because the tests for doing a table
821 lookup were testing for characters less than 127 instead of 255.
822
823 27. If a pattern contains capturing parentheses that are not used in a match,
824 their slots in the ovector are set to -1. For those that are higher than
825 any matched groups, this happens at the end of processing. In the case when
826 there were back references that the ovector was too small to contain
827 (causing temporary malloc'd memory to be used during matching), and the
828 highest capturing number was not used, memory off the end of the ovector
829 was incorrectly being set to -1. (It was using the size of the temporary
830 memory instead of the true size.)
831
832 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
833 ovector size, it uses memory at the end of the block that it has got.
834
835 29. Check for an overlong MARK name and give an error at compile time. The
836 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
837
838 30. JIT compiler update.
839
840 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
841 Rill for the patch.
842
843 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
844
845 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
846
847 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
848 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
849 was enabled.
850
851 35. Improve JIT code generation for greedy plus quantifier.
852
853 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
854 1 to "aa" instead of to an empty string. The bug affected repeated groups
855 that could potentially match an empty string.
856
857 37. Optimizing single character iterators in JIT.
858
859 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
860 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
861 codepoints that are too big for the mode are faulted, and in a UTF mode,
862 disallowed codepoints are also faulted.
863
864 39. If PCRE was compiled with UTF support, in three places in the DFA
865 matcher there was code that should only have been obeyed in UTF mode, but
866 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
867 processing when bytes with values greater than 127 were present. In 16-bit
868 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
869 both cases the values are those that cannot be the first data item in a UTF
870 character. The three items that might have provoked this were recursions,
871 possessively repeated groups, and atomic groups.
872
873 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
874 and pcregrep, because some OS require shared objects to be explicitly
875 passed to ld, causing the link step to fail if they are not.
876
877 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
878 mode, patterns that started with \h* or \R* might be incorrectly matched.
879
880
881 Version 8.30 04-February-2012
882 -----------------------------
883
884 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
885 name is defined in ctype.h.
886
887 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
888 only in quite long subpatterns.
889
890 3. Removed the function pcre_info(), which has been obsolete and deprecated
891 since it was replaced by pcre_fullinfo() in February 2000.
892
893 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
894 match a (*MARK), and the match failed at the start of the subject, a
895 reference to memory before the start of the subject could occur. This bug
896 was introduced by fix 17 of release 8.21.
897
898 5. A reference to an unset group with zero minimum repetition was giving
899 totally wrong answers (in non-JavaScript-compatibility mode). For example,
900 /(another)?(\1?)test/ matched against "hello world test". This bug was
901 introduced in release 8.13.
902
903 6. Add support for 16-bit character strings (a large amount of work involving
904 many changes and refactorings).
905
906 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
907 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
908 from a file.
909
910 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
911 rounding is not applied in this particular case).
912
913 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
914 if they appear, or are escaped, in patterns.
915
916 10. Get rid of a number of -Wunused-but-set-variable warnings.
917
918 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
919 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
920 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
921 also returns the mark "x". This bug applied to capturing parentheses,
922 non-capturing parentheses, and atomic parentheses. It also applied to some
923 assertions.
924
925 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
926 information out of configure.ac instead of relying on pcre.h.generic, which
927 is not stored in the repository.
928
929 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
930 -lreadline.
931
932 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
933
934 15. Applied Graycode's patch to put the top-level frame on the stack rather
935 than the heap when not using the stack for recursion. This gives a
936 performance improvement in many cases when recursion is not deep.
937
938 16. Experimental code added to "pcretest -C" to output the stack frame size.
939
940
941 Version 8.21 12-Dec-2011
942 ------------------------
943
944 1. Updating the JIT compiler.
945
946 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
947 are added as well.
948
949 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
950 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
951 calling _pcre_jit_exec. Some extra comments are added.
952
953 4. (*MARK) settings inside atomic groups that do not contain any capturing
954 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
955 was introduced by change 18 for 8.20.
956
957 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
958 ECMA-262 standard.
959
960 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
961 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
962 This bug was probably introduced by change 9 of 8.13.
963
964 7. While fixing 6 above, I noticed that a number of other items were being
965 incorrectly rejected as "not fixed length". This arose partly because newer
966 opcodes had not been added to the fixed-length checking code. I have (a)
967 corrected the bug and added tests for these items, and (b) arranged for an
968 error to occur if an unknown opcode is encountered while checking for fixed
969 length instead of just assuming "not fixed length". The items that were
970 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
971 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
972 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
973
974 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
975 being incorrectly compiled and would have given unpredicatble results.
976
977 9. A possessively repeated subpattern with minimum repeat count greater than
978 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
979 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
980 the first (A) could occur when it should not.
981
982 10. Add a cast and remove a redundant test from the code.
983
984 11. JIT should use pcre_malloc/pcre_free for allocation.
985
986 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
987 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
988 is anything other than /usr, -L is still shown).
989
990 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
991
992 14. Perl does not support \N without a following name in a [] class; PCRE now
993 also gives an error.
994
995 15. If a forward reference was repeated with an upper limit of around 2000,
996 it caused the error "internal error: overran compiling workspace". The
997 maximum number of forward references (including repeats) was limited by the
998 internal workspace, and dependent on the LINK_SIZE. The code has been
999 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1000 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1001 of around 200,000 forward references. While doing this, I also speeded up
1002 the filling in of repeated forward references.
1003
1004 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1005 incorrectly expecting the subject to contain another "a" after the start.
1006
1007 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1008 in the match, the SKIP should be ignored. This was not happening; instead
1009 the SKIP was being treated as NOMATCH. For patterns such as
1010 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1011 tested.
1012
1013 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1014 now much more compatible with Perl, in particular in cases where the result
1015 is a non-match for a non-anchored pattern. For example, if
1016 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1017 "m", where previously it did not return a name. A side effect of this
1018 change is that for partial matches, the last encountered mark name is
1019 returned, as for non matches. A number of tests that were previously not
1020 Perl-compatible have been moved into the Perl-compatible test files. The
1021 refactoring has had the pleasing side effect of removing one argument from
1022 the match() function, thus reducing its stack requirements.
1023
1024 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1025 subsequent uses of /S (without +) incorrectly behaved like /S+.
1026
1027 21. Retrieve executable code size support for the JIT compiler and fixing
1028 some warnings.
1029
1030 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1031 not work when the shorter character appeared right at the end of the
1032 subject string.
1033
1034 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1035 systems.
1036
1037 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1038 output it when the /M option is used in pcretest.
1039
1040 25. The CheckMan script was not being included in the distribution. Also, added
1041 an explicit "perl" to run Perl scripts from the PrepareRelease script
1042 because this is reportedly needed in Windows.
1043
1044 26. If study data was being save in a file and studying had not found a set of
1045 "starts with" bytes for the pattern, the data written to the file (though
1046 never used) was taken from uninitialized memory and so caused valgrind to
1047 complain.
1048
1049 27. Updated RunTest.bat as provided by Sheri Pierce.
1050
1051 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1052
1053 29. Computation of memory usage for the table of capturing group names was
1054 giving an unnecessarily large value.
1055
1056
1057 Version 8.20 21-Oct-2011
1058 ------------------------
1059
1060 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1061 a POSIX class. After further experiments with Perl, which convinced me that
1062 Perl has bugs and confusions, a closing square bracket is no longer allowed
1063 in a POSIX name. This bug also affected patterns with classes that started
1064 with full stops.
1065
1066 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1067 captured substring, but while checking the failing first alternative,
1068 substring 1 is temporarily captured. If the output vector supplied to
1069 pcre_exec() was not big enough for this capture, the yield of the function
1070 was still zero ("insufficient space for captured substrings"). This cannot
1071 be totally fixed without adding another stack variable, which seems a lot
1072 of expense for a edge case. However, I have improved the situation in cases
1073 such as /(a)(b)x|abc/ matched against "abc", where the return code
1074 indicates that fewer than the maximum number of slots in the ovector have
1075 been set.
1076
1077 3. Related to (2) above: when there are more back references in a pattern than
1078 slots in the output vector, pcre_exec() uses temporary memory during
1079 matching, and copies in the captures as far as possible afterwards. It was
1080 using the entire output vector, but this conflicts with the specification
1081 that only 2/3 is used for passing back captured substrings. Now it uses
1082 only the first 2/3, for compatibility. This is, of course, another edge
1083 case.
1084
1085 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1086 main code base, and can be used by building with --enable-jit. When this is
1087 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1088 runtime --no-jit option is given.
1089
1090 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1091 ovector, the return from the function was zero, implying that there were
1092 other matches that did not fit. The correct "exactly full" value is now
1093 returned.
1094
1095 6. If a subpattern that was called recursively or as a subroutine contained
1096 (*PRUNE) or any other control that caused it to give a non-standard return,
1097 invalid errors such as "Error -26 (nested recursion at the same subject
1098 position)" or even infinite loops could occur.
1099
1100 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1101 computing the minimum length on reaching *ACCEPT, and so ended up with the
1102 wrong value of 1 rather than 0. Further investigation indicates that
1103 computing a minimum subject length in the presence of *ACCEPT is difficult
1104 (think back references, subroutine calls), and so I have changed the code
1105 so that no minimum is registered for a pattern that contains *ACCEPT.
1106
1107 8. If (*THEN) was present in the first (true) branch of a conditional group,
1108 it was not handled as intended. [But see 16 below.]
1109
1110 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1111 Sheri Pierce.
1112
1113 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1114 the first byte in a match must be "a".
1115
1116 11. Change 17 for 8.13 increased the recursion depth for patterns like
1117 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1118 pattern contains any instances of (*THEN). If it does not, the old
1119 optimizations are restored. It would be nice to do this on a per-group
1120 basis, but at the moment that is not feasible.
1121
1122 12. In some environments, the output of pcretest -C is CRLF terminated. This
1123 broke RunTest's code that checks for the link size. A single white space
1124 character after the value is now allowed for.
1125
1126 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1127 For "fr", it uses the Windows-specific input and output files.
1128
1129 14. If (*THEN) appeared in a group that was called recursively or as a
1130 subroutine, it did not work as intended. [But see next item.]
1131
1132 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1133 pattern fragments (but not containing any | characters). If A and B are
1134 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1135 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1136 D. In other words, Perl considers parentheses that do not contain any |
1137 characters to be part of a surrounding alternative, whereas PCRE was
1138 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1139 differently. PCRE now behaves in the same way as Perl, except in the case
1140 of subroutine/recursion calls such as (?1) which have in any case always
1141 been different (but PCRE had them first :-).
1142
1143 16. Related to 15 above: Perl does not treat the | in a conditional group as
1144 creating alternatives. Such a group is treated in the same way as an
1145 ordinary group without any | characters when processing (*THEN). PCRE has
1146 been changed to match Perl's behaviour.
1147
1148 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1149 RunGrepTest script failed.
1150
1151 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1152 inevitable for groups that contain captures, but it can lead to a lot of
1153 stack use in large patterns. The old behaviour has been restored for atomic
1154 groups that do not contain any capturing parentheses.
1155
1156 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1157 suppress the check for a minimum subject length at run time. (If it was
1158 given to pcre_exec() or pcre_dfa_exec() it did work.)
1159
1160 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1161 fail to work when decoding hex characters in data strings in EBCDIC
1162 environments.
1163
1164 21. It appears that in at least one Mac OS environment, the isxdigit() function
1165 is implemented as a macro that evaluates to its argument more than once,
1166 contravening the C 90 Standard (I haven't checked a later standard). There
1167 was an instance in pcretest which caused it to go wrong when processing
1168 \x{...} escapes in subject strings. The has been rewritten to avoid using
1169 things like p++ in the argument of isxdigit().
1170
1171
1172 Version 8.13 16-Aug-2011
1173 ------------------------
1174
1175 1. The Unicode data tables have been updated to Unicode 6.0.0.
1176
1177 2. Two minor typos in pcre_internal.h have been fixed.
1178
1179 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1180 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1181 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1182
1183 4. There were a number of related bugs in the code for matching backrefences
1184 caselessly in UTF-8 mode when codes for the characters concerned were
1185 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1186 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1187 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1188 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1189 2-byte code at the end of the subject (it thought there wasn't enough data
1190 left).
1191
1192 5. Comprehensive information about what went wrong is now returned by
1193 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1194 as the output vector has at least 2 elements. The offset of the start of
1195 the failing character and a reason code are placed in the vector.
1196
1197 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1198 now returned is for the first byte of the failing character, instead of the
1199 last byte inspected. This is an incompatible change, but I hope it is small
1200 enough not to be a problem. It makes the returned offset consistent with
1201 pcre_exec() and pcre_dfa_exec().
1202
1203 7. pcretest now gives a text phrase as well as the error number when
1204 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1205 failure, the offset and reason code are output.
1206
1207 8. When \R was used with a maximizing quantifier it failed to skip backwards
1208 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1209 back over a single character (\n). This seems wrong (because it treated the
1210 two characters as a single entity when going forwards), conflicts with the
1211 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1212 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1213 has been changed.
1214
1215 9. Some internal refactoring has changed the processing so that the handling
1216 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1217 time (the PCRE_DOTALL option was changed this way some time ago: version
1218 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1219 which was always a bit of a fudge. It also means that there is one less
1220 argument for the match() function, which reduces its stack requirements
1221 slightly. This change also fixes an incompatibility with Perl: the pattern
1222 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1223
1224 10. More internal refactoring has drastically reduced the number of recursive
1225 calls to match() for possessively repeated groups such as (abc)++ when
1226 using pcre_exec().
1227
1228 11. While implementing 10, a number of bugs in the handling of groups were
1229 discovered and fixed:
1230
1231 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1232 (a|)*(?1) gave a compile-time internal error.
1233 ((a|)+)+ did not notice that the outer group could match an empty string.
1234 (^a|^)+ was not marked as anchored.
1235 (.*a|.*)+ was not marked as matching at start or after a newline.
1236
1237 12. Yet more internal refactoring has removed another argument from the match()
1238 function. Special calls to this function are now indicated by setting a
1239 value in a variable in the "match data" data block.
1240
1241 13. Be more explicit in pcre_study() instead of relying on "default" for
1242 opcodes that mean there is no starting character; this means that when new
1243 ones are added and accidentally left out of pcre_study(), testing should
1244 pick them up.
1245
1246 14. The -s option of pcretest has been documented for ages as being an old
1247 synonym of -m (show memory usage). I have changed it to mean "force study
1248 for every regex", that is, assume /S for every regex. This is similar to -i
1249 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1250 using it. It makes it easier to run collections of tests with and without
1251 study enabled, and thereby test pcre_study() more easily. All the standard
1252 tests are now run with and without -s (but some patterns can be marked as
1253 "never study" - see 20 below).
1254
1255 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1256 restoration of the capturing data to the outer values was not happening
1257 correctly.
1258
1259 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1260 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1261 pattern had matched an empty string, and so incorrectly returned a no
1262 match.
1263
1264 17. There was optimizing code for the last branch of non-capturing parentheses,
1265 and also for the obeyed branch of a conditional subexpression, which used
1266 tail recursion to cut down on stack usage. Unfortunately, now that there is
1267 the possibility of (*THEN) occurring in these branches, tail recursion is
1268 no longer possible because the return has to be checked for (*THEN). These
1269 two optimizations have therefore been removed. [But see 8.20/11 above.]
1270
1271 18. If a pattern containing \R was studied, it was assumed that \R always
1272 matched two bytes, thus causing the minimum subject length to be
1273 incorrectly computed because \R can also match just one byte.
1274
1275 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1276 was incorrectly computed.
1277
1278 20. If /S is present twice on a test pattern in pcretest input, it now
1279 *disables* studying, thereby overriding the use of -s on the command line
1280 (see 14 above). This is necessary for one or two tests to keep the output
1281 identical in both cases.
1282
1283 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1284 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1285
1286 22. When an atomic group that contained a capturing parenthesis was
1287 successfully matched, but the branch in which it appeared failed, the
1288 capturing was not being forgotten if a higher numbered group was later
1289 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1290 group 1 to "a", when in fact it should be unset. This applied to multi-
1291 branched capturing and non-capturing groups, repeated or not, and also to
1292 positive assertions (capturing in negative assertions does not happen
1293 in PCRE) and also to nested atomic groups.
1294
1295 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1296 subject after a captured substring, to make it easier to tell which of a
1297 number of identical substrings has been captured.
1298
1299 24. The way atomic groups are processed by pcre_exec() has been changed so that
1300 if they are repeated, backtracking one repetition now resets captured
1301 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1302 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1303 "aaa". Previously, it would have been "a". As part of this code
1304 refactoring, the way recursive calls are handled has also been changed.
1305
1306 25. If an assertion condition captured any substrings, they were not passed
1307 back unless some other capturing happened later. For example, if
1308 (?(?=(a))a) was matched against "a", no capturing was returned.
1309
1310 26. When studying a pattern that contained subroutine calls or assertions,
1311 the code for finding the minimum length of a possible match was handling
1312 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1313 group 1 called group 2 while simultaneously a separate group 2 called group
1314 1). A stack overflow occurred in this case. I have fixed this by limiting
1315 the recursion depth to 10.
1316
1317 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1318 Fortmann. This supports explicit test numbers on the command line, and has
1319 argument validation and error reporting.
1320
1321 28. An instance of \X with an unlimited repeat could fail if at any point the
1322 first character it looked at was a mark character.
1323
1324 29. Some minor code refactoring concerning Unicode properties and scripts
1325 should reduce the stack requirement of match() slightly.
1326
1327 30. Added the '=' option to pcretest to check the setting of unused capturing
1328 slots at the end of the pattern, which are documented as being -1, but are
1329 not included in the return count.
1330
1331 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1332 compiled something random. Now it gives a compile-time error (as does
1333 Perl).
1334
1335 32. A *MARK encountered during the processing of a positive assertion is now
1336 recorded and passed back (compatible with Perl).
1337
1338 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1339 had alternative anchored branches, the search for a second match in a line
1340 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1341 matched the line "0102" twice. The same bug affected patterns that started
1342 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1343 twice.
1344
1345 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1346 does, and because of capturing effects, quantifying parenthesized
1347 assertions may at times be useful. Quantifiers are now allowed for
1348 parenthesized assertions.
1349
1350 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1351
1352 36. \g was being checked for fancy things in a character class, when it should
1353 just be a literal "g".
1354
1355 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1356 appearance of a nested POSIX class supersedes an apparent external class.
1357 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1358 unescaped square brackets may also appear as part of class names. For
1359 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1360 more like Perl. (But see 8.20/1 above.)
1361
1362 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1363 was because it thought it was \N{name}, which is not supported).
1364
1365 39. Add minix to OS list not supporting the -S option in pcretest.
1366
1367 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1368 cannot analyze patterns in sufficient detail to catch mutual recursions
1369 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1370 subgroup is called recursively as a subpattern for a second time at the
1371 same position in the subject string. In previous releases this might have
1372 been caught by the recursion limit, or it might have run out of stack.
1373
1374 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1375 happen only once. PCRE was, however incorrectly giving a compile time error
1376 "recursive call could loop indefinitely" because it cannot analyze the
1377 pattern in sufficient detail. The compile time test no longer happens when
1378 PCRE is compiling a conditional subpattern, but actual runaway loops are
1379 now caught at runtime (see 40 above).
1380
1381 42. It seems that Perl allows any characters other than a closing parenthesis
1382 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1383 has been changed to be the same.
1384
1385 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1386 as not to get warnings when autogen.sh is called. Also changed
1387 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1388
1389 44. To help people who use pcregrep to scan files containing exceedingly long
1390 lines, the following changes have been made:
1391
1392 (a) The default value of the buffer size parameter has been increased from
1393 8K to 20K. (The actual buffer used is three times this size.)
1394
1395 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1396 PCRE is built.
1397
1398 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1399 to be set at run time.
1400
1401 (d) Numerical values in pcregrep options can be followed by K or M, for
1402 example --buffer-size=50K.
1403
1404 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1405 given and the return code is set to 2.
1406
1407 45. Add a pointer to the latest mark to the callout data block.
1408
1409 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1410 partial match of an empty string instead of no match. This was specific to
1411 the use of ".".
1412
1413 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1414 complete match instead of a partial match. This bug was dependent on both
1415 the PCRE_UTF8 and PCRE_DOTALL options being set.
1416
1417 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1418 starting byte set, because \b was not being ignored.
1419
1420
1421 Version 8.12 15-Jan-2011
1422 ------------------------
1423
1424 1. Fixed some typos in the markup of the man pages, and wrote a script that
1425 checks for such things as part of the documentation building process.
1426
1427 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1428 --match-limit and --recursion-limit options (added for 8.11). In
1429 particular, this made one of the standard tests fail. (The integer value
1430 went into the wrong half of a long int.)
1431
1432 3. If the --colour option was given to pcregrep with -v (invert match), it
1433 did strange things, either producing crazy output, or crashing. It should,
1434 of course, ignore a request for colour when reporting lines that do not
1435 match.
1436
1437 4. Another pcregrep bug caused similar problems if --colour was specified with
1438 -M (multiline) and the pattern match finished with a line ending.
1439
1440 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1441 matched in multiline mode, the following line was shown as part of the
1442 match. This seems wrong, so I have changed it.
1443
1444 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1445 the check for further matches in the same line (so they could be coloured)
1446 to overrun the end of the current line. If another match was found, it was
1447 incorrectly shown (and then shown again when found in the next line).
1448
1449 7. If pcregrep was compiled under Windows, there was a reference to the
1450 function pcregrep_exit() before it was defined. I am assuming this was
1451 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1452 reported by a user. I've moved the definition above the reference.
1453
1454
1455 Version 8.11 10-Dec-2010
1456 ------------------------
1457
1458 1. (*THEN) was not working properly if there were untried alternatives prior
1459 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1460 backtracked to try for "b" instead of moving to the next alternative branch
1461 at the same level (in this case, to look for "c"). The Perl documentation
1462 is clear that when (*THEN) is backtracked onto, it goes to the "next
1463 alternative in the innermost enclosing group".
1464
1465 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1466 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1467 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1468 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1469 (*THEN).
1470
1471 3. If \s appeared in a character class, it removed the VT character from
1472 the class, even if it had been included by some previous item, for example
1473 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1474 of \s, but is part of the POSIX "space" class.)
1475
1476 4. A partial match never returns an empty string (because you can always
1477 match an empty string at the end of the subject); however the checking for
1478 an empty string was starting at the "start of match" point. This has been
1479 changed to the "earliest inspected character" point, because the returned
1480 data for a partial match starts at this character. This means that, for
1481 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1482 (previously it gave "no match").
1483
1484 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1485 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1486 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1487 has an implication that the given string is incomplete (because a partial
1488 match is preferred over a full match). For this reason, these items now
1489 give a partial match in this situation. [Aside: previously, the one case
1490 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1491 match rather than a full match, which was wrong by the old rules, but is
1492 now correct.]
1493
1494 6. There was a bug in the handling of #-introduced comments, recognized when
1495 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1496 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1497 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1498 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1499 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1500 places in pcre_compile().
1501
1502 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1503 comments when looking ahead for named forward references to subpatterns,
1504 the only newline sequence it recognized was NL. It now handles newlines
1505 according to the set newline convention.
1506
1507 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1508 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1509 cater for a lack of strerror(). These oversights have been fixed.
1510
1511 9. Added --match-limit and --recursion-limit to pcregrep.
1512
1513 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1514
1515 11. When the -o option was used, pcregrep was setting a return code of 1, even
1516 when matches were found, and --line-buffered was not being honoured.
1517
1518 12. Added an optional parentheses number to the -o and --only-matching options
1519 of pcregrep.
1520
1521 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1522 can match an empty string. The code to do it in pcretest and pcredemo
1523 needed fixing:
1524
1525 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1526 only one byte after an empty string match just before CRLF (this case
1527 just got forgotten; "any" and "anycrlf" were OK).
1528
1529 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1530 mode when an empty string match preceded an ASCII character followed by
1531 a non-ASCII character. (The code for advancing by one character rather
1532 than one byte was nonsense.)
1533
1534 (c) The pcredemo.c sample program did not have any code at all to handle
1535 the cases when CRLF is a valid newline sequence.
1536
1537 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1538 as a starting offset was within the subject string. There is now a new
1539 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1540 negative or greater than the length of the string. In order to test this,
1541 pcretest is extended to allow the setting of negative starting offsets.
1542
1543 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1544 starting offset points to the beginning of a UTF-8 character was
1545 unnecessarily clumsy. I tidied it up.
1546
1547 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1548 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1549
1550 17. Nobody had reported that the --include_dir option, which was added in
1551 release 7.7 should have been called --include-dir (hyphen, not underscore)
1552 for compatibility with GNU grep. I have changed it to --include-dir, but
1553 left --include_dir as an undocumented synonym, and the same for
1554 --exclude-dir, though that is not available in GNU grep, at least as of
1555 release 2.5.4.
1556
1557 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1558 characters from a string of bytes) have been redefined so as not to use
1559 loops, in order to improve performance in some environments. At the same
1560 time, I abstracted some of the common code into auxiliary macros to save
1561 repetition (this should not affect the compiled code).
1562
1563 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1564 compile-time error is now given if \c is not followed by an ASCII
1565 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1566 different, and any byte value is allowed.)
1567
1568 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1569 START_OPTIMIZE option, which is now allowed at compile time - but just
1570 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1571 to pcregrep and other applications that have no direct access to PCRE
1572 options. The new /Y option in pcretest sets this option when calling
1573 pcre_compile().
1574
1575 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1576 back references. Groups containing recursive back references were forced to
1577 be atomic by that change, but in the case of named groups, the amount of
1578 memory required was incorrectly computed, leading to "Failed: internal
1579 error: code overflow". This has been fixed.
1580
1581 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1582 pcretest.c, to avoid build problems in some Borland environments.
1583
1584
1585 Version 8.10 25-Jun-2010
1586 ------------------------
1587
1588 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1589 THEN.
1590
1591 2. (*ACCEPT) was not working when inside an atomic group.
1592
1593 3. Inside a character class, \B is treated as a literal by default, but
1594 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1595 causes the error). The code is unchanged, but I tidied the documentation.
1596
1597 4. Inside a character class, PCRE always treated \R and \X as literals,
1598 whereas Perl faults them if its -w option is set. I have changed PCRE so
1599 that it faults them when PCRE_EXTRA is set.
1600
1601 5. Added support for \N, which always matches any character other than
1602 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1603
1604 6. When compiling pcregrep with newer versions of gcc which may have
1605 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1606 declared with attribute warn_unused_result" were given. Just casting the
1607 result to (void) does not stop the warnings; a more elaborate fudge is
1608 needed. I've used a macro to implement this.
1609
1610 7. Minor change to pcretest.c to avoid a compiler warning.
1611
1612 8. Added four artifical Unicode properties to help with an option to make
1613 \s etc use properties (see next item). The new properties are: Xan
1614 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1615
1616 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1617 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1618 this option. Modified pcretest to add /W to test this facility. Added
1619 REG_UCP to make it available via the POSIX interface.
1620
1621 10. Added --line-buffered to pcregrep.
1622
1623 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1624 studied, and the match started with a letter with a code point greater than
1625 127 whose first byte was different to the first byte of the other case of
1626 the letter, the other case of this starting letter was not recognized
1627 (#976).
1628
1629 12. If a pattern that was studied started with a repeated Unicode property
1630 test, for example, \p{Nd}+, there was the theoretical possibility of
1631 setting up an incorrect bitmap of starting bytes, but fortunately it could
1632 not have actually happened in practice until change 8 above was made (it
1633 added property types that matched character-matching opcodes).
1634
1635 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1636 possible starting bytes for non-anchored patterns.
1637
1638 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1639 \R, and also a number of cases that involve Unicode properties, both
1640 explicit and implicit when PCRE_UCP is set.
1641
1642 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1643 input, it could crash or give wrong results if characters with values
1644 greater than 0xc0 were present in the subject string. (Detail: it assumed
1645 UTF-8 input when processing these items.)
1646
1647 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1648 size_t is 64-bit (#991).
1649
1650 17. Added a check for running out of memory when PCRE is compiled with
1651 --disable-stack-for-recursion (#990).
1652
1653 18. If the last data line in a file for pcretest does not have a newline on
1654 the end, a newline was missing in the output.
1655
1656 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1657 less than 128) in its various bitmaps. However, there is a facility for
1658 generating tables according to the current locale when PCRE is compiled. It
1659 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1660 characters, are recognized by isspace() and therefore were getting set in
1661 these tables, and indeed these tables seem to approximate to ISO 8859. This
1662 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1663 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1664 which of course cannot start UTF-8 characters. I have changed the code so
1665 that only real ASCII characters (less than 128) and the correct starting
1666 bytes for UTF-8 encodings are set for characters greater than 127 when in
1667 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1668 altogether.)
1669
1670 20. Added the /T option to pcretest so as to be able to run tests with non-
1671 standard character tables, thus making it possible to include the tests
1672 used for 19 above in the standard set of tests.
1673
1674 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1675 reference to a subpattern the other side of a comment that contains an
1676 opening parenthesis caused either an internal compiling error, or a
1677 reference to the wrong subpattern.
1678
1679
1680 Version 8.02 19-Mar-2010
1681 ------------------------
1682
1683 1. The Unicode data tables have been updated to Unicode 5.2.0.
1684
1685 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1686 configured.
1687
1688 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1689 original author of that file, following a query about its status.
1690
1691 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1692 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1693
1694 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1695 quantifier applied to a forward-referencing subroutine call, could compile
1696 incorrect code or give the error "internal error: previously-checked
1697 referenced subpattern not found".
1698
1699 6. Both MS Visual Studio and Symbian OS have problems with initializing
1700 variables to point to external functions. For these systems, therefore,
1701 pcre_malloc etc. are now initialized to local functions that call the
1702 relevant global functions.
1703
1704 7. There were two entries missing in the vectors called coptable and poptable
1705 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1706 I've fixed the data, and added a kludgy way of testing at compile time that
1707 the lengths are correct (equal to the number of opcodes).
1708
1709 8. Following on from 7, I added a similar kludge to check the length of the
1710 eint vector in pcreposix.c.
1711
1712 9. Error texts for pcre_compile() are held as one long string to avoid too
1713 much relocation at load time. To find a text, the string is searched,
1714 counting zeros. There was no check for running off the end of the string,
1715 which could happen if a new error number was added without updating the
1716 string.
1717
1718 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1719
1720 11. \K was not working if it appeared in an atomic group or in a group that
1721 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1722 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1723 the assertion is positive, but not if it is negative.
1724
1725 12. Change 11 fortuitously reduced the size of the stack frame used in the
1726 "match()" function of pcre_exec.c by one pointer. Forthcoming
1727 implementation of support for (*MARK) will need an extra pointer on the
1728 stack; I have reserved it now, so that the stack frame size does not
1729 decrease.
1730
1731 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1732 item in branch that calls a recursion is a subroutine call - as in the
1733 second branch in the above example - was incorrectly given the compile-
1734 time error "recursive call could loop indefinitely" because pcre_compile()
1735 was not correctly checking the subroutine for matching a non-empty string.
1736
1737 14. The checks for overrunning compiling workspace could trigger after an
1738 overrun had occurred. This is a "should never occur" error, but it can be
1739 triggered by pathological patterns such as hundreds of nested parentheses.
1740 The checks now trigger 100 bytes before the end of the workspace.
1741
1742 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1743
1744
1745 Version 8.01 19-Jan-2010
1746 ------------------------
1747
1748 1. If a pattern contained a conditional subpattern with only one branch (in
1749 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1750 computed the wrong minimum data length (which is of course zero for such
1751 subpatterns). This could cause incorrect "no match" results.
1752
1753 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1754 the pattern is reset in the first branch, pcre_compile() failed with
1755 "internal error: code overflow at offset...". This happened only when
1756 the reset was to the original external option setting. (An optimization
1757 abstracts leading options settings into an external setting, which was the
1758 cause of this.)
1759
1760 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1761 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1762 assertion pattern did not match (meaning that the assertion was true), it
1763 was incorrectly treated as false if the SKIP had been reached during the
1764 matching. This also applied to assertions used as conditions.
1765
1766 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1767 assertion subpattern, including such a pattern used as a condition,
1768 unpredictable results occurred, instead of the error return
1769 PCRE_ERROR_DFA_UITEM.
1770
1771 5. The C++ GlobalReplace function was not working like Perl for the special
1772 situation when an empty string is matched. It now does the fancy magic
1773 stuff that is necessary.
1774
1775 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1776 removed. (These were left over from very, very early versions of PCRE.)
1777
1778 7. Some cosmetic changes to the code to make life easier when compiling it
1779 as part of something else:
1780
1781 (a) Change DEBUG to PCRE_DEBUG.
1782
1783 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1784 called "current" as "current_branch", to prevent a collision with the
1785 Linux macro when compiled as a kernel module.
1786
1787 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1788 prevent a collision with the Linux macro when compiled as a kernel
1789 module.
1790
1791 8. In pcre_compile() there are some checks for integer overflows that used to
1792 cast potentially large values to (double). This has been changed to that
1793 when building, a check for int64_t is made, and if it is found, it is used
1794 instead, thus avoiding the use of floating point arithmetic. (There is no
1795 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1796 double.
1797
1798 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1799 2005 (difference between two addresses compared to an unsigned value).
1800
1801 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1802 custom one, because of the following reported problem in Windows:
1803
1804 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1805 under Win32.
1806 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1807 therefore missing the function definition.
1808 - The compiler thus generates a "C" signature for the test function.
1809 - The linker fails to find the "C" function.
1810 - PCRE fails to configure if asked to do so against libbz2.
1811
1812 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1813 messages were output:
1814
1815 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1816 rerunning libtoolize, to keep the correct libtool macros in-tree.
1817 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1818
1819 I have done both of these things.
1820
1821 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1822 most of the time, it *can* run out if it is given a pattern that contains a
1823 runaway infinite recursion. I updated the discussion in the pcrestack man
1824 page.
1825
1826 13. Now that we have gone to the x.xx style of version numbers, the minor
1827 version may start with zero. Using 08 or 09 is a bad idea because users
1828 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1829 interpreted as invalid octal numbers. I've updated the previous comment in
1830 configure.ac, and also added a check that gives an error if 08 or 09 are
1831 used.
1832
1833 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1834 causing partial matching to fail when the end of the subject matched \W
1835 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1836
1837 15. There were some discrepancies between the declarations in pcre_internal.h
1838 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1839 their definitions. The declarations used "const uschar *" and the
1840 definitions used USPTR. Even though USPTR is normally defined as "const
1841 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1842 reported that: "This difference in casting confuses some C++ compilers, for
1843 example, SunCC recognizes above declarations as different functions and
1844 generates broken code for hbpcre." I have changed the declarations to use
1845 USPTR.
1846
1847 16. GNU libtool is named differently on some systems. The autogen.sh script now
1848 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1849 (FreeBSD).
1850
1851 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1852 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1853 comment: "Figure out how to create a longlong from a string: strtoll and
1854 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1855 instance, but it only takes 2 args instead of 3!"
1856
1857 18. A subtle bug concerned with back references has been fixed by a change of
1858 specification, with a corresponding code fix. A pattern such as
1859 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1860 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1861 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1862 same bug. Such groups have to be quantified to be useful, or contained
1863 inside another quantified group. (If there's no repetition, the reference
1864 can never match.) The problem arises because, having left the group and
1865 moved on to the rest of the pattern, a later failure that backtracks into
1866 the group uses the captured value from the final iteration of the group
1867 rather than the correct earlier one. I have fixed this in PCRE by forcing
1868 any group that contains a reference to itself to be an atomic group; that
1869 is, there cannot be any backtracking into it once it has completed. This is
1870 similar to recursive and subroutine calls.
1871
1872
1873 Version 8.00 19-Oct-09
1874 ----------------------
1875
1876 1. The table for translating pcre_compile() error codes into POSIX error codes
1877 was out-of-date, and there was no check on the pcre_compile() error code
1878 being within the table. This could lead to an OK return being given in
1879 error.
1880
1881 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1882 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1883 in a Windows environment.
1884
1885 3. The pcregrep --count option prints the count for each file even when it is
1886 zero, as does GNU grep. However, pcregrep was also printing all files when
1887 --files-with-matches was added. Now, when both options are given, it prints
1888 counts only for those files that have at least one match. (GNU grep just
1889 prints the file name in this circumstance, but including the count seems
1890 more useful - otherwise, why use --count?) Also ensured that the
1891 combination -clh just lists non-zero counts, with no names.
1892
1893 4. The long form of the pcregrep -F option was incorrectly implemented as
1894 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1895 but it seems right to fix it, and I didn't think it was worth preserving
1896 the old behaviour.
1897
1898 5. The command line items --regex=pattern and --regexp=pattern were not
1899 recognized by pcregrep, which required --regex pattern or --regexp pattern
1900 (with a space rather than an '='). The man page documented the '=' forms,
1901 which are compatible with GNU grep; these now work.
1902
1903 6. No libpcreposix.pc file was created for pkg-config; there was just
1904 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1905
1906 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1907 when UCP support is not needed, by modifying the Python script that
1908 generates it from Unicode data files. This should not matter if the module
1909 is correctly used as a library, but I received one complaint about 50K of
1910 unwanted data. My guess is that the person linked everything into his
1911 program rather than using a library. Anyway, it does no harm.
1912
1913 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1914 was a minimum greater than 1 for a wide character in a possessive
1915 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1916 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1917 character. Chaos in the form of incorrect output or a compiling loop could
1918 result.
1919
1920 9. The restrictions on what a pattern can contain when partial matching is
1921 requested for pcre_exec() have been removed. All patterns can now be
1922 partially matched by this function. In addition, if there are at least two
1923 slots in the offset vector, the offset of the earliest inspected character
1924 for the match and the offset of the end of the subject are set in them when
1925 PCRE_ERROR_PARTIAL is returned.
1926
1927 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1928 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1929 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1930 and may be more useful for multi-segment matching.
1931
1932 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1933 used to be given if ever the end of the subject was reached; now it is
1934 given only if matching could not proceed because another character was
1935 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1936 string "Z", which now yields "no match" instead of "partial match". In the
1937 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1938 final character ended with (*FAIL).
1939
1940 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1941 if the pattern had a "must contain" character that was already found in the
1942 earlier partial match, unless partial matching was again requested. For
1943 example, with the pattern /dog.(body)?/, the "must contain" character is
1944 "g". If the first part-match was for the string "dog", restarting with
1945 "sbody" failed. This bug has been fixed.
1946
1947 13. The string returned by pcre_dfa_exec() after a partial match has been
1948 changed so that it starts at the first inspected character rather than the
1949 first character of the match. This makes a difference only if the pattern
1950 starts with a lookbehind assertion or \b or \B (\K is not supported by
1951 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1952 matching functions compatible, and I think it's the right thing to do.
1953
1954 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1955 so that the demonstration program is easily available in environments where
1956 PCRE has not been installed from source.
1957
1958 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1959 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1960 library.
1961
1962 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1963 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1964 is not the first non-POSIX option to be added. Clearly some people find
1965 these options useful.
1966
1967 17. If a caller to the POSIX matching function regexec() passes a non-zero
1968 value for nmatch with a NULL value for pmatch, the value of
1969 nmatch is forced to zero.
1970
1971 18. RunGrepTest did not have a test for the availability of the -u option of
1972 the diff command, as RunTest does. It now checks in the same way as
1973 RunTest, and also checks for the -b option.
1974
1975 19. If an odd number of negated classes containing just a single character
1976 interposed, within parentheses, between a forward reference to a named
1977 subpattern and the definition of the subpattern, compilation crashed with
1978 an internal error, complaining that it could not find the referenced
1979 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1980 [The bug was that it was starting one character too far in when skipping
1981 over the character class, thus treating the ] as data rather than
1982 terminating the class. This meant it could skip too much.]
1983
1984 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1985 /g option in pcretest when the pattern contains \K, which makes it possible
1986 to have an empty string match not at the start, even when the pattern is
1987 anchored. Updated pcretest and pcredemo to use this option.
1988
1989 21. If the maximum number of capturing subpatterns in a recursion was greater
1990 than the maximum at the outer level, the higher number was returned, but
1991 with unset values at the outer level. The correct (outer level) value is
1992 now given.
1993
1994 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1995 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1996 make it do so. The string so far is captured, making this feature
1997 compatible with Perl.
1998
1999 23. The tests have been re-organized, adding tests 11 and 12, to make it
2000 possible to check the Perl 5.10 features against Perl 5.10.
2001
2002 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2003 pattern matches a fixed length string. PCRE did not allow this; now it
2004 does. Neither allows recursion.
2005
2006 25. I finally figured out how to implement a request to provide the minimum
2007 length of subject string that was needed in order to match a given pattern.
2008 (It was back references and recursion that I had previously got hung up
2009 on.) This code has now been added to pcre_study(); it finds a lower bound
2010 to the length of subject needed. It is not necessarily the greatest lower
2011 bound, but using it to avoid searching strings that are too short does give
2012 some useful speed-ups. The value is available to calling programs via
2013 pcre_fullinfo().
2014
2015 26. While implementing 25, I discovered to my embarrassment that pcretest had
2016 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2017 study optimizations had never been tested with that matching function.
2018 Oops. What is worse, even when it was passed study data, there was a bug in
2019 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2020 were also very few tests of studied patterns with pcre_dfa_exec().
2021
2022 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2023 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2024 on the other side of the coin, they are no longer allowed to have different
2025 names, because these cannot be distinguished in PCRE, and this has caused
2026 confusion. (This is a difference from Perl.)
2027
2028 28. When duplicate subpattern names are present (necessarily with different
2029 numbers, as required by 27 above), and a test is made by name in a
2030 conditional pattern, either for a subpattern having been matched, or for
2031 recursion in such a pattern, all the associated numbered subpatterns are
2032 tested, and the overall condition is true if the condition is true for any
2033 one of them. This is the way Perl works, and is also more like the way
2034 testing by number works.
2035
2036
2037 Version 7.9 11-Apr-09
2038 ---------------------
2039
2040 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2041 (pcretest), all targets were linked against these libraries. This included
2042 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2043 libraries. This caused unwanted dependencies to be created. This problem
2044 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2045 pcretest is linked with readline.
2046
2047 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2048 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2049 moved outside it again, because FALSE and TRUE are already defined in AIX,
2050 but BOOL is not.
2051
2052 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2053 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2054
2055 4. The pcregrep documentation said spaces were inserted as well as colons (or
2056 hyphens) following file names and line numbers when outputting matching
2057 lines. This is not true; no spaces are inserted. I have also clarified the
2058 wording for the --colour (or --color) option.
2059
2060 5. In pcregrep, when --colour was used with -o, the list of matching strings
2061 was not coloured; this is different to GNU grep, so I have changed it to be
2062 the same.
2063
2064 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2065 each matching line was coloured. Now it goes on to look for further matches
2066 of any of the test patterns, which is the same behaviour as GNU grep.
2067
2068 7. A pattern that could match an empty string could cause pcregrep to loop; it
2069 doesn't make sense to accept an empty string match in pcregrep, so I have
2070 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2071 seems to be how GNU grep behaves. [But see later change 40 for release
2072 8.33.]
2073
2074 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2075 start or after a newline", because the conditional assertion was not being
2076 correctly handled. The rule now is that both the assertion and what follows
2077 in the first alternative must satisfy the test.
2078
2079 9. If auto-callout was enabled in a pattern with a conditional group whose
2080 condition was an assertion, PCRE could crash during matching, both with
2081 pcre_exec() and pcre_dfa_exec().
2082
2083 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2084 used for matching.
2085
2086 11. Unicode property support in character classes was not working for
2087 characters (bytes) greater than 127 when not in UTF-8 mode.
2088
2089 12. Added the -M command line option to pcretest.
2090
2091 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2092
2093 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2094
2095 16. Added comments and documentation about mis-use of no_arg in the C++
2096 wrapper.
2097
2098 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2099 from Martin Jerabek that uses macro names for all relevant character and
2100 string constants.
2101
2102 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2103 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2104 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2105 these, but not everybody uses configure.
2106
2107 19. A conditional group that had only one branch was not being correctly
2108 recognized as an item that could match an empty string. This meant that an
2109 enclosing group might also not be so recognized, causing infinite looping
2110 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2111 with the subject "ab", where knowledge that the repeated group can match
2112 nothing is needed in order to break the loop.
2113
2114 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2115 exec(), but without supplying a callout function, matching went wrong.
2116
2117 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2118 leak if the size of the offset vector was greater than 30. When the vector
2119 is smaller, the saved offsets during recursion go onto a local stack
2120 vector, but for larger vectors malloc() is used. It was failing to free
2121 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2122 error, in fact).
2123
2124 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2125 heapframe that is used only when UTF-8 support is enabled. This caused no
2126 problem, but was untidy.
2127
2128 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2129 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2130 included within another project.
2131
2132 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2133 slightly modified by me:
2134
2135 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2136 not building pcregrep.
2137
2138 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2139 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2140
2141 25. Forward references, both numeric and by name, in patterns that made use of
2142 duplicate group numbers, could behave incorrectly or give incorrect errors,
2143 because when scanning forward to find the reference group, PCRE was not
2144 taking into account the duplicate group numbers. A pattern such as
2145 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2146
2147 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2148 the feature of a custom pointer more persuasive (as requested by a user).
2149
2150 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2151 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2152 pre-defined.
2153
2154 28. Added support for (*UTF8) at the start of a pattern.
2155
2156 29. Arrange for flags added by the "release type" setting in CMake to be shown
2157 in the configuration summary.
2158
2159
2160 Version 7.8 05-Sep-08
2161 ---------------------
2162
2163 1. Replaced UCP searching code with optimized version as implemented for Ad
2164 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2165 stage table and inline lookup instead of a function, giving speed ups of 2
2166 to 5 times on some simple patterns that I tested. Permission was given to
2167 distribute the MultiStage2.py script that generates the tables (it's not in
2168 the tarball, but is in the Subversion repository).
2169
2170 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2171 scripts.
2172
2173 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2174 a group with a zero qualifier. The result of the study could be incorrect,
2175 or the function might crash, depending on the pattern.
2176
2177 4. Caseless matching was not working for non-ASCII characters in back
2178 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2179 It now works when Unicode Property Support is available.
2180
2181 5. In pcretest, an escape such as \x{de} in the data was always generating
2182 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2183 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2184 truncation.
2185
2186 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2187
2188 7. Added two (int) casts to pcregrep when printing the difference of two
2189 pointers, in case they are 64-bit values.
2190
2191 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2192 test 2 if it fails.
2193
2194 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2195 and a #define of that name to empty if it is not externally set. This is to
2196 allow users of MSVC to set it if necessary.
2197
2198 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2199 the convenience functions in the pcre_get.c source file.
2200
2201 11. An option change at the start of a pattern that had top-level alternatives
2202 could cause overwriting and/or a crash. This command provoked a crash in
2203 some environments:
2204
2205 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2206
2207 This potential security problem was recorded as CVE-2008-2371.
2208
2209 12. For a pattern where the match had to start at the beginning or immediately
2210 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2211 pcre_dfa_exec() could read past the end of the passed subject if there was
2212 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2213 pcretest so that it places the subject at the end of its malloc-ed buffer.
2214
2215 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2216 exec() might read past the end of the data buffer in UTF-8 mode.
2217
2218 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2219 the data contained the byte 0x85 as part of a UTF-8 character within its
2220 first line. This applied both to normal and DFA matching.
2221
2222 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2223 /^[^d]*?$/8 failed to match "abc".
2224
2225 16. Added a missing copyright notice to pcrecpp_internal.h.
2226
2227 17. Make it more clear in the documentation that values returned from
2228 pcre_exec() in ovector are byte offsets, not character counts.
2229
2230 18. Tidied a few places to stop certain compilers from issuing warnings.
2231
2232 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2233 supplied by Stefan Weber. I made a further small update for 7.8 because
2234 there is a change of source arrangements: the pcre_searchfuncs.c module is
2235 replaced by pcre_ucd.c.
2236
2237
2238 Version 7.7 07-May-08
2239 ---------------------
2240
2241 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2242 a string to a long long, pretend we don't even have a long long." This is
2243 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2244
2245 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2246 pre-7.6 versions, which defined a global no_arg variable instead of putting
2247 it in the RE class. (See also #8 below.)
2248
2249 3. Remove a line of dead code, identified by coverity and reported by Nuno
2250 Lopes.
2251
2252 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2253
2254 (1) The include/exclude patterns were being applied to the whole pathnames
2255 of files, instead of just to the final components.
2256
2257 (2) If there was more than one level of directory, the subdirectories were
2258 skipped unless they satisfied the include/exclude conditions. This is
2259 inconsistent with GNU grep (and could even be seen as contrary to the
2260 pcregrep specification - which I improved to make it absolutely clear).
2261 The action now is always to scan all levels of directory, and just
2262 apply the include/exclude patterns to regular files.
2263
2264 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2265 --exclude_dir in the tests to avoid scanning .svn directories.
2266
2267 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2268 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2269 doesn't support NULs in patterns.
2270
2271 7. Added some missing "const"s to declarations of static tables in
2272 pcre_compile.c and pcre_dfa_exec.c.
2273
2274 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2275 caused by fix #2 above. (Subsequently also a second patch to fix the
2276 first patch. And a third patch - this was a messy problem.)
2277
2278 9. Applied Craig's patch to remove the use of push_back().
2279
2280 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2281 matching function regexec().
2282
2283 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2284 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2285 references. PCRE supports relative numbers with this syntax (I don't think
2286 Oniguruma does).
2287
2288 12. Previously, a group with a zero repeat such as (...){0} was completely
2289 omitted from the compiled regex. However, this means that if the group
2290 was called as a subroutine from elsewhere in the pattern, things went wrong
2291 (an internal error was given). Such groups are now left in the compiled
2292 pattern, with a new opcode that causes them to be skipped at execution
2293 time.
2294
2295 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2296 to the way PCRE behaves:
2297
2298 (a) A lone ] character is dis-allowed (Perl treats it as data).
2299
2300 (b) A back reference to an unmatched subpattern matches an empty string
2301 (Perl fails the current match path).
2302
2303 (c) A data ] in a character class must be notated as \] because if the
2304 first data character in a class is ], it defines an empty class. (In
2305 Perl it is not possible to have an empty class.) The empty class []
2306 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2307 The negative empty class [^] matches any one character, independently
2308 of the DOTALL setting.
2309
2310 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2311 non-existent subpattern following a character class starting with ']' and
2312 containing () gave an internal compiling error instead of "reference to
2313 non-existent subpattern". Fortunately, when the pattern did exist, the
2314 compiled code was correct. (When scanning forwards to check for the
2315 existence of the subpattern, it was treating the data ']' as terminating
2316 the class, so got the count wrong. When actually compiling, the reference
2317 was subsequently set up correctly.)
2318
2319 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2320 it was being rejected as not supported by pcre_dfa_exec(), even though
2321 other assertions are supported. I have made pcre_dfa_exec() support
2322 (*FAIL).
2323
2324 16. The implementation of 13c above involved the invention of a new opcode,
2325 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2326 cannot be changed at match time, I realized I could make a small
2327 improvement to matching performance by compiling OP_ALLANY instead of
2328 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2329 on the OP_ANY path.
2330
2331 17. Compiling pcretest on Windows with readline support failed without the
2332 following two fixes: (1) Make the unistd.h include conditional on
2333 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2334
2335 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2336 ncurses library to be included for pcretest when ReadLine support is
2337 requested, but also to allow for it to be overridden. This patch came from
2338 Daniel Bergström.
2339
2340 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2341 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2342 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2343 spotting this.
2344
2345
2346 Version 7.6 28-Jan-08
2347 ---------------------
2348
2349 1. A character class containing a very large number of characters with
2350 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2351 overflow.
2352
2353 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2354 HAVE_LONG_LONG is not defined.
2355
2356 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2357 bring them up to date and include new features. This patch includes:
2358
2359 - Fixed PH's badly added libz and libbz2 support.
2360 - Fixed a problem with static linking.
2361 - Added pcredemo. [But later removed - see 7 below.]
2362 - Fixed dftables problem and added an option.
2363 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2364 HAVE_LONG_LONG.
2365 - Added readline support for pcretest.
2366 - Added an listing of the option settings after cmake has run.
2367
2368 4. A user submitted a patch to Makefile that makes it easy to create
2369 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2370 Makefile.am that cause it to include this special target, without
2371 affecting anything else. Note that the same mingw target plus all
2372 the other distribution libraries and programs are now supported
2373 when configuring with CMake (see 6 below) instead of with
2374 Configure/Make.
2375
2376 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2377 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2378 exported in the Windows port". It has not yet been confirmed that the patch
2379 solves the problem, but it does no harm.
2380
2381 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2382 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2383 with CMake, and also correct the comment about stack recursion.
2384
2385 7. Remove the automatic building of pcredemo from the ./configure system and
2386 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2387 of a program that users should build themselves after PCRE is installed, so
2388 building it automatically is not really right. What is more, it gave
2389 trouble in some build environments.
2390
2391 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2392
2393
2394 Version 7.5 10-Jan-08
2395 ---------------------
2396
2397 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2398 values in parens when parsing an RE using the C++ wrapper."
2399
2400 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2401 Characters greater than 255 were excluded from the class instead of being
2402 included.
2403
2404 3. The same bug as (2) above applied to negated POSIX classes such as
2405 [:^space:].
2406
2407 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2408 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2409 I have changed it.
2410
2411 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2412 first named subpattern) and a construct such as (?&a) would reference the
2413 first named subpattern whose name started with "a" (in other words, the
2414 length check was missing). Both these problems are fixed. "Subpattern name
2415 expected" is now given for (?&) (a zero-length name), and this patch also
2416 makes it give the same error for \k'' (previously it complained that that
2417 was a reference to a non-existent subpattern).
2418
2419 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2420 this is right because (?- can be followed by option settings as well as by
2421 digits. I have, however, made the messages clearer.
2422
2423 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2424 than the number used in the conditional) now cause a compile-time error.
2425 This is actually not compatible with Perl, which accepts such patterns, but
2426 treats the conditional as always being FALSE (as PCRE used to), but it
2427 seems to me that giving a diagnostic is better.
2428
2429 8. Change "alphameric" to the more common word "alphanumeric" in comments
2430 and messages.
2431
2432 9. Fix two occurrences of "backslash" in comments that should have been
2433 "backspace".
2434
2435 10. Remove two redundant lines of code that can never be obeyed (their function
2436 was moved elsewhere).
2437
2438 11. The program that makes PCRE's Unicode character property table had a bug
2439 which caused it to generate incorrect table entries for sequences of
2440 characters that have the same character type, but are in different scripts.
2441 It amalgamated them into a single range, with the script of the first of
2442 them. In other words, some characters were in the wrong script. There were
2443 thirteen such cases, affecting characters in the following ranges:
2444
2445 U+002b0 - U+002c1
2446 U+0060c - U+0060d
2447 U+0061e - U+00612
2448 U+0064b - U+0065e
2449 U+0074d - U+0076d
2450 U+01800 - U+01805
2451 U+01d00 - U+01d77
2452 U+01d9b - U+01dbf
2453 U+0200b - U+0200f
2454 U+030fc - U+030fe
2455 U+03260 - U+0327f
2456 U+0fb46 - U+0fbb1
2457 U+10450 - U+1049d
2458
2459 12. The -o option (show only the matching part of a line) for pcregrep was not
2460 compatible with GNU grep in that, if there was more than one match in a
2461 line, it showed only the first of them. It now behaves in the same way as
2462 GNU grep.
2463
2464 13. If the -o and -v options were combined for pcregrep, it printed a blank
2465 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2466 does the same. The return code can be used to tell if there were any
2467 non-matching lines.
2468
2469 14. Added --file-offsets and --line-offsets to pcregrep.
2470
2471 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2472 infinitely looping recursion. The bug was that positive lookaheads were not
2473 being skipped when checking for a possible empty match (negative lookaheads
2474 and both kinds of lookbehind were skipped).
2475
2476 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2477 inclusion of <windows.h> to before rather than after the definition of
2478 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2479
2480 17. Specifying a possessive quantifier with a specific limit for a Unicode
2481 character property caused pcre_compile() to compile bad code, which led at
2482 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2483 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2484 caused the error; without that there was no problem.
2485
2486 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2487
2488 19. Added --enable-pcretest-libreadline.
2489
2490 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2491 RE::GlobalReplace(). As a result, the number of replacements returned was
2492 double what it should be. I removed one of the increments, but Craig sent a
2493 later patch that removed the other one (the right fix) and added unit tests
2494 that check the return values (which was not done before).
2495
2496 21. Several CMake things:
2497
2498 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2499 the names libpcre and libpcreposix, not just pcre and pcreposix.
2500
2501 (2) The above change means that pcretest and pcregrep are now correctly
2502 linked with the newly-built libraries, not previously installed ones.
2503
2504 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2505
2506 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2507 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2508 UTF-8 newline character). The key issue is that the pattern starts .*;
2509 this means that the match must be either at the beginning, or after a
2510 newline. The bug was in the code for advancing after a failed match and
2511 checking that the new position followed a newline. It was not taking
2512 account of UTF-8 characters correctly.
2513
2514 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2515 character classes. PCRE was not treating the sequence [:...:] as a
2516 character class unless the ... were all letters. Perl, however, seems to
2517 allow any characters between [: and :], though of course it rejects as
2518 unknown any "names" that contain non-letters, because all the known class
2519 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2520 for example, whereas PCRE did not - it did not recognize a POSIX character
2521 class. This seemed a bit dangerous, so the code has been changed to be
2522 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2523 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2524 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2525 Perl does, and where it didn't before.
2526
2527 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2528 Windows environments %n is disabled by default.
2529
2530
2531 Version 7.4 21-Sep-07
2532 ---------------------
2533
2534 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2535 means that a class such as [\s] counted as "explicit reference to CR or
2536 LF". That isn't really right - the whole point of the change was to try to
2537 help when there was an actual mention of one of the two characters. So now
2538 the change happens only if \r or \n (or a literal CR or LF) character is
2539 encountered.
2540
2541 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2542 of both had grown to the point where there were only 3 bits left.
2543 Fortunately, there was spare space in the data structure, and so I have
2544 moved the internal flags into a new 16-bit field to free up more option
2545 bits.
2546
2547 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2548 but did not set the internal JCHANGED flag - either of these is enough to
2549 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2550 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2551 start sets both bits.
2552
2553 4. Added options (at build time, compile time, exec time) to change \R from
2554 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2555
2556 5. doc/pcresyntax.html was missing from the distribution.
2557
2558 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2559 compatibility, even though it is no longer used.
2560
2561 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2562 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2563 windows.h file is present (where different names are used). [This was
2564 reversed later after testing - see 16 below.]
2565
2566 8. Changed all #include <config.h> to #include "config.h". There were also
2567 some further <pcre.h> cases that I changed to "pcre.h".
2568
2569 9. When pcregrep was used with the --colour option, it missed the line ending
2570 sequence off the lines that it output.
2571
2572 10. It was pointed out to me that arrays of string pointers cause lots of
2573 relocations when a shared library is dynamically loaded. A technique of
2574 using a single long string with a table of offsets can drastically reduce
2575 these. I have refactored PCRE in four places to do this. The result is
2576 dramatic:
2577
2578 Originally: 290
2579 After changing UCP table: 187
2580 After changing error message table: 43
2581 After changing table of "verbs" 36
2582 After changing table of Posix names 22
2583
2584 Thanks to the folks working on Gregex for glib for this insight.
2585
2586 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2587 unicode-properties was also set.
2588
2589 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2590
2591 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2592 checked only for CRLF.
2593
2594 14. Added casts to pcretest.c to avoid compiler warnings.
2595
2596 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2597
2598 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2599 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2600 entirely. This removes changes made in 7 above.
2601
2602 17. The CMake files have been updated, and there is now more information about
2603 building with CMake in the NON-UNIX-USE document.
2604
2605
2606 Version 7.3 28-Aug-07
2607 ---------------------
2608
2609 1. In the rejigging of the build system that eventually resulted in 7.1, the
2610 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2611 brackets there is not right, since it causes compilers to look for an
2612 installed pcre.h, not the version that is in the source that is being
2613 compiled (which of course may be different). I have changed it back to:
2614
2615 #include "pcre.h"
2616
2617 I have a vague recollection that the change was concerned with compiling in
2618 different directories, but in the new build system, that is taken care of
2619 by the VPATH setting the Makefile.
2620
2621 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2622 when the subject happened to end in the byte 0x85 (e.g. if the last
2623 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2624 characters but of course it shouldn't be taken as a newline when it is part
2625 of another character. The bug was that, for an unlimited repeat of . in
2626 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2627 characters when looking for a newline.
2628
2629 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2630
2631 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2632 in debug output.
2633
2634 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2635 long printing in the pcrecpp unittest when running under MinGW.
2636
2637 6. ESC_K was left out of the EBCDIC table.
2638
2639 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2640 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2641 limit also applies to "virtual nesting" when a pattern is recursive, and in
2642 this case 1000 isn't so big. I have been able to remove this limit at the
2643 expense of backing off one optimization in certain circumstances. Normally,
2644 when pcre_exec() would call its internal match() function recursively and
2645 immediately return the result unconditionally, it uses a "tail recursion"
2646 feature to save stack. However, when a subpattern that can match an empty
2647 string has an unlimited repetition quantifier, it no longer makes this
2648 optimization. That gives it a stack frame in which to save the data for
2649 checking that an empty string has been matched. Previously this was taken
2650 from the 1000-entry workspace that had been reserved. So now there is no
2651 explicit limit, but more stack is used.
2652
2653 8. Applied Daniel's patches to solve problems with the import/export magic
2654 syntax that is required for Windows, and which was going wrong for the
2655 pcreposix and pcrecpp parts of the library. These were overlooked when this
2656 problem was solved for the main library.
2657
2658 9. There were some crude static tests to avoid integer overflow when computing
2659 the size of patterns that contain repeated groups with explicit upper
2660 limits. As the maximum quantifier is 65535, the maximum group length was
2661 set at 30,000 so that the product of these two numbers did not overflow a
2662 32-bit integer. However, it turns out that people want to use groups that
2663 are longer than 30,000 bytes (though not repeat them that many times).
2664 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2665 made it possible to implement the integer overflow checks in a much more
2666 dynamic way, which I have now done. The artificial limitation on group
2667 length has been removed - we now have only the limit on the total length of
2668 the compiled pattern, which depends on the LINK_SIZE setting.
2669
2670 10. Fixed a bug in the documentation for get/copy named substring when
2671 duplicate names are permitted. If none of the named substrings are set, the
2672 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2673 empty string.
2674
2675 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2676 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2677 because the ] is interpreted as the first data character and the
2678 terminating ] is not found. PCRE has been made compatible with Perl in this
2679 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2680 cause memory overwriting.
2681
2682 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2683 string has been matched (to stop an infinite loop). It was not recognizing
2684 a conditional subpattern that could match an empty string if that
2685 subpattern was within another subpattern. For example, it looped when
2686 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2687 condition was not nested. This bug has been fixed.
2688
2689 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2690 past the start of the subject in the presence of bytes with the top bit
2691 set, for example "\x8aBCD".
2692
2693 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2694 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2695
2696 14. Optimized (?!) to (*FAIL).
2697
2698 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2699 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2700 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2701 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2702 does: it's just the validity check that is more restrictive.
2703
2704 16. Inserted checks for integer overflows during escape sequence (backslash)
2705 processing, and also fixed erroneous offset values for syntax errors during
2706 backslash processing.
2707
2708 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2709 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2710
2711 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2712 caused an overrun.
2713
2714 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2715 something other than just ASCII characters) inside a group that had an
2716 unlimited repeat caused a loop at compile time (while checking to see
2717 whether the group could match an empty string).
2718
2719 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2720 [\P{Any}] did so. (Error in the code for printing property names.)
2721
2722 21. An orphan \E inside a character class could cause a crash.
2723
2724 22. A repeated capturing bracket such as (A)? could cause a wild memory
2725 reference during compilation.
2726
2727 23. There are several functions in pcre_compile() that scan along a compiled
2728 expression for various reasons (e.g. to see if it's fixed length for look
2729 behind). There were bugs in these functions when a repeated \p or \P was
2730 present in the pattern. These operators have additional parameters compared
2731 with \d, etc, and these were not being taken into account when moving along
2732 the compiled data. Specifically:
2733
2734 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2735 length.
2736
2737 (b) An item such as \pL+ within a repeated group could cause crashes or
2738 loops.
2739
2740 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2741 "reference to non-existent subpattern" error.
2742
2743 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2744
2745 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2746 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2747
2748 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2749
2750 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2751 character were causing crashes (broken optimization).
2752
2753 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2754 \p or \P) caused a compile-time loop.
2755
2756 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2757 break. For example, the unstudied pattern [\r\n]A does not match the string
2758 "\r\nA" because change 7.0/46 below moves the current point on by two
2759 characters after failing to match at the start. However, the pattern \nA
2760 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2761 the same is true. There doesn't seem any very clean way out of this, but
2762 what I have chosen to do makes the common cases work: PCRE now takes note
2763 of whether there can be an explicit match for \r or \n anywhere in the
2764 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2765 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2766 pattern has explicit CR or LF references.
2767
2768 29. Added (*CR) etc for changing newline setting at start of pattern.
2769
2770
2771 Version 7.2 19-Jun-07
2772 ---------------------
2773
2774 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2775 which is apparently normally available under Windows.
2776
2777 2. Re-jig the pcregrep tests with different newline settings in an attempt
2778 to make them independent of the local environment's newline setting.
2779
2780 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2781
2782 4. Some of the "internals" tests were previously cut out when the link size
2783 was not 2, because the output contained actual offsets. The recent new
2784 "Z" feature of pcretest means that these can be cut out, making the tests
2785 usable with all link sizes.
2786
2787 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2788 stack recursion. This gives a massive performance boost under BSD, but just
2789 a small improvement under Linux. However, it saves one field in the frame
2790 in all cases.
2791
2792 6. Added more features from the forthcoming Perl 5.10:
2793
2794 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2795 recursion call. It refers to the nth most recently opened parentheses.
2796
2797 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2798 to be opened parentheses.
2799
2800 (c) Conditions that refer to capturing parentheses can be specified
2801 relatively, for example, (?(-2)... or (?(+3)...
2802
2803 (d) \K resets the start of the current match so that everything before
2804 is not part of it.
2805
2806 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2807
2808 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2809 reference syntax.
2810
2811 (g) (?| introduces a group in which the numbering of parentheses in each
2812 alternative starts with the same number.
2813
2814 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2815
2816 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2817 PCRE_INFO_JCHANGED.
2818
2819 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2820 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2821 for detecting groups that can match an empty string.
2822
2823 9. A pattern with a very large number of alternatives (more than several
2824 hundred) was running out of internal workspace during the pre-compile
2825 phase, where pcre_compile() figures out how much memory will be needed. A
2826 bit of new cunning has reduced the workspace needed for groups with
2827 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2828 workspace instead of running out of the 4096 that are available.
2829
2830 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2831
2832 11. Applied patch from Google to remove an optimization that didn't quite work.
2833 The report of the bug said:
2834
2835 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2836 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2837 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2838
2839 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2840 it matched the wrong number of bytes.
2841
2842
2843 Version 7.1 24-Apr-07
2844 ---------------------
2845
2846 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2847 that is more "standard", making use of automake and other Autotools. There
2848 is some re-arrangement of the files and adjustment of comments consequent
2849 on this.
2850
2851 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2852 for recursive directory scanning broke on some systems because the files
2853 are not scanned in any specific order and on different systems the order
2854 was different. A call to "sort" has been inserted into RunGrepTest for the
2855 approprate test as a short-term fix. In the longer term there may be an
2856 alternative.
2857
2858 3. I had an email from Eric Raymond about problems translating some of PCRE's
2859 man pages to HTML (despite the fact that I distribute HTML pages, some
2860 people do their own conversions for various reasons). The problems
2861 concerned the use of low-level troff macros .br and .in. I have therefore
2862 removed all such uses from the man pages (some were redundant, some could
2863 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2864 HTML has been updated to handle .nf/.fi and to complain if it encounters
2865 .br or .in.
2866
2867 4. Updated comments in configure.ac that get placed in config.h.in and also
2868 arranged for config.h to be included in the distribution, with the name
2869 config.h.generic, for the benefit of those who have to compile without
2870 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2871
2872 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2873 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2874 makevp.bat for the current PCRE, using the additional files
2875 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2876
2877 6. A Windows user reported a minor discrepancy with test 2, which turned out
2878 to be caused by a trailing space on an input line that had got lost in his
2879 copy. The trailing space was an accident, so I've just removed it.
2880
2881 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2882 that is needed.
2883
2884 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2885 as "const" (a) because they are and (b) because it helps the PHP
2886 maintainers who have recently made a script to detect big data structures
2887 in the php code that should be moved to the .rodata section. I remembered
2888 to update Builducptable as well, so it won't revert if ucptable.h is ever
2889 re-created.
2890
2891 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2892 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2893 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2894 support is not required. This saves 1.5-2K of code, which is important in
2895 some applications.
2896
2897 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2898 so as not to refer to the tables, even though these functions will never be
2899 called when UTF-8 support is disabled. Otherwise there are problems with a
2900 shared library.
2901
2902 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2903
2904 (a) It was defining its arguments as char * instead of void *.
2905
2906 (b) It was assuming that all moves were upwards in memory; this was true
2907 a long time ago when I wrote it, but is no longer the case.
2908
2909 The emulated memove() is provided for those environments that have neither
2910 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2911 is clearly not the case, as these two bugs were recently reported.
2912
2913 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2914 and Detrail to create the HTML documentation, the .txt form of the man
2915 pages, and it removes trailing spaces from listed files. It also creates
2916 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2917 case, it wraps all the #defines with #ifndefs. This script should be run
2918 before "make dist".
2919
2920 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2921 with Unicode property support.
2922
2923 (a) For a maximizing quantifier, if the two different cases of the
2924 character were of different lengths in their UTF-8 codings (there are
2925 some cases like this - I found 11), and the matching function had to
2926 back up over a mixture of the two cases, it incorrectly assumed they
2927 were both the same length.
2928
2929 (b) When PCRE was configured to use the heap rather than the stack for
2930 recursion during matching, it was not correctly preserving the data for
2931 the other case of a UTF-8 character when checking ahead for a match
2932 while processing a minimizing repeat. If the check also involved
2933 matching a wide character, but failed, corruption could cause an
2934 erroneous result when trying to check for a repeat of the original
2935 character.
2936
2937 13. Some tidying changes to the testing mechanism:
2938
2939 (a) The RunTest script now detects the internal link size and whether there
2940 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2941 values substituted by "configure". (The RunGrepTest script already did
2942 this for UTF-8.) The configure.ac script no longer substitutes the
2943 relevant variables.
2944
2945 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2946 with length and offset values. This means that the output is different
2947 for different internal link sizes. Test 2 is skipped for link sizes
2948 other than 2 because of this, bypassing the problem. Unfortunately,
2949 there was also a test in test 3 (the locale tests) that used /B and
2950 failed for link sizes other than 2. Rather than cut the whole test out,
2951 I have added a new /Z option to pcretest that replaces the length and
2952 offset values with spaces. This is now used to make test 3 independent
2953 of link size. (Test 2 will be tidied up later.)
2954
2955 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2956 segmentation fault instead of returning the appropriate error message.
2957
2958 15. In multiline mode when the newline sequence was set to "any", the pattern
2959 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2960 This doesn't seem right; it now treats the CRLF combination as the line
2961 ending, and so does not match in that case. It's only a pattern such as ^$
2962 that would hit this one: something like ^ABC$ would have failed after \r
2963 and then tried again after \r\n.
2964
2965 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2966 in an attempt to make files that differ only in their line terminators
2967 compare equal. This works on Linux.
2968
2969 17. Under certain error circumstances pcregrep might try to free random memory
2970 as it exited. This is now fixed, thanks to valgrind.
2971
2972 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2973 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2974 was because its rules for how to advance for /g after matching an empty
2975 string at the end of a line did not allow for this case. They now check for
2976 it specially.
2977
2978 20. pcretest is supposed to handle patterns and data of any length, by
2979 extending its buffers when necessary. It was getting this wrong when the
2980 buffer for a data line had to be extended.
2981
2982 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2983 CRLF as a newline sequence.
2984
2985 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2986 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2987 I have nevertheless tidied it up.
2988
2989 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2990
2991 24. Added a man page for pcre-config.
2992
2993
2994 Version 7.0 19-Dec-06
2995 ---------------------
2996
2997 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2998 moving to gcc 4.1.1.
2999
3000 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3001 sys/time.h, which is documented as needed for this function. It doesn't
3002 seem to matter on Linux, but it showed up on some releases of OS X.
3003
3004 3. It seems that there are systems where bytes whose values are greater than
3005 127 match isprint() in the "C" locale. The "C" locale should be the
3006 default when a C program starts up. In most systems, only ASCII printing
3007 characters match isprint(). This difference caused the output from pcretest
3008 to vary, making some of the tests fail. I have changed pcretest so that:
3009
3010 (a) When it is outputting text in the compiled version of a pattern, bytes
3011 other than 32-126 are always shown as hex escapes.
3012
3013 (b) When it is outputting text that is a matched part of a subject string,
3014 it does the same, unless a different locale has been set for the match
3015 (using the /L modifier). In this case, it uses isprint() to decide.
3016
3017 4. Fixed a major bug that caused incorrect computation of the amount of memory
3018 required for a compiled pattern when options that changed within the
3019 pattern affected the logic of the preliminary scan that determines the
3020 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3021 that the computed length was too small. The symptoms of this bug were
3022 either the PCRE error "internal error: code overflow" from pcre_compile(),
3023 or a glibc crash with a message such as "pcretest: free(): invalid next
3024 size (fast)". Examples of patterns that provoked this bug (shown in
3025 pcretest format) are:
3026
3027 /(?-x: )/x
3028 /(?x)(?-x: \s*#\s*)/
3029 /((?i)[\x{c0}])/8
3030 /(?i:[\x{c0}])/8
3031
3032 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3033 is now done differently.
3034
3035 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3036 wrapper classes; (b) implement a new function in the C++ scanner that is
3037 more efficient than the old way of doing things because it avoids levels of
3038 recursion in the regex matching; (c) add a paragraph to the documentation
3039 for the FullMatch() function.
3040
3041 6. The escape sequence \n was being treated as whatever was defined as
3042 "newline". Not only was this contrary to the documentation, which states
3043 that \n is character 10 (hex 0A), but it also went horribly wrong when
3044 "newline" was defined as CRLF. This has been fixed.
3045
3046 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3047 was being set to -1 for the "end of line" case (supposedly a value that no
3048 character can have). Though this value is never used (the check for end of
3049 line is "zero bytes in current character"), it caused compiler complaints.
3050 I've changed it to 0xffffffff.
3051
3052 8. In pcre_version.c, the version string was being built by a sequence of
3053 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3054 string (as it is for production releases) called a macro with an empty
3055 argument. The C standard says the result of this is undefined. The gcc
3056 compiler treats it as an empty string (which was what was wanted) but it is
3057 reported that Visual C gives an error. The source has been hacked around to
3058 avoid this problem.
3059
3060 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3061 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3062 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3063 of them did).
3064
3065 10. Originally, pcretest opened its input and output without "b"; then I was
3066 told that "b" was needed in some environments, so it was added for release
3067 5.0 to both the input and output. (It makes no difference on Unix-like
3068 systems.) Later I was told that it is wrong for the input on Windows. I've
3069 now abstracted the modes into two macros, to make it easier to fiddle with
3070 them, and removed "b" from the input mode under Windows.
3071
3072 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3073
3074 12. Added -help and --help to pcretest as an official way of being reminded
3075 of the options.
3076
3077 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3078 and pcrecpp.cc because they annoy compilers at high warning levels.
3079
3080 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3081
3082 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3083 scripts are not C programs :-) and which was not noticed because it works
3084 on Linux.
3085
3086 16. pcretest is supposed to handle any length of pattern and data line (as one
3087 line or as a continued sequence of lines) by extending its input buffer if
3088 necessary. This feature was broken for very long pattern lines, leading to
3089 a string of junk being passed to pcre_compile() if the pattern was longer
3090 than about 50K.
3091
3092 17. I have done a major re-factoring of the way pcre_compile() computes the
3093 amount of memory needed for a compiled pattern. Previously, there was code
3094 that made a preliminary scan of the pattern in order to do this. That was
3095 OK when PCRE was new, but as the facilities have expanded, it has become
3096 harder and harder to keep it in step with the real compile phase, and there
3097 have been a number of bugs (see for example, 4 above). I have now found a
3098 cunning way of running the real compile function in a "fake" mode that
3099 enables it to compute how much memory it would need, while actually only
3100 ever using a few hundred bytes of working memory and without too many
3101 tests of the mode. This should make future maintenance and development
3102 easier. A side effect of this work is that the limit of 200 on the nesting
3103 depth of parentheses has been removed (though this was never a serious
3104 limitation, I suspect). However, there is a downside: pcre_compile() now
3105 runs more slowly than before (30% or more, depending on the pattern). I
3106 hope this isn't a big issue. There is no effect on runtime performance.
3107
3108 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3109 newline (only possible for the last line of a file) and it was a
3110 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3111
3112 19. Added additional timing features to pcretest. (1) The -tm option now times
3113 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3114 separate command line item, by a number that specifies the number of
3115 repeats to use when timing. The default is 50000; this gives better
3116 precision, but takes uncomfortably long for very large patterns.
3117
3118 20. Extended pcre_study() to be more clever in cases where a branch of a
3119 subpattern has no definite first character. For example, (a*|b*)[cd] would
3120 previously give no result from pcre_study(). Now it recognizes that the
3121 first character must be a, b, c, or d.
3122
3123 21. There was an incorrect error "recursive call could loop indefinitely" if
3124 a subpattern (or the entire pattern) that was being tested for matching an
3125 empty string contained only one non-empty item after a nested subpattern.
3126 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3127 incorrectly, because the \d was being skipped in the check.
3128
3129 22. The pcretest program now has a new pattern option /B and a command line
3130 option -b, which is equivalent to adding /B to every pattern. This causes
3131 it to show the compiled bytecode, without the additional information that
3132 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3133 is the same as /B/I).
3134
3135 23. A new optimization is now able automatically to treat some sequences such
3136 as a*b as a*+b. More specifically, if something simple (such as a character
3137 or a simple class like \d) has an unlimited quantifier, and is followed by
3138 something that cannot possibly match the quantified thing, the quantifier
3139 is automatically "possessified".
3140
3141 24. A recursive reference to a subpattern whose number was greater than 39
3142 went wrong under certain circumstances in UTF-8 mode. This bug could also
3143 have affected the operation of pcre_study().
3144
3145 25. Realized that a little bit of performance could be had by replacing
3146 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3147
3148 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3149
3150 27. Possessive quantifiers such as a++ were previously implemented by turning
3151 them into atomic groups such as ($>a+). Now they have their own opcodes,
3152 which improves performance. This includes the automatically created ones
3153 from 23 above.
3154
3155 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3156 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3157 the first matched character to be a colon. This applied both to named and
3158 numbered groups.
3159
3160 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3161
3162 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3163 building PCRE on Windows easier, so I have included it in the distribution.
3164
3165 31. There is now a check in pcretest against a ridiculously large number being
3166 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3167 loop, the loop is abandoned.
3168
3169 32. Forward references to subpatterns in conditions such as (?(2)...) where
3170 subpattern 2 is defined later cause pcre_compile() to search forwards in
3171 the pattern for the relevant set of parentheses. This search went wrong
3172 when there were unescaped parentheses in a character class, parentheses
3173 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3174
3175 33. "Subroutine" calls and backreferences were previously restricted to
3176 referencing subpatterns earlier in the regex. This restriction has now
3177 been removed.
3178
3179 34. Added a number of extra features that are going to be in Perl 5.10. On the
3180 whole, these are just syntactic alternatives for features that PCRE had
3181 previously implemented using the Python syntax or my own invention. The
3182 other formats are all retained for compatibility.
3183
3184 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3185 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3186 also .NET compatible.
3187
3188 (b) A recursion or subroutine call to a named group can now be defined as
3189 (?&name) as well as (?P>name).
3190
3191 (c) A backreference to a named group can now be defined as \k<name> or
3192 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3193 5.10, are also .NET compatible.
3194
3195 (d) A conditional reference to a named group can now use the syntax
3196 (?(<name>) or (?('name') as well as (?(name).
3197
3198 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3199 groups (named and numbered) that are never evaluated inline, but can be
3200 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3201 is always false. There may be only one alternative in such a group.
3202
3203 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3204 as the simple (?(R). The condition is true only if the most recent
3205 recursion is that of the given number or name. It does not search out
3206 through the entire recursion stack.
3207
3208 (g) The escape \gN or \g{N} has been added, where N is a positive or
3209 negative number, specifying an absolute or relative reference.
3210
3211 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3212 some "unreachable code" warnings.
3213
3214 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3215 things, this adds five new scripts.
3216
3217 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3218 There were also incompatibilities regarding the handling of \Q..\E inside
3219 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3220 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3221
3222 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3223 matches an empty string, and forcibly breaks the loop. There were bugs in
3224 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3225 against aaaa the result was just "a" rather than "aaaa", for example. Two
3226 separate and independent bugs (that affected different cases) have been
3227 fixed.
3228
3229 39. Refactored the code to abolish the use of different opcodes for small
3230 capturing bracket numbers. This is a tidy that I avoided doing when I
3231 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3232 The new approach is not only tidier, it makes it possible to reduce the
3233 memory needed to fix the previous bug (38).
3234
3235 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3236 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3237 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3238 mode.
3239
3240 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3241 report.
3242
3243 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3244 copy construction and assignment in the C++ wrapper.
3245
3246 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3247 couple of bugs that could have given wrong results in the "--newline=crlf"
3248 case.
3249
3250 44. Added a number of casts and did some reorganization of signed/unsigned int
3251 variables following suggestions from Dair Grant. Also renamed the variable
3252 "this" as "item" because it is a C++ keyword.
3253
3254 45. Arranged for dftables to add
3255
3256 #include "pcre_internal.h"
3257
3258 to pcre_chartables.c because without it, gcc 4.x may remove the array
3259 definition from the final binary if PCRE is built into a static library and
3260 dead code stripping is activated.
3261
3262 46. For an unanchored pattern, if a match attempt fails at the start of a
3263 newline sequence, and the newline setting is CRLF or ANY, and the next two
3264 characters are CRLF, advance by two characters instead of one.
3265
3266
3267 Version 6.7 04-Jul-06
3268 ---------------------
3269
3270 1. In order to handle tests when input lines are enormously long, pcretest has
3271 been re-factored so that it automatically extends its buffers when
3272 necessary. The code is crude, but this _is_ just a test program. The
3273 default size has been increased from 32K to 50K.
3274
3275 2. The code in pcre_study() was using the value of the re argument before
3276 testing it for NULL. (Of course, in any sensible call of the function, it
3277 won't be NULL.)
3278
3279 3. The memmove() emulation function in pcre_internal.h, which is used on
3280 systems that lack both memmove() and bcopy() - that is, hardly ever -
3281 was missing a "static" storage class specifier.
3282
3283 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3284 containing an extended class (one that cannot be represented by a bitmap
3285 because it contains high-valued characters or Unicode property items, e.g.
3286 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3287 pattern, but PCRE should not loop if you do not (it no longer does).
3288 [Detail: two cases were found: (a) a repeated subpattern containing an
3289 extended class; (b) a recursive reference to a subpattern that followed a
3290 previous extended class. It wasn't skipping over the extended class
3291 correctly when UTF-8 mode was not set.]
3292
3293 5. A negated single-character class was not being recognized as fixed-length
3294 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3295 compile error "lookbehind assertion is not fixed length".
3296
3297 6. The RunPerlTest auxiliary script was showing an unexpected difference
3298 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3299 write a Perl script that can interpret lines of an input file either as
3300 byte characters or as UTF-8, which is what "perltest" was being required to
3301 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3302 can't do is switch easily at run time between having the "use utf8;" pragma
3303 or not. In the end, I fudged it by using the RunPerlTest script to insert
3304 "use utf8;" explicitly for the UTF-8 tests.
3305
3306 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3307 the end of the subject string, contrary to the documentation and to what
3308 Perl does. This was true of both matching functions. Now it matches only at
3309 the start of the subject and immediately after *internal* newlines.
3310
3311 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3312 a pointer to an int instead of a pointer to an unsigned long int. This
3313 caused problems on 64-bit systems.
3314
3315 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3316 instance of the 'standard' template library not being so standard".
3317
3318 10. There was no check on the number of named subpatterns nor the maximum
3319 length of a subpattern name. The product of these values is used to compute
3320 the size of the memory block for a compiled pattern. By supplying a very
3321 long subpattern name and a large number of named subpatterns, the size
3322 computation could be caused to overflow. This is now prevented by limiting
3323 the length of names to 32 characters, and the number of named subpatterns
3324 to 10,000.
3325
3326 11. Subpatterns that are repeated with specific counts have to be replicated in
3327 the compiled pattern. The size of memory for this was computed from the
3328 length of the subpattern and the repeat count. The latter is limited to
3329 65535, but there was no limit on the former, meaning that integer overflow
3330 could in principle occur. The compiled length of a repeated subpattern is
3331 now limited to 30,000 bytes in order to prevent this.
3332
3333 12. Added the optional facility to have named substrings with the same name.
3334
3335 13. Added the ability to use a named substring as a condition, using the
3336 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3337 are numbers (not recommended). Forward references are permitted.
3338
3339 14. Added forward references in named backreferences (if you see what I mean).
3340
3341 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3342 pattern could run off the end of the subject. For example, the pattern
3343 "(?s)(.{1,5})"8 did this with the subject "ab".
3344
3345 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3346 PCRE_CASELESS was set when matching characters that were quantified with ?
3347 or *.
3348
3349 17. A character class other than a single negated character that had a minimum
3350 but no maximum quantifier - for example [ab]{6,} - was not handled
3351 correctly by pce_dfa_exec(). It would match only one character.
3352
3353 18. A valid (though odd) pattern that looked like a POSIX character
3354 class but used an invalid character after [ (for example [[,abc,]]) caused
3355 pcre_compile() to give the error "Failed: internal error: code overflow" or
3356 in some cases to crash with a glibc free() error. This could even happen if
3357 the pattern terminated after [[ but there just happened to be a sequence of
3358 letters, a binary zero, and a closing ] in the memory that followed.
3359
3360 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3361 over the years. Originally (before any Unicode support), just the bottom 8
3362 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3363 output from "man perlunicode" includes this:
3364
3365 The regular expression compiler produces polymorphic opcodes. That
3366 is, the pattern adapts to the data and automatically switches to
3367 the Unicode character scheme when presented with Unicode data--or
3368 instead uses a traditional byte scheme when presented with byte
3369 data.
3370
3371 Sadly, a wide octal escape does not cause a switch, and in a string with
3372 no other multibyte characters, these octal escapes are treated as before.
3373 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3374 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3375 Unicode string.
3376
3377 I have not perpetrated such confusion in PCRE. Up till now, it took just
3378 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3379 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3380 translate to the appropriate multibyte character.
3381
3382 29. Applied some refactoring to reduce the number of warnings from Microsoft
3383 and Borland compilers. This has included removing the fudge introduced
3384 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3385 a warning about an unused variable.
3386
3387 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3388 characters since release 4.0, because Perl (from release 5.004) does not.
3389 [Or at least, is documented not to: some releases seem to be in conflict
3390 with the documentation.] However, when a pattern was studied with
3391 pcre_study() and all its branches started with \s, PCRE still included VT
3392 as a possible starting character. Of course, this did no harm; it just
3393 caused an unnecessary match attempt.
3394
3395 22. Removed a now-redundant internal flag bit that recorded the fact that case
3396 dependency changed within the pattern. This was once needed for "required
3397 byte" processing, but is no longer used. This recovers a now-scarce options
3398 bit. Also moved the least significant internal flag bit to the most-
3399 significant bit of the word, which was not previously used (hangover from
3400 the days when it was an int rather than a uint) to free up another bit for
3401 the future.
3402
3403 23. Added support for CRLF line endings as well as CR and LF. As well as the
3404 default being selectable at build time, it can now be changed at runtime
3405 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3406 specify that it is scanning data with non-default line endings.
3407
3408 24. Changed the definition of CXXLINK to make it agree with the definition of
3409 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3410
3411 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3412 recursions. This makes a big different to stack usage for some patterns.
3413
3414 26. If a subpattern containing a named recursion or subroutine reference such
3415 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3416 the space required for the compiled pattern went wrong and gave too small a
3417 value. Depending on the environment, this could lead to "Failed: internal
3418 error: code overflow at offset 49" or "glibc detected double free or
3419 corruption" errors.
3420
3421 27. Applied patches from Google (a) to support the new newline modes and (b) to
3422 advance over multibyte UTF-8 characters in GlobalReplace.
3423
3424 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3425 difference for some implementation of PCRE in some Windows version.
3426
3427 29. Added some extra testing facilities to pcretest:
3428
3429 \q<number> in a data line sets the "match limit" value
3430 \Q<number> in a data line sets the "match recursion limt" value
3431 -S <number> sets the stack size, where <number> is in megabytes
3432
3433 The -S option isn't available for Windows.
3434
3435
3436 Version 6.6 06-Feb-06
3437 ---------------------
3438
3439 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3440 in pcreposix.h. I have copied the definition from pcre.h.
3441
3442 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3443 because pcre.h is no longer a built file.
3444
3445 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3446 not normally included in the compiled code.
3447
3448
3449 Version 6.5 01-Feb-06
3450 ---------------------
3451
3452 1. When using the partial match feature with pcre_dfa_exec(), it was not
3453 anchoring the second and subsequent partial matches at the new starting
3454 point. This could lead to incorrect results. For example, with the pattern
3455 /1234/, partially matching against "123" and then "a4" gave a match.
3456
3457 2. Changes to pcregrep:
3458
3459 (a) All non-match returns from pcre_exec() were being treated as failures
3460 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3461 error message is output. Some extra information is given for the
3462 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3463 probably the only errors that are likely to be caused by users (by
3464 specifying a regex that has nested indefinite repeats, for instance).
3465 If there are more than 20 of these errors, pcregrep is abandoned.
3466
3467 (b) A binary zero was treated as data while matching, but terminated the
3468 output line if it was written out. This has been fixed: binary zeroes
3469 are now no different to any other data bytes.
3470
3471 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3472 used to set a locale for matching. The --locale=xxxx long option has
3473 been added (no short equivalent) to specify a locale explicitly on the
3474 pcregrep command, overriding the environment variables.
3475
3476 (d) When -B was used with -n, some line numbers in the output were one less
3477 than they should have been.
3478
3479 (e) Added the -o (--only-matching) option.
3480
3481 (f) If -A or -C was used with -c (count only), some lines of context were
3482 accidentally printed for the final match.
3483
3484 (g) Added the -H (--with-filename) option.
3485
3486 (h) The combination of options -rh failed to suppress file names for files
3487 that were found from directory arguments.
3488
3489 (i) Added the -D (--devices) and -d (--directories) options.
3490
3491 (j) Added the -F (--fixed-strings) option.
3492
3493 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3494
3495 (l) Added the --colo(u)r option.
3496
3497 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3498 is not present by default.
3499
3500 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3501 items such as (?R) or (?1), when the recursion could match a number of
3502 alternatives. If it matched one of the alternatives, but subsequently,
3503 outside the recursion, there was a failure, the code tried to back up into
3504 the recursion. However, because of the way PCRE is implemented, this is not
3505 possible, and the result was an incorrect result from the match.
3506
3507 In order to prevent this happening, the specification of recursion has
3508 been changed so that all such subpatterns are automatically treated as
3509 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3510
3511 4. I had overlooked the fact that, in some locales, there are characters for
3512 which isalpha() is true but neither isupper() nor islower() are true. In
3513 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3514 and ordfeminine) are like this. This affected the treatment of \w and \W
3515 when they appeared in character classes, but not when they appeared outside
3516 a character class. The bit map for "word" characters is now created
3517 separately from the results of isalnum() instead of just taking it from the
3518 upper, lower, and digit maps. (Plus the underscore character, of course.)
3519
3520 5. The above bug also affected the handling of POSIX character classes such as
3521 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3522 permanent tables. Instead, the bit maps for such a class were previously
3523 created as the appropriate unions of the upper, lower, and digit bitmaps.
3524 Now they are created by subtraction from the [[:word:]] class, which has
3525 its own bitmap.
3526
3527 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3528 It is created by subtracting the vertical space characters (\x09, \x0a,
3529 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3530 subtraction was done in the overall bitmap for a character class, meaning
3531 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3532 be recognized. This bug has been fixed.
3533
3534 7. Patches from the folks at Google:
3535
3536 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3537 real life, but is still worth protecting against".
3538
3539 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3540 regular expressions".
3541
3542 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3543 have it.
3544
3545 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3546 "configure" and the latter not, in order to fix a problem somebody had
3547 with compiling the Arg class on HP-UX.
3548
3549 (e) Improve the error-handling of the C++ wrapper a little bit.
3550
3551 (f) New tests for checking recursion limiting.
3552
3553 8. The pcre_memmove() function, which is used only if the environment does not
3554 have a standard memmove() function (and is therefore rarely compiled),
3555 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3556 returning a result (though PCRE never actually uses the result).
3557
3558 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3559 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3560 returned instead of calling malloc() with an overflowing number that would
3561 most likely cause subsequent chaos.
3562
3563 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3564
3565 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3566 with this option is matched, the nmatch and pmatch options of regexec() are
3567 ignored.
3568
3569 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3570 provided in case anyone wants to the the POSIX interface with UTF-8
3571 strings.
3572
3573 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3574 C++ linking (needed for some HP-UX environments).
3575
3576 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3577 (unused parameter) and in the pcre_printint() function (omitted "default"
3578 switch label when the default is to do nothing).
3579
3580 15. Added some code to make it possible, when PCRE is compiled as a C++
3581 library, to replace subject pointers for pcre_exec() with a smart pointer
3582 class, thus making it possible to process discontinuous strings.
3583
3584 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3585 much the same function. They were added by different people who were trying
3586 to make PCRE easy to compile on non-Unix systems. It has been suggested
3587 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3588 for compiling on Windows systems. I have therefore replaced it with
3589 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3590 defaults to "extern" for C or "extern C" for C++, which works fine on
3591 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3592 SCOPE with something explicit in config.h. In addition:
3593
3594 (a) pcreposix.h still had just "extern" instead of either of these macros;
3595 I have replaced it with PCRE_DATA_SCOPE.
3596
3597 (b) Functions such as _pcre_xclass(), which are internal to the library,
3598 but external in the C sense, all had PCRE_EXPORT in their definitions.
3599 This is apparently wrong for the Windows case, so I have removed it.
3600 (It makes no difference on Unix-like systems.)
3601
3602 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3603 of recursive calls to match(). This is different to MATCH_LIMIT because
3604 that limits the total number of calls to match(), not all of which increase
3605 the depth of recursion. Limiting the recursion depth limits the amount of
3606 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3607 when PCRE is compiled, and changed at run time. A patch from Google adds
3608 this functionality to the C++ interface.
3609
3610 18. Changes to the handling of Unicode character properties:
3611
3612 (a) Updated the table to Unicode 4.1.0.
3613
3614 (b) Recognize characters that are not in the table as "Cn" (undefined).
3615
3616 (c) I revised the way the table is implemented to a much improved format
3617 which includes recognition of ranges. It now supports the ranges that
3618 are defined in UnicodeData.txt, and it also amalgamates other
3619 characters into ranges. This has reduced the number of entries in the
3620 table from around 16,000 to around 3,000, thus reducing its size
3621 considerably. I realized I did not need to use a tree structure after
3622 all - a binary chop search is just as efficient. Having reduced the
3623 number of entries, I extended their size from 6 bytes to 8 bytes to
3624 allow for more data.
3625
3626 (d) Added support for Unicode script names via properties such as \p{Han}.
3627
3628 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3629 matching that character.
3630
3631 20. When matching a repeated Unicode property with a minimum greater than zero,
3632 (for example \pL{2,}), PCRE could look past the end of the subject if it
3633 reached it while seeking the minimum number of characters. This could
3634 happen only if some of the characters were more than one byte long, because
3635 there is a check for at least the minimum number of bytes.
3636
3637 21. Refactored the implementation of \p and \P so as to be more general, to
3638 allow for more different types of property in future. This has changed the
3639 compiled form incompatibly. Anybody with saved compiled patterns that use
3640 \p or \P will have to recompile them.
3641
3642 22. Added "Any" and "L&" to the supported property types.
3643
3644 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3645 but give a compile time error if the value is greater than 0xff.
3646
3647 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3648 accidentally not being installed or uninstalled.
3649
3650 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3651 made were to insert the current release number. This seemed silly, because
3652 it made things harder for people building PCRE on systems that don't run
3653 "configure". I have turned pcre.h into a distributed file, no longer built
3654 by "configure", with the version identification directly included. There is
3655 no longer a pcre.h.in file.
3656
3657 However, this change necessitated a change to the pcre-config script as
3658 well. It is built from pcre-config.in, and one of the substitutions was the
3659 release number. I have updated configure.ac so that ./configure now finds
3660 the release number by grepping pcre.h.
3661
3662 26. Added the ability to run the tests under valgrind.
3663
3664
3665 Version 6.4 05-Sep-05
3666 ---------------------
3667
3668 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3669 "--" to be printed when multiple files were scanned, even when none of the
3670 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3671 consider it to be a bug, and have restored the previous behaviour.
3672
3673 2. A couple of code tidies to get rid of compiler warnings.
3674
3675 3. The pcretest program used to cheat by referring to symbols in the library
3676 whose names begin with _pcre_. These are internal symbols that are not
3677 really supposed to be visible externally, and in some environments it is
3678 possible to suppress them. The cheating is now confined to including
3679 certain files from the library's source, which is a bit cleaner.
3680
3681 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3682 file's purpose clearer.
3683
3684 5. Reorganized pcre_ucp_findchar().
3685
3686
3687 Version 6.3 15-Aug-05
3688 ---------------------
3689
3690 1. The file libpcre.pc.in did not have general read permission in the tarball.
3691
3692 2. There were some problems when building without C++ support:
3693
3694 (a) If C++ support was not built, "make install" and "make test" still
3695 tried to test it.
3696
3697 (b) There were problems when the value of CXX was explicitly set. Some
3698 changes have been made to try to fix these, and ...
3699
3700 (c) --disable-cpp can now be used to explicitly disable C++ support.
3701
3702 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3703 backslash in a target when C++ was disabled. This confuses some
3704 versions of "make", apparently. Using an intermediate variable solves
3705 this. (Same for CPP_LOBJ.)
3706
3707 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3708 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3709 necessary on certain architectures.
3710
3711 4. Added a setting of -export-symbols-regex to the link command to remove
3712 those symbols that are exported in the C sense, but actually are local
3713 within the library, and not documented. Their names all begin with
3714 "_pcre_". This is not a perfect job, because (a) we have to except some
3715 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3716 available (and never for static libraries). I have made a note to try to
3717 find a way round (a) in the future.
3718
3719
3720 Version 6.2 01-Aug-05
3721 ---------------------
3722
3723 1. There was no test for integer overflow of quantifier values. A construction
3724 such as {1111111111111111} would give undefined results. What is worse, if
3725 a minimum quantifier for a parenthesized subpattern overflowed and became
3726 negative, the calculation of the memory size went wrong. This could have
3727 led to memory overwriting.
3728
3729 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3730
3731 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3732 operating environments where this matters.
3733
3734 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3735 PCRE options from within the C++ wrapper.
3736
3737 5. Named capturing subpatterns were not being correctly counted when a pattern
3738 was compiled. This caused two problems: (a) If there were more than 100
3739 such subpatterns, the calculation of the memory needed for the whole
3740 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3741 back references of the form \12, where the number was greater than 9, were
3742 not recognized as back references, even though there were sufficient
3743 previous subpatterns.
3744
3745 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3746 versions of gcc, e.g. 2.95.4.
3747
3748
3749 Version 6.1 21-Jun-05
3750 ---------------------
3751
3752 1. There was one reference to the variable "posix" in pcretest.c that was not
3753 surrounded by "#if !defined NOPOSIX".
3754
3755 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3756 the cross-check on the old pcre_info() function, for the benefit of the
3757 cut-down version of PCRE that is currently imported into Exim.
3758
3759 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3760 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3761 patterns that start (?i) but I don't think that matters. The use of (?i) is
3762 just an example; this all applies to the other options as well.
3763
3764 4. Since libtool seems to echo the compile commands it is issuing, the output
3765 from "make" can be reduced a bit by putting "@" in front of each libtool
3766 compile command.
3767
3768 5. Patch from the folks at Google for configure.in to be a bit more thorough
3769 in checking for a suitable C++ installation before trying to compile the
3770 C++ stuff. This should fix a reported problem when a compiler was present,
3771 but no suitable headers.
3772
3773 6. The man pages all had just "PCRE" as their title. I have changed them to
3774 be the relevant file name. I have also arranged that these names are
3775 retained in the file doc/pcre.txt, which is a concatenation in text format
3776 of all the man pages except the little individual ones for each function.
3777
3778 7. The NON-UNIX-USE file had not been updated for the different set of source
3779 files that come with release 6. I also added a few comments about the C++
3780 wrapper.
3781
3782
3783 Version 6.0 07-Jun-05
3784 ---------------------
3785
3786 1. Some minor internal re-organization to help with my DFA experiments.
3787
3788 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3789 didn't matter for the library itself when fully configured, but did matter
3790 when compiling without UCP support, or within Exim, where the ucp files are
3791 not imported.
3792
3793 3. Refactoring of the library code to split up the various functions into
3794 different source modules. The addition of the new DFA matching code (see
3795 below) to a single monolithic source would have made it really too
3796 unwieldy, quite apart from causing all the code to be include in a
3797 statically linked application, when only some functions are used. This is
3798 relevant even without the DFA addition now that patterns can be compiled in
3799 one application and matched in another.
3800
3801 The downside of splitting up is that there have to be some external
3802 functions and data tables that are used internally in different modules of
3803 the library but which are not part of the API. These have all had their
3804 names changed to start with "_pcre_" so that they are unlikely to clash
3805 with other external names.
3806
3807 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3808 a different (DFA) algorithm. Although it is slower than the original
3809 function, it does have some advantages for certain types of matching
3810 problem.
3811
3812 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3813 including restarting after a partial match.
3814
3815 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3816 defined when compiling for Windows was sent to me. I have put it into the
3817 code, though I have no means of testing or verifying it.
3818
3819 7. Added the pcre_refcount() auxiliary function.
3820
3821 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3822 match before or at the first newline in the subject string. In pcretest,
3823 the /f option on a pattern can be used to set this.
3824
3825 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3826 would behave wrongly. This has been present in PCRE since release 4.0.
3827
3828 10. A number of changes to the pcregrep command:
3829
3830 (a) Refactored how -x works; insert ^(...)$ instead of setting
3831 PCRE_ANCHORED and checking the length, in preparation for adding
3832 something similar for -w.
3833
3834 (b) Added the -w (match as a word) option.
3835
3836 (c) Refactored the way lines are read and buffered so as to have more
3837 than one at a time available.
3838
3839 (d) Implemented a pcregrep test script.
3840
3841 (e) Added the -M (multiline match) option. This allows patterns to match
3842 over several lines of the subject. The buffering ensures that at least
3843 8K, or the rest of the document (whichever is the shorter) is available
3844 for matching (and similarly the previous 8K for lookbehind assertions).
3845
3846 (f) Changed the --help output so that it now says
3847
3848 -w, --word-regex(p)
3849
3850 instead of two lines, one with "regex" and the other with "regexp"
3851 because that confused at least one person since the short forms are the
3852 same. (This required a bit of code, as the output is generated
3853 automatically from a table. It wasn't just a text change.)
3854
3855 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3856 option but starts with a hyphen. Could be a pattern or a path name
3857 starting with a hyphen, for instance.
3858
3859 (h) "-" can be given as a file name to represent stdin.
3860
3861 (i) When file names are being printed, "(standard input)" is used for
3862 the standard input, for compatibility with GNU grep. Previously
3863 "<stdin>" was used.
3864
3865 (j) The option --label=xxx can be used to supply a name to be used for
3866 stdin when file names are being printed. There is no short form.
3867
3868 (k) Re-factored the options decoding logic because we are going to add
3869 two more options that take data. Such options can now be given in four
3870 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3871
3872 (l) Added the -A, -B, and -C options for requesting that lines of context
3873 around matches be printed.
3874
3875 (m) Added the -L option to print the names of files that do not contain
3876 any matching lines, that is, the complement of -l.
3877
3878 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3879 continue to scan other files.
3880
3881 (o) The -s option was incorrectly implemented. For compatibility with other
3882 greps, it now suppresses the error message for a non-existent or non-
3883 accessible file (but not the return code). There is a new option called
3884 -q that suppresses the output of matching lines, which was what -s was
3885 previously doing.
3886
3887 (p) Added --include and --exclude options to specify files for inclusion
3888 and exclusion when recursing.
3889
3890 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3891 Hopefully, it now does.
3892
3893 12. Missing cast in pcre_study().
3894
3895 13. Added an "uninstall" target to the makefile.
3896
3897 14. Replaced "extern" in the function prototypes in Makefile.in with
3898 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3899 world, but is set differently for Windows.
3900
3901 15. Added a second compiling function called pcre_compile2(). The only
3902 difference is that it has an extra argument, which is a pointer to an
3903 integer error code. When there is a compile-time failure, this is set
3904 non-zero, in addition to the error test pointer being set to point to an
3905 error message. The new argument may be NULL if no error number is required
3906 (but then you may as well call pcre_compile(), which is now just a
3907 wrapper). This facility is provided because some applications need a
3908 numeric error indication, but it has also enabled me to tidy up the way
3909 compile-time errors are handled in the POSIX wrapper.
3910
3911 16. Added VPATH=.libs to the makefile; this should help when building with one
3912 prefix path and installing with another. (Or so I'm told by someone who
3913 knows more about this stuff than I do.)
3914
3915 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3916 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3917 match everything, including newlines. This is not POSIX-compatible, but
3918 somebody wanted the feature. From pcretest it can be activated by using
3919 both the P and the s flags.
3920
3921 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3922
3923 19. libpcre.pc was being incorrectly installed as executable.
3924
3925 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3926 it now also looks for '\r' so that it will work unmodified on Windows.
3927
3928 21. Added Google's contributed C++ wrapper to the distribution.
3929
3930 22. Added some untidy missing memory free() calls in pcretest, to keep
3931 Electric Fence happy when testing.
3932
3933
3934
3935 Version 5.0 13-Sep-04
3936 ---------------------
3937
3938 1. Internal change: literal characters are no longer packed up into items
3939 containing multiple characters in a single byte-string. Each character
3940 is now matched using a separate opcode. However, there may be more than one
3941 byte in the character in UTF-8 mode.
3942
3943 2. The pcre_callout_block structure has two new fields: pattern_position and
3944 next_item_length. These contain the offset in the pattern to the next match
3945 item, and its length, respectively.
3946
3947 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3948 insertion of callouts before each pattern item. Added the /C option to
3949 pcretest to make use of this.
3950
3951 4. On the advice of a Windows user, the lines
3952
3953 #if defined(_WIN32) || defined(WIN32)
3954 _setmode( _fileno( stdout ), 0x8000 );
3955 #endif /* defined(_WIN32) || defined(WIN32) */
3956
3957 have been added to the source of pcretest. This apparently does useful
3958 magic in relation to line terminators.
3959
3960 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3961 for the benefit of those environments where the "b" makes a difference.
3962
3963 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3964 to know about it. I have put a hack into configure.in that adds in code
3965 to set GCC=yes if CC=icc. This seems to end up at a point in the
3966 generated configure script that is early enough to affect the setting of
3967 compiler options, which is what is needed, but I have no means of testing
3968 whether it really works. (The user who reported this had patched the
3969 generated configure script, which of course I cannot do.)
3970
3971 LATER: After change 22 below (new libtool files), the configure script
3972 seems to know about icc (and also ecc). Therefore, I have commented out
3973 this hack in configure.in.
3974
3975 7. Added support for pkg-config (2 patches were sent in).
3976
3977 8. Negated POSIX character classes that used a combination of internal tables
3978 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3979 [[:^ascii]]. Typically, they would match almost any characters. The other
3980 POSIX classes were not broken in this way.
3981
3982 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3983 to find the match, as PCRE was deluded into thinking that the match had to
3984 start at the start point or following a newline. The same bug applied to
3985 patterns with negative forward assertions or any backward assertions
3986 preceding ".*" at the start, unless the pattern required a fixed first
3987 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3988
3989 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3990 starting at the last subject character, bytes beyond the end of the subject
3991 string were read.
3992
3993 11. Renamed the variable "class" as "classbits" to make life easier for C++
3994 users. (Previously there was a macro definition, but it apparently wasn't
3995 enough.)
3996
3997 12. Added the new field "tables" to the extra data so that tables can be passed
3998 in at exec time, or the internal tables can be re-selected. This allows
3999 a compiled regex to be saved and re-used at a later time by a different
4000 program that might have everything at different addresses.
4001
4002 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4003 -R library as well as a -L library.
4004
4005 14. The debugging options of pcretest (-d on the command line or D on a
4006 pattern) showed incorrect output for anything following an extended class
4007 that contained multibyte characters and which was followed by a quantifier.
4008
4009 15. Added optional support for general category Unicode character properties
4010 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4011 support. It adds about 90K to the size of the library. The meanings of the
4012 inbuilt class escapes such as \d and \s have NOT been changed.
4013
4014 16. Updated pcredemo.c to include calls to free() to release the memory for the
4015 compiled pattern.
4016
4017 17. The generated file chartables.c was being created in the source directory
4018 instead of in the building directory. This caused the build to fail if the
4019 source directory was different from the building directory, and was
4020 read-only.
4021
4022 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4023 file. No doubt somebody will tell me if they don't make sense... Also added
4024 Dan Mooney's comments about building on OpenVMS.
4025
4026 19. Added support for partial matching via the PCRE_PARTIAL option for
4027 pcre_exec() and the \P data escape in pcretest.
4028
4029 20. Extended pcretest with 3 new pattern features:
4030
4031 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4032 write the compiled pattern to the file whose name is "rest-of-line".
4033 This is a straight binary dump of the data, with the saved pointer to
4034 the character tables forced to be NULL. The study data, if any, is
4035 written too. After writing, pcretest reads a new pattern.
4036
4037 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4038 compiled pattern from the given file. There must not be any
4039 occurrences of "<" in the file name (pretty unlikely); if there are,
4040 pcretest will instead treat the initial "<" as a pattern delimiter.
4041 After reading in the pattern, pcretest goes on to read data lines as
4042 usual.
4043
4044 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4045 and 16-bit fields in a compiled pattern, to simulate a pattern that
4046 was compiled on a host of opposite endianness.
4047
4048 21. The pcre-exec() function can now cope with patterns that were compiled on
4049 hosts of opposite endianness, with this restriction:
4050
4051 As for any compiled expression that is saved and used later, the tables
4052 pointer field cannot be preserved; the extra_data field in the arguments
4053 to pcre_exec() should be used to pass in a tables address if a value
4054 other than the default internal tables were used at compile time.
4055
4056 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4057 now diagnosed as an error. Previously, most of the time, a negative number
4058 would have been treated as zero, but if in addition "ovector" was passed as
4059 NULL, a crash could occur.
4060
4061 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4062 new versions from the libtool 1.5 distribution (the last one is a copy of
4063 a file called libtool.m4). This seems to have fixed the need to patch
4064 "configure" to support Darwin 1.3 (which I used to do). However, I still
4065 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4066 workstation).
4067
4068 24. Changed the PCRE licence to be the more standard "BSD" licence.
4069
4070
4071 Version 4.5 01-Dec-03
4072 ---------------------
4073
4074 1. There has been some re-arrangement of the code for the match() function so
4075 that it can be compiled in a version that does not call itself recursively.
4076 Instead, it keeps those local variables that need separate instances for
4077 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4078 needs to "recurse". Keeping track of where control must go is done by means
4079 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4080 hide most of the details from the main code, and operates only if
4081 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4082 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4083 operating.
4084
4085 To make it easier for callers to provide specially tailored get/free
4086 functions for this usage, two new functions, pcre_stack_malloc, and
4087 pcre_stack_free, are used. They are always called in strict stacking order,
4088 and the size of block requested is always the same.
4089
4090 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4091 PCRE has been compiled to use the stack or the heap for recursion. The
4092 -C option of pcretest uses this to show which version is compiled.
4093
4094 A new data escape \S, is added to pcretest; it causes the amounts of store
4095 obtained and freed by both kinds of malloc/free at match time to be added
4096 to the output.
4097
4098 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4099 what's available on my current Linux desktop machine.
4100
4101 3. When matching a UTF-8 string, the test for a valid string at the start has
4102 been extended. If start_offset is not zero, PCRE now checks that it points
4103 to a byte that is the start of a UTF-8 character. If not, it returns
4104 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4105 this is necessary because there may be backward assertions in the pattern.
4106 When matching the same subject several times, it may save resources to use
4107 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4108
4109 4. The code for checking the validity of UTF-8 strings has been tightened so
4110 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4111 containing "overlong sequences".
4112
4113 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4114 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4115 should have been "&", but it just so happened that all the cases this let
4116 through by mistake were picked up later in the function.
4117
4118 6. I had used a variable called "isblank" - this is a C99 function, causing
4119 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4120
4121 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4122 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4123 so the version is shown; (c) stop "make test" echoing "./RunTest".
4124
4125 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4126
4127 9. The prototype for memmove() for systems that don't have it was using
4128 size_t, but the inclusion of the header that defines size_t was later. I've
4129 moved the #includes for the C headers earlier to avoid this.
4130
4131 10. Added some adjustments to the code to make it easier to compiler on certain
4132 special systems:
4133
4134 (a) Some "const" qualifiers were missing.
4135 (b) Added the macro EXPORT before all exported functions; by default this
4136 is defined to be empty.
4137 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4138 that it reads its output file name as an argument instead of writing
4139 to the standard output and assuming this can be redirected.
4140
4141 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4142 class containing characters with values greater than 255, PCRE compilation
4143 went into a loop.
4144
4145 12. A recursive reference to a subpattern that was within another subpattern
4146 that had a minimum quantifier of zero caused PCRE to crash. For example,
4147 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4148 recursion. If the recursively-called subpattern itself had a zero repeat,
4149 that was OK.
4150
4151 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4152 buffer into which it was copied (for escape processing) was still set at
4153 1024, so long lines caused crashes.
4154
4155 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4156 "internal error: code overflow...". This applied to any character class
4157 that was followed by a possessive quantifier.
4158
4159 15. Modified the Makefile to add libpcre.la as a prerequisite for
4160 libpcreposix.la because I was told this is needed for a parallel build to
4161 work.
4162
4163 16. If a pattern that contained .* following optional items at the start was
4164 studied, the wrong optimizing data was generated, leading to matching
4165 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4166 matching string must start with a or b or c. The correct conclusion for
4167 this pattern is that a match can start with any character.
4168
4169
4170 Version 4.4 13-Aug-03
4171 ---------------------
4172
4173 1. In UTF-8 mode, a character class containing characters with values between
4174 127 and 255 was not handled correctly if the compiled pattern was studied.
4175 In fixing this, I have also improved the studying algorithm for such
4176 classes (slightly).
4177
4178 2. Three internal functions had redundant arguments passed to them. Removal
4179 might give a very teeny performance improvement.
4180
4181 3. Documentation bug: the value of the capture_top field in a callout is *one
4182 more than* the number of the hightest numbered captured substring.
4183
4184 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4185 in incorrectly linking with a previously installed version. They now link
4186 explicitly with libpcre.la.
4187
4188 5. configure.in no longer needs to recognize Cygwin specially.
4189
4190 6. A problem in pcre.in for Windows platforms is fixed.
4191
4192 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4193 pcretest, it used to include the size of the study block as part of its
4194 output. Unfortunately, the structure contains a field that has a different
4195 size on different hardware architectures. This meant that the tests that
4196 showed this size failed. As the block is currently always of a fixed size,
4197 this information isn't actually particularly useful in pcretest output, so
4198 I have just removed it.
4199
4200 8. Three pre-processor statements accidentally did not start in column 1.
4201 Sadly, there are *still* compilers around that complain, even though
4202 standard C has not required this for well over a decade. Sigh.
4203
4204 9. In pcretest, the code for checking callouts passed small integers in the
4205 callout_data field, which is a void * field. However, some picky compilers
4206 complained about the casts involved for this on 64-bit systems. Now
4207 pcretest passes the address of the small integer instead, which should get
4208 rid of the warnings.
4209
4210 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4211 both compile and run time, and gives an error if an invalid UTF-8 sequence
4212 is found. There is a option for disabling this check in cases where the
4213 string is known to be correct and/or the maximum performance is wanted.
4214
4215 11. In response to a bug report, I changed one line in Makefile.in from
4216
4217 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4218 to
4219 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4220
4221 to look similar to other lines, but I have no way of telling whether this
4222 is the right thing to do, as I do not use Windows. No doubt I'll get told
4223 if it's wrong...
4224
4225
4226 Version 4.3 21-May-03
4227 ---------------------
4228
4229 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4230 Makefile.
4231
4232 2. Some refactoring to improve the quality of the code:
4233
4234 (i) The utf8_table... variables are now declared "const".
4235
4236 (ii) The code for \cx, which used the "case flipping" table to upper case
4237 lower case letters, now just substracts 32. This is ASCII-specific,
4238 but the whole concept of \cx is ASCII-specific, so it seems
4239 reasonable.
4240
4241 (iii) PCRE was using its character types table to recognize decimal and
4242 hexadecimal digits in the pattern. This is silly, because it handles
4243 only 0-9, a-f, and A-F, but the character types table is locale-
4244 specific, which means strange things might happen. A private
4245 table is now used for this - though it costs 256 bytes, a table is
4246 much faster than multiple explicit tests. Of course, the standard
4247 character types table is still used for matching digits in subject
4248 strings against \d.
4249
4250 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4251 ending in _t are). So I've renamed it as ESC_tee.
4252
4253 3. The first argument for regexec() in the POSIX wrapper should have been
4254 defined as "const".
4255
4256 4. Changed pcretest to use malloc() for its buffers so that they can be
4257 Electric Fenced for debugging.
4258
4259 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4260 to read one or more bytes before the start of the subject string. Often this
4261 had no effect on PCRE's behaviour, but in some circumstances it could
4262 provoke a segmentation fault.
4263
4264 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4265 to try to read one or more bytes before the start of the subject string.
4266
4267 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4268 UTF-8 support could misbehave in various ways if the subject string
4269 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4270 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4271 back over UTF-8 characters.)
4272
4273
4274 Version 4.2 14-Apr-03
4275 ---------------------
4276
4277 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4278
4279 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4280 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4281 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4282 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4283 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4284 and BUILD_EXEEXT
4285 Note: automatic setting of the BUILD variables is not yet working
4286 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4287 compile-time but not at link-time
4288 [LINK]: use for linking executables only
4289 make different versions for Windows and non-Windows
4290 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4291 libraries
4292 [LINK_FOR_BUILD]: new variable
4293 [OBJEXT]: use throughout
4294 [EXEEXT]: use throughout
4295 <winshared>: new target
4296 <wininstall>: new target
4297 <dftables.o>: use native compiler
4298 <dftables>: use native linker
4299 <install>: handle Windows platform correctly
4300 <clean>: ditto
4301 <check>: ditto
4302 copy DLL to top builddir before testing
4303
4304 As part of these changes, -no-undefined was removed again. This was reported
4305 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4306 in any case.
4307
4308 3. Some tidies to get rid of compiler warnings:
4309
4310 . In the match_data structure, match_limit was an unsigned long int, whereas
4311 match_call_count was an int. I've made them both unsigned long ints.
4312
4313 . In pcretest the fact that a const uschar * doesn't automatically cast to
4314 a void * provoked a warning.
4315
4316 . Turning on some more compiler warnings threw up some "shadow" variables
4317 and a few more missing casts.
4318
4319 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4320 option, a class that contained a single character with a value between 128
4321 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4322
4323 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4324 option, a class that contained several characters, but with at least one
4325 whose value was between 128 and 255 caused PCRE to crash.
4326
4327
4328 Version 4.1 12-Mar-03
4329 ---------------------
4330
4331 1. Compiling with gcc -pedantic found a couple of places where casts were
4332 needed, and a string in dftables.c that was longer than standard compilers are
4333 required to support.
4334
4335 2. Compiling with Sun's compiler found a few more places where the code could
4336 be tidied up in order to avoid warnings.
4337
4338 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4339 first of these names is deprecated in the latest Autoconf in favour of the name
4340 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4341 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4342 analogy I have changed it to CFLAGS_FOR_BUILD.
4343
4344 4. Added -no-undefined to the linking command in the Makefile, because this is
4345 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4346 linking step for the pcreposix library.
4347
4348 5. PCRE was failing to diagnose the case of two named groups with the same
4349 name.
4350
4351 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4352 literal character that is needed in the subject for a match, and scans along to
4353 ensure that it is present before embarking on the full matching process. This
4354 saves time in cases of nested unlimited repeats that are never going to match.
4355 Problem: the scan can take a lot of time if the subject is very long (e.g.
4356 megabytes), thus penalizing straightforward matches. It is now done only if the
4357 amount of subject to be scanned is less than 1000 bytes.
4358
4359 7. A lesser problem with the same optimization is that it was recording the
4360 first character of an anchored pattern as "needed", thus provoking a search
4361 right along the subject, even when the first match of the pattern was going to
4362 fail. The "needed" character is now not set for anchored patterns, unless it
4363 follows something in the pattern that is of non-fixed length. Thus, it still
4364 fulfils its original purpose of finding quick non-matches in cases of nested
4365 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4366
4367
4368 Version 4.0 17-Feb-03
4369 ---------------------
4370
4371 1. If a comment in an extended regex that started immediately after a meta-item
4372 extended to the end of string, PCRE compiled incorrect data. This could lead to
4373 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4374
4375 2. Moved to autoconf 2.53 and libtool 1.4.2.
4376
4377 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4378 the special perltest8 script is no longer needed - all the tests can be run
4379 from a single perltest script.
4380
4381 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4382 by \s. It has now been removed in PCRE. This means it isn't recognized as
4383 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4384 class [:space:] *does* include VT, thereby creating a mess.
4385
4386 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4387 space and tab.
4388
4389 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4390 its new features into the main test script, reducing the number of scripts.
4391
4392 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4393 were backward compatible, and made the (?i) apply to the whole pattern, as if
4394 /i were given. Now it behaves more logically, and applies the option setting
4395 only to what follows. PCRE has been changed to follow suit. However, if it
4396 finds options settings right at the start of the pattern, it extracts them into
4397 the global options, as before. Thus, they show up in the info data.
4398
4399 8. Added support for the \Q...\E escape sequence. Characters in between are
4400 treated as literals. This is slightly different from Perl in that $ and @ are
4401 also handled as literals inside the quotes. In Perl, they will cause variable
4402 interpolation. Note the following examples:
4403
4404 Pattern PCRE matches Perl matches
4405
4406 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4407 \Qabc\$xyz\E abc\$xyz abc\$xyz
4408 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4409
4410 For compatibility with Perl, \Q...\E sequences are recognized inside character
4411 classes as well as outside them.
4412
4413 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4414 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4415 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4416 signed/unsigned warnings.
4417
4418 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4419 option for pcretest, so I've replaced it by a simple function that does just
4420 that job.
4421
4422 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4423 "pcregrep -".
4424
4425 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4426 Java package. This provides some syntactic sugar for simple cases of what my
4427 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4428 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4429 item, you can use this simplified notation. Note that only makes sense with
4430 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4431 greediness, whatever the setting of the PCRE_UNGREEDY option.
4432
4433 13. A change of greediness default within a pattern was not taking effect at
4434 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4435 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4436 was abstracted outside.
4437
4438 14. PCRE now supports the \G assertion. It is true when the current matching
4439 position is at the start point of the match. This differs from \A when the
4440 starting offset is non-zero. Used with the /g option of pcretest (or similar
4441 code), it works in the same way as it does for Perl's /g option. If all
4442 alternatives of a regex begin with \G, the expression is anchored to the start
4443 match position, and the "anchored" flag is set in the compiled expression.
4444
4445 15. Some bugs concerning the handling of certain option changes within patterns
4446 have been fixed. These applied to options other than (?ims). For example,
4447 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4448 been the other way round. Some of this was related to change 7 above.
4449
4450 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4451 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4452 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4453 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4454
4455 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4456 mode. Unlike ".", it always matches newline, whatever the setting of
4457 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4458 assertions. Perl allows it, but it doesn't (in general) work because it can't
4459 calculate the length of the lookbehind. At least, that's the case for Perl
4460 5.8.0 - I've been told they are going to document that it doesn't work in
4461 future.
4462
4463 18. Added an error diagnosis for escapes that PCRE does not support: these are
4464 \L, \l, \N, \P, \p, \U, \u, and \X.
4465
4466 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4467 reading past the end of the pattern in cases such as /[abcd/.
4468
4469 20. PCRE was getting more memory than necessary for patterns with classes that
4470 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4471
4472 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4473 compiling PCRE for use with Virtual Pascal.
4474
4475 22. Small fix to the Makefile to make it work properly if the build is done
4476 outside the source tree.
4477
4478 23. Added a new extension: a condition to go with recursion. If a conditional
4479 subpattern starts with (?(R) the "true" branch is used if recursion has
4480 happened, whereas the "false" branch is used only at the top level.
4481
4482 24. When there was a very long string of literal characters (over 255 bytes
4483 without UTF support, over 250 bytes with UTF support), the computation of how
4484 much memory was required could be incorrect, leading to segfaults or other
4485 strange effects.
4486
4487 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4488 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4489 there was a subsequent back reference to those brackets. This meant that, for
4490 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4491 possible to check for precisely this case. All we can do is abandon the
4492 optimization if .* occurs inside capturing brackets when there are any back
4493 references whatsoever. (See below for a better fix that came later.)
4494
4495 26. The handling of the optimization for finding the first character of a
4496 non-anchored pattern, and for finding a character that is required later in the
4497 match were failing in some cases. This didn't break the matching; it just
4498 failed to optimize when it could. The way this is done has been re-implemented.
4499
4500 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4501
4502 28. Added a new feature that provides some of the functionality that Perl
4503 provides with (?{...}). The facility is termed a "callout". The way it is done
4504 in PCRE is for the caller to provide an optional function, by setting
4505 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4506 global variable. By default it is unset, which disables all calling out. To get
4507 the function called, the regex must include (?C) at appropriate points. This
4508 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4509 This provides a means of identifying different callout points. When PCRE
4510 reaches such a point in the regex, if pcre_callout has been set, the external
4511 function is called. It is provided with data in a structure called
4512 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4513 matching continues; if it returns a non-zero value, the match at the current
4514 point fails. However, backtracking will occur if possible. [This was changed
4515 later and other features added - see item 49 below.]
4516
4517 29. pcretest is upgraded to test the callout functionality. It provides a
4518 callout function that displays information. By default, it shows the start of
4519 the match and the current position in the text. There are some new data escapes
4520 to vary what happens:
4521
4522 \C+ in addition, show current contents of captured substrings
4523 \C- do not supply a callout function
4524 \C!n return 1 when callout number n is reached
4525 \C!n!m return 1 when callout number n is reached for the mth time
4526
4527 30. If pcregrep was called with the -l option and just a single file name, it
4528 output "<stdin>" if a match was found, instead of the file name.
4529
4530 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4531 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4532 pcre_exec(). This saves a malloc/free per call. The default value of
4533 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4534 when configuring.
4535
4536 32. The default maximum size of a compiled pattern is 64K. There have been a
4537 few cases of people hitting this limit. The code now uses macros to handle the
4538 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4539 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4540 configuring. Tests 2 and 5 work only with 2-byte links because they output
4541 debugging information about compiled patterns.
4542
4543 33. Internal code re-arrangements:
4544
4545 (a) Moved the debugging function for printing out a compiled regex into
4546 its own source file (printint.c) and used #include to pull it into
4547 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4548 separate copies.
4549
4550 (b) Defined the list of op-code names for debugging as a macro in
4551 internal.h so that it is next to the definition of the opcodes.
4552
4553 (c) Defined a table of op-code lengths for simpler skipping along compiled
4554 code. This is again a macro in internal.h so that it is next to the
4555 definition of the opcodes.
4556
4557 34. Added support for recursive calls to individual subpatterns, along the
4558 lines of Robin Houston's patch (but implemented somewhat differently).
4559
4560 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4561 allow it to read and process whole directories in Win32. This code was
4562 contributed by Lionel Fourquaux; it has not been tested by me.
4563
4564 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4565 used to name a group. Names consist of alphanumerics and underscores, and must
4566 be unique. Back references use the syntax (?P=name) and recursive calls use
4567 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4568 numbers. The function pcre_fullinfo() can be used after compilation to extract
4569 a name/number map. There are three relevant calls:
4570
4571 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4572 PCRE_INFO_NAMECOUNT yields the number of entries
4573 PCRE_INFO_NAMETABLE yields a pointer to the map.
4574
4575 The map is a vector of fixed-size entries. The size of each entry depends on
4576 the length of the longest name used. The first two bytes of each entry are the
4577 group number, most significant byte first. There follows the corresponding
4578 name, zero terminated. The names are in alphabetical order.
4579
4580 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4581 case instead of 255. Making it the same both with and without UTF-8 support
4582 means that the same test output works with both.
4583
4584 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4585 calling malloc() with a zero argument.
4586
4587 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4588 optimization. I've improved things by keeping a bitmap of backreferences with
4589 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4590 fact referenced, the optimization can be applied. It is unlikely that a
4591 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4592 the match to follow \n) will appear inside brackets with a number greater than
4593 31, but if it does, any back reference > 31 suppresses the optimization.
4594
4595 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4596 of disabling numbered capturing parentheses. Any opening parenthesis that is
4597 not followed by ? behaves as if it were followed by ?: but named parentheses
4598 can still be used for capturing (and they will acquire numbers in the usual
4599 way).
4600
4601 41. Redesigned the return codes from the match() function into yes/no/error so
4602 that errors can be passed back from deep inside the nested calls. A malloc
4603 failure while inside a recursive subpattern call now causes the
4604 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4605
4606 42. It is now possible to set a limit on the number of times the match()
4607 function is called in a call to pcre_exec(). This facility makes it possible to
4608 limit the amount of recursion and backtracking, though not in a directly
4609 obvious way, because the match() function is used in a number of different
4610 circumstances. The count starts from zero for each position in the subject
4611 string (for non-anchored patterns). The default limit is, for compatibility, a
4612 large number, namely 10 000 000. You can change this in two ways:
4613
4614 (a) When configuring PCRE before making, you can use --with-match-limit=n
4615 to set a default value for the compiled library.
4616
4617 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4618 a different value is set. See 45 below.
4619
4620 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4621
4622 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4623 of things that can be changed at compile time. The first argument specifies
4624 what is wanted and the second points to where the information is to be placed.
4625 The current list of available information is:
4626
4627 PCRE_CONFIG_UTF8
4628
4629 The output is an integer that is set to one if UTF-8 support is available;
4630 otherwise it is set to zero.
4631
4632 PCRE_CONFIG_NEWLINE
4633
4634 The output is an integer that it set to the value of the code that is used for
4635 newline. It is either LF (10) or CR (13).
4636
4637 PCRE_CONFIG_LINK_SIZE
4638
4639 The output is an integer that contains the number of bytes used for internal
4640 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4641
4642 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4643
4644 The output is an integer that contains the threshold above which the POSIX
4645 interface uses malloc() for output vectors. See item 31 above.
4646
4647 PCRE_CONFIG_MATCH_LIMIT
4648
4649 The output is an unsigned integer that contains the default limit of the number
4650 of match() calls in a pcre_exec() execution. See 42 above.
4651
4652 44. pcretest has been upgraded by the addition of the -C option. This causes it
4653 to extract all the available output from the new pcre_config() function, and to
4654 output it. The program then exits immediately.
4655
4656 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4657 order to support additional features. One way would have been to define
4658 pcre_exec2() (for example) with extra arguments, but this would not have been
4659 extensible, and would also have required all calls to the original function to
4660 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4661 is used for passing in "extra" data from pcre_study().
4662
4663 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4664 contains the following fields:
4665
4666 flags a bitmap indicating which of the following fields are set
4667 study_data opaque data from pcre_study()
4668 match_limit a way of specifying a limit on match() calls for a specific
4669 call to pcre_exec()
4670 callout_data data for callouts (see 49 below)
4671
4672 The flag bits are also defined in pcre.h, and are
4673
4674 PCRE_EXTRA_STUDY_DATA
4675 PCRE_EXTRA_MATCH_LIMIT
4676 PCRE_EXTRA_CALLOUT_DATA
4677
4678 The pcre_study() function now returns one of these new pcre_extra blocks, with
4679 the actual study data pointed to by the study_data field, and the
4680 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4681 before. That is, this change is entirely upwards-compatible and requires no
4682 change to existing code.
4683
4684 If you want to pass in additional data to pcre_exec(), you can either place it
4685 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4686 block.
4687
4688 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4689 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4690 times with different match limits, until it finds the minimum value needed for
4691 pcre_exec() to complete. The value is then output. This can be instructive; for
4692 most simple matches the number is quite small, but for pathological cases it
4693 gets very large very quickly.
4694
4695 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4696 returns the size of the data block pointed to by the study_data field in a
4697 pcre_extra block, that is, the value that was passed as the argument to
4698 pcre_malloc() when PCRE was getting memory in which to place the information
4699 created by pcre_study(). The fourth argument should point to a size_t variable.
4700 pcretest has been extended so that this information is shown after a successful
4701 pcre_study() call when information about the compiled regex is being displayed.
4702
4703 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4704 because what follows is always an absolute path. (Later: it turns out that this
4705 is more than cosmetic for MinGW, because it doesn't like empty path
4706 components.)
4707
4708 49. Some changes have been made to the callout feature (see 28 above):
4709
4710 (i) A callout function now has three choices for what it returns:
4711
4712 0 => success, carry on matching
4713 > 0 => failure at this point, but backtrack if possible
4714 < 0 => serious error, return this value from pcre_exec()
4715
4716 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4717 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4718 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4719 use by callout functions. It will never be used by PCRE itself.
4720
4721 (ii) The pcre_extra structure (see 45 above) has a void * field called
4722 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4723 pcre_callout_block structure has a field of the same name. The contents of
4724 the field passed in the pcre_extra structure are passed to the callout
4725 function in the corresponding field in the callout block. This makes it
4726 easier to use the same callout-containing regex from multiple threads. For
4727 testing, the pcretest program has a new data escape
4728
4729 \C*n pass the number n (may be negative) as callout_data
4730
4731 If the callout function in pcretest receives a non-zero value as
4732 callout_data, it returns that value.
4733
4734 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4735 there were some redundant $(CFLAGS) in commands that are now specified as
4736 $(LINK), which already includes $(CFLAGS).
4737
4738 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4739 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4740 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4741 one-byte characters throughout. Note that case-insensitive matching applies
4742 only to characters whose values are less than 256. PCRE doesn't support the
4743 notion of cases for higher-valued characters.
4744
4745 (i) A character class whose characters are all within 0-255 is handled as
4746 a bit map, and the map is inverted for negative classes. Previously, a
4747 character > 255 always failed to match such a class; however it should
4748 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4749
4750 (ii) A negated character class with a single character < 255 is coded as
4751 "not this character" (OP_NOT). This wasn't working properly when the test
4752 character was multibyte, either singly or repeated.
4753
4754 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4755 mode, for example: \x{100}{2,3}.
4756
4757 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4758 singly or repeated) now correctly test multibyte characters. However,
4759 PCRE doesn't recognize any characters with values greater than 255 as
4760 digits, spaces, or word characters. Such characters always match \D, \S,
4761 and \W, and never match \d, \s, or \w.
4762
4763 (v) Classes may now contain characters and character ranges with values
4764 greater than 255. For example: [ab\x{100}-\x{400}].
4765
4766 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4767 PCRE in UTF-8 mode.
4768
4769 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4770 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4771 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4772 value.)
4773
4774 53. The single man page has become too large. I have therefore split it up into
4775 a number of separate man pages. These also give rise to individual HTML pages;
4776 these are now put in a separate directory, and there is an index.html page that
4777 lists them all. Some hyperlinking between the pages has been installed.
4778
4779 54. Added convenience functions for handling named capturing parentheses.
4780
4781 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4782 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4783 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4784 are faulted.
4785
4786 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4787 calling configure. These values are used when compiling the dftables.c program
4788 which is run to generate the source of the default character tables. They
4789 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4790 you will need to set these values.
4791
4792 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4793
4794
4795 Version 3.9 02-Jan-02
4796 ---------------------
4797
4798 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4799
4800 2. If --disable-static was given, the building process failed when trying to
4801 build pcretest and pcregrep. (For some reason it was using libtool to compile
4802 them, which is not right, as they aren't part of the library.)
4803
4804
4805 Version 3.8 18-Dec-01
4806 ---------------------
4807
4808 1. The experimental UTF-8 code was completely screwed up. It was packing the
4809 bytes in the wrong order. How dumb can you get?
4810
4811
4812 Version 3.7 29-Oct-01
4813 ---------------------
4814
4815 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4816 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4817 this didn't happen under Solaris 8, where I normally test things.
4818
4819 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4820 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4821 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4822 this fix an hour or so after the initial 3.7 release.)
4823
4824
4825 Version 3.6 23-Oct-01
4826 ---------------------
4827
4828 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4829 offsets passed as NULL with zero offset count.
4830
4831 2. The config.guess and config.sub files had not been updated when I moved to
4832 the latest autoconf.
4833
4834
4835 Version 3.5 15-Aug-01
4836 ---------------------
4837
4838 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4839 had been forgotten.
4840
4841 2. By using declared but undefined structures, we can avoid using "void"
4842 definitions in pcre.h while keeping the internal definitions of the structures
4843 private.
4844
4845 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4846 user point of view, this means that both static and shared libraries are built
4847 by default, but this can be individually controlled. More of the work of
4848 handling this static/shared cases is now inside libtool instead of PCRE's make
4849 file.
4850
4851 4. The pcretest utility is now installed along with pcregrep because it is
4852 useful for users (to test regexs) and by doing this, it automatically gets
4853 relinked by libtool. The documentation has been turned into a man page, so
4854 there are now .1, .txt, and .html versions in /doc.
4855
4856 5. Upgrades to pcregrep:
4857 (i) Added long-form option names like gnu grep.
4858 (ii) Added --help to list all options with an explanatory phrase.
4859 (iii) Added -r, --recursive to recurse into sub-directories.
4860 (iv) Added -f, --file to read patterns from a file.
4861
4862 6. pcre_exec() was referring to its "code" argument before testing that
4863 argument for NULL (and giving an error if it was NULL).
4864
4865 7. Upgraded Makefile.in to allow for compiling in a different directory from
4866 the source directory.
4867
4868 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4869 options bits, the pointer it was passed was to an int instead of to an unsigned
4870 long int. This mattered only on 64-bit systems.
4871
4872 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4873 generated) instead of pcre.in, which it its source. Also made the same change
4874 in several of the .c files.
4875
4876 10. A new release of gcc defines printf() as a macro, which broke pcretest
4877 because it had an ifdef in the middle of a string argument for printf(). Fixed
4878 by using separate calls to printf().
4879
4880 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4881 script, to force use of CR or LF instead of \n in the source. On non-Unix
4882 systems, the value can be set in config.h.
4883
4884 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4885 absolute limit. Changed the text of the error message to make this clear, and
4886 likewise updated the man page.
4887
4888 13. The limit of 99 on the number of capturing subpatterns has been removed.
4889 The new limit is 65535, which I hope will not be a "real" limit.
4890
4891
4892 Version 3.4 22-Aug-00
4893 ---------------------
4894
4895 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4896
4897 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4898
4899
4900 Version 3.3 01-Aug-00
4901 ---------------------
4902
4903 1. If an octal character was given, but the value was greater than \377, it
4904 was not getting masked to the least significant bits, as documented. This could
4905 lead to crashes in some systems.
4906
4907 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4908 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4909
4910 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4911 These just pass their arguments on to (pcre_free)(), but they are provided
4912 because some uses of PCRE bind it to non-C systems that can call its functions,
4913 but cannot call free() or pcre_free() directly.
4914
4915 4. Add "make test" as a synonym for "make check". Corrected some comments in
4916 the Makefile.
4917
4918 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4919 Makefile.
4920
4921 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4922 command called pgrep for grepping around the active processes.
4923
4924 7. Added the beginnings of support for UTF-8 character strings.
4925
4926 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4927 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4928 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4929 out for the ar command.)
4930
4931
4932 Version 3.2 12-May-00
4933 ---------------------
4934
4935 This is purely a bug fixing release.
4936
4937 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4938 of ZA. This was just one example of several cases that could provoke this bug,
4939 which was introduced by change 9 of version 2.00. The code for breaking
4940 infinite loops after an iteration that matches an empty string was't working
4941 correctly.
4942
4943 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4944 when matched against abbab (for example). After matching an empty string, it
4945 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4946 caused it to match further down the string than it should.
4947
4948 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4949 was there because it doesn't seem to be needed, and it causes trouble on some
4950 systems, as it is not a Standard C header. It has been removed.
4951
4952 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4953 were reported on the Macintosh. The changes were from
4954
4955 while ((c = *(++ptr)) != 0 && c != '\n');
4956 to
4957 while ((c = *(++ptr)) != 0 && c != '\n') ;
4958
4959 Totally extraordinary, but if that's what it takes...
4960
4961 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4962 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4963 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4964 assumes the way PCRE uses memmove() (always moving upwards).
4965
4966 6. PCRE is being used in one environment where strchr() is not available. There
4967 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4968 faster code anyway.
4969
4970
4971 Version 3.1 09-Feb-00
4972 ---------------------
4973
4974 The only change in this release is the fixing of some bugs in Makefile.in for
4975 the "install" target:
4976
4977 (1) It was failing to install pcreposix.h.
4978
4979 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4980
4981
4982 Version 3.0 01-Feb-00
4983 ---------------------
4984
4985 1. Add support for the /+ modifier to perltest (to output $` like it does in
4986 pcretest).
4987
4988 2. Add support for the /g modifier to perltest.
4989
4990 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4991 matches null strings.
4992
4993 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4994 pattern. Perl treats empty patterns specially - it reuses the most recent
4995 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4996 effect.
4997
4998 5. The POSIX interface was broken in that it was just handing over the POSIX
4999 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5000 required a bigger vector, with some working space on the end. This means that
5001 the POSIX wrapper now has to get and free some memory, and copy the results.
5002
5003 6. Added some simple autoconf support, placing the test data and the
5004 documentation in separate directories, re-organizing some of the
5005 information files, and making it build pcre-config (a GNU standard). Also added
5006 libtool support for building PCRE as a shared library, which is now the
5007 default.
5008
5009 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5010 09 are not valid octal constants. Single digits will be used for minor values
5011 less than 10.
5012
5013 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5014 existing programs that set these in the POSIX interface can use PCRE without
5015 modification.
5016
5017 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5018 return all that pcre_info() returns, plus additional data. The pcre_info()
5019 function is retained for compatibility, but is considered to be obsolete.
5020
5021 10. Added experimental recursion feature (?R) to handle one common case that
5022 Perl 5.6 will be able to do with (?p{...}).
5023
5024 11. Added support for POSIX character classes like [:alpha:], which Perl is
5025 adopting.
5026
5027
5028 Version 2.08 31-Aug-99
5029 ----------------------
5030
5031 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5032 trying to match at the startoffset position, but instead was moving forward to
5033 the next newline as if a previous match had failed.
5034
5035 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5036 and could get into a loop if a null string was matched other than at the start
5037 of the subject.
5038
5039 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5040 be distinguished at compile time, and for completeness also added PCRE_DATE.
5041
5042 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5043 in GnuWin32 environments.
5044
5045
5046 Version 2.07 29-Jul-99
5047 ----------------------
5048
5049 1. The documentation is now supplied in plain text form and HTML as well as in
5050 the form of man page sources.
5051
5052 2. C++ compilers don't like assigning (void *) values to other pointer types.
5053 In particular this affects malloc(). Although there is no problem in Standard
5054 C, I've put in casts to keep C++ compilers happy.
5055
5056 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5057 should be (const char *).
5058
5059 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5060 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5061 However, I haven't made this a standard facility. The documentation doesn't
5062 mention it, and the Makefile doesn't support it.
5063
5064 5. The Makefile now contains an "install" target, with editable destinations at
5065 the top of the file. The pcretest program is not installed.
5066
5067 6. pgrep -V now gives the PCRE version number and date.
5068
5069 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5070 causing the entire string to be ignored, instead of just the last character.
5071
5072 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5073 non-matching string, it can take a very, very long time, even for strings of
5074 quite modest length, because of the nested recursion. PCRE now does better in
5075 some of these cases. It does this by remembering the last required literal
5076 character in the pattern, and pre-searching the subject to ensure it is present
5077 before running the real match. In other words, it applies a heuristic to detect
5078 some types of certain failure quickly, and in the above example, if presented
5079 with a string that has no trailing " it gives "no match" very quickly.
5080
5081 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5082 other alternatives are tried instead.
5083
5084
5085 Version 2.06 09-Jun-99
5086 ----------------------
5087
5088 1. Change pcretest's output for amount of store used to show just the code
5089 space, because the remainder (the data block) varies in size between 32-bit and
5090 64-bit systems.
5091
5092 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5093 start matching at. This allows lookbehinds to work when searching for multiple
5094 occurrences in a string.
5095
5096 3. Added additional options to pcretest for testing multiple occurrences:
5097
5098 /+ outputs the rest of the string that follows a match
5099 /g loops for multiple occurrences, using the new startoffset argument
5100 /G loops for multiple occurrences by passing an incremented pointer
5101
5102 4. PCRE wasn't doing the "first character" optimization for patterns starting
5103 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5104 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5105 the letter 'x'. On long subject strings, this gives a significant speed-up.
5106
5107
5108 Version 2.05 21-Apr-99
5109 ----------------------
5110
5111 1. Changed the type of magic_number from int to long int so that it works
5112 properly on 16-bit systems.
5113
5114 2. Fixed a bug which caused patterns starting with .* not to work correctly
5115 when the subject string contained newline characters. PCRE was assuming
5116 anchoring for such patterns in all cases, which is not correct because .* will
5117 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5118 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5119 must be retried after every newline in the subject.
5120
5121
5122 Version 2.04 18-Feb-99
5123 ----------------------
5124
5125 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5126 computation of the store needed to hold the pattern was incorrect (too large).
5127 If such patterns were nested a few deep, this could multiply and become a real
5128 problem.
5129
5130 2. Added /M option to pcretest to show the memory requirement of a specific
5131 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5132
5133 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5134 compiled in such a way that the backtracking after subsequent failure was
5135 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5136 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5137
5138
5139 Version 2.03 02-Feb-99
5140 ----------------------
5141
5142 1. Fixed typo and small mistake in man page.
5143
5144 2. Added 4th condition (GPL supersedes if conflict) and created separate
5145 LICENCE file containing the conditions.
5146
5147 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5148 Perl, that is the internal \ allows the delimiter to be included in the
5149 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5150 the final delimiter, add \ to the end of the pattern (to test the error).
5151
5152 4. Added the convenience functions for extracting substrings after a successful
5153 match. Updated pcretest to make it able to test these functions.
5154
5155
5156 Version 2.02 14-Jan-99
5157 ----------------------
5158
5159 1. Initialized the working variables associated with each extraction so that
5160 their saving and restoring doesn't refer to uninitialized store.
5161
5162 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5163 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5164 fix the problem.
5165
5166 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5167 calls, and wasn't printing the correct value for compiling calls. Increased the
5168 default value of LOOPREPEAT, and the number of significant figures in the
5169 times.
5170
5171 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5172
5173 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5174 a building problem on Windows NT with a FAT file system.
5175
5176
5177 Version 2.01 21-Oct-98
5178 ----------------------
5179
5180 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5181 to character tables built by pcre_maketables() in the current locale. If NULL
5182 is passed, the default tables are used.
5183
5184
5185 Version 2.00 24-Sep-98
5186 ----------------------
5187
5188 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5189 it any more.
5190
5191 2. Allow quantification of (?>) groups, and make it work correctly.
5192
5193 3. The first character computation wasn't working for (?>) groups.
5194
5195 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5196 end of the subject) and add 5.005's \z, which really does match only at the
5197 very end of the subject.
5198
5199 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5200
5201 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5202 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5203 localized options. All options to pcre_study() were also removed.
5204
5205 7. Add other new features from 5.005:
5206
5207 $(?<= positive lookbehind
5208 $(?<! negative lookbehind
5209 (?imsx-imsx) added the unsetting capability
5210 such a setting is global if at outer level; local otherwise
5211 (?imsx-imsx:) non-capturing groups with option setting
5212 (?(cond)re|re) conditional pattern matching
5213
5214 A backreference to itself in a repeated group matches the previous
5215 captured string.
5216
5217 8. General tidying up of studying (both automatic and via "study")
5218 consequential on the addition of new assertions.
5219
5220 9. As in 5.005, unlimited repeated groups that could match an empty substring
5221 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5222 runtime if any iteration does actually match an empty substring.
5223
5224 10. Include the RunTest script in the distribution.
5225
5226 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5227 discrepancies, some of which were old and were also with respect to 5.004. They
5228 have now been fixed.
5229
5230
5231 Version 1.09 28-Apr-98
5232 ----------------------
5233
5234 1. A negated single character class followed by a quantifier with a minimum
5235 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5236 program crashes, or just wrong answers. This did not apply to negated classes
5237 containing more than one character, or to minima other than one.
5238
5239
5240 Version 1.08 27-Mar-98
5241 ----------------------
5242
5243 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5244
5245 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5246 latter must appear before anything that relies on it in the pattern.
5247
5248
5249 Version 1.07 16-Feb-98
5250 ----------------------
5251
5252 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5253 repeat of a potentially empty string).
5254
5255
5256 Version 1.06 23-Jan-98
5257 ----------------------
5258
5259 1. Added Markus Oberhumer's little patches for C++.
5260
5261 2. Literal strings longer than 255 characters were broken.
5262
5263
5264 Version 1.05 23-Dec-97
5265 ----------------------
5266
5267 1. Negated character classes containing more than one character were failing if
5268 PCRE_CASELESS was set at run time.
5269
5270
5271 Version 1.04 19-Dec-97
5272 ----------------------
5273
5274 1. Corrected the man page, where some "const" qualifiers had been omitted.
5275
5276 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5277 input syntax.
5278
5279 3. Fixed memory leak which occurred when a regex with back references was
5280 matched with an offsets vector that wasn't big enough. The temporary memory
5281 that is used in this case wasn't being freed if the match failed.
5282
5283 4. Tidied pcretest to ensure it frees memory that it gets.
5284
5285 5. Temporary memory was being obtained in the case where the passed offsets
5286 vector was exactly big enough.
5287
5288 6. Corrected definition of offsetof() from change 5 below.
5289
5290 7. I had screwed up change 6 below and broken the rules for the use of
5291 setjmp(). Now fixed.
5292
5293
5294 Version 1.03 18-Dec-97
5295 ----------------------
5296
5297 1. A erroneous regex with a missing opening parenthesis was correctly
5298 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5299 on some systems.
5300
5301 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5302 it was reported that one broken compiler failed on the former because "code" is
5303 also an independent variable.
5304
5305 3. The erroneous regex a[]b caused an array overrun reference.
5306
5307 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5308 fail on data ending with that character. (It was going on too far, and checking
5309 the next character, typically a binary zero.) This was specific to the
5310 optimized code for single-character negative classes.
5311
5312 5. Added a contributed patch from the TIN world which does the following:
5313
5314 + Add an undef for memmove, in case the the system defines a macro for it.
5315
5316 + Add a definition of offsetof(), in case there isn't one. (I don't know
5317 the reason behind this - offsetof() is part of the ANSI standard - but
5318 it does no harm).
5319
5320 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5321 most of the places where whitespace preceded '#'. I have given up and
5322 allowed the remaining 2 cases to be at the margin.
5323
5324 + Rename some variables in pcre to eliminate shadowing. This seems very
5325 pedantic, but does no harm, of course.
5326
5327 6. Moved the call to setjmp() into its own function, to get rid of warnings
5328 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5329
5330 7. Constructs such as \d{8,} were compiling into the equivalent of
5331 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5332 outcome, but in this particular case used more store than had been allocated,
5333 which caused the bug to be discovered because it threw up an internal error.
5334
5335 8. The debugging code in both pcre and pcretest for outputting the compiled
5336 form of a regex was going wrong in the case of back references followed by
5337 curly-bracketed repeats.
5338
5339
5340 Version 1.02 12-Dec-97
5341 ----------------------
5342
5343 1. Typos in pcre.3 and comments in the source fixed.
5344
5345 2. Applied a contributed patch to get rid of places where it used to remove
5346 'const' from variables, and fixed some signed/unsigned and uninitialized
5347 variable warnings.
5348
5349 3. Added the "runtest" target to Makefile.
5350
5351 4. Set default compiler flag to -O2 rather than just -O.
5352
5353
5354 Version 1.01 19-Nov-97
5355 ----------------------
5356
5357 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5358 like /([ab]*)*/, that is, for classes with more than one character in them.
5359
5360 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5361 as /((?>a*))*/ (a PCRE_EXTRA facility).
5362
5363
5364 Version 1.00 18-Nov-97
5365 ----------------------
5366
5367 1. Added compile-time macros to support systems such as SunOS4 which don't have
5368 memmove() or strerror() but have other things that can be used instead.
5369
5370 2. Arranged that "make clean" removes the executables.
5371
5372
5373 Version 0.99 27-Oct-97
5374 ----------------------
5375
5376 1. Fixed bug in code for optimizing classes with only one character. It was
5377 initializing a 32-byte map regardless, which could cause it to run off the end
5378 of the memory it had got.
5379
5380 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5381
5382
5383 Version 0.98 22-Oct-97
5384 ----------------------
5385
5386 1. Fixed bug in code for handling temporary memory usage when there are more
5387 back references than supplied space in the ovector. This could cause segfaults.
5388
5389
5390 Version 0.97 21-Oct-97
5391 ----------------------
5392
5393 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5394
5395 2. Optimized negated single characters not to use a bit map.
5396
5397 3. Brought error texts together as macro definitions; clarified some of them;
5398 fixed one that was wrong - it said "range out of order" when it meant "invalid
5399 escape sequence".
5400
5401 4. Changed some char * arguments to const char *.
5402
5403 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5404
5405 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5406 pcretest.
5407
5408
5409 Version 0.96 16-Oct-97
5410 ----------------------
5411
5412 1. Added a simple "pgrep" utility to the distribution.
5413
5414 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5415 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5416 where "ddd" means "one or more decimal digits".
5417
5418 3. Fixed serious bug. If a pattern had a back reference, but the call to
5419 pcre_exec() didn't supply a large enough ovector to record the related
5420 identifying subpattern, the match always failed. PCRE now remembers the number
5421 of the largest back reference, and gets some temporary memory in which to save
5422 the offsets during matching if necessary, in order to ensure that
5423 backreferences always work.
5424
5425 4. Increased the compatibility with Perl in a number of ways:
5426
5427 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5428 to request this handling. The option can be set at compile or exec time.
5429
5430 (b) $ matches before a terminating newline by default; an option
5431 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5432 mode). The option can be set at compile or exec time.
5433
5434 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5435 the same as Perl's. If the decimal number it represents is less than 10
5436 or there aren't that many previous left capturing parentheses, an octal
5437 escape is read. Inside a character class, it's always an octal escape,
5438 even if it is a single digit.
5439
5440 (d) An escaped but undefined alphabetic character is taken as a literal,
5441 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5442 escapes.
5443
5444 (e) {0} is now permitted. (The previous item is removed from the compiled
5445 pattern).
5446
5447 5. Changed all the names of code files so that the basic parts are no longer
5448 than 10 characters, and abolished the teeny "globals.c" file.
5449
5450 6. Changed the handling of character classes; they are now done with a 32-byte
5451 bit map always.
5452
5453 7. Added the -d and /D options to pcretest to make it possible to look at the
5454 internals of compilation without having to recompile pcre.
5455
5456
5457 Version 0.95 23-Sep-97
5458 ----------------------
5459
5460 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5461 \x20 at the start of a run of normal characters. These were being treated as
5462 real characters, instead of the source characters being re-checked.
5463
5464
5465 Version 0.94 18-Sep-97
5466 ----------------------
5467
5468 1. The functions are now thread-safe, with the caveat that the global variables
5469 containing pointers to malloc() and free() or alternative functions are the
5470 same for all threads.
5471
5472 2. Get pcre_study() to generate a bitmap of initial characters for non-
5473 anchored patterns when this is possible, and use it if passed to pcre_exec().
5474
5475
5476 Version 0.93 15-Sep-97
5477 ----------------------
5478
5479 1. /(b)|(:+)/ was computing an incorrect first character.
5480
5481 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5482 but not actually doing anything yet.
5483
5484 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5485 as Perl does (e.g. [-az] or [az-]).
5486
5487 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5488 all possible positions.
5489
5490 5. Split up into different modules to avoid including unneeded functions in a
5491 compiled binary. However, compile and exec are still in one module. The "study"
5492 function is split off.
5493
5494 6. The character tables are now in a separate module whose source is generated
5495 by an auxiliary program - but can then be edited by hand if required. There are
5496 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5497 toupper() in the code.
5498
5499 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5500 make them global. Abolish the function for setting them, as the caller can now
5501 set them directly.
5502
5503
5504 Version 0.92 11-Sep-97
5505 ----------------------
5506
5507 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5508 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5509
5510 2. Caseless matching was not working in character classes if the characters in
5511 the pattern were in upper case.
5512
5513 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5514
5515 4. Make PCRE_ANCHORED public and accept as a compile option.
5516
5517 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5518 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5519 pass them.
5520
5521 6. Give an error if bad option bits passed at compile or run time.
5522
5523 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5524 pcretest to cause it to pass that flag.
5525
5526 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5527 options, and the first character, if set.
5528
5529 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5530
5531
5532 Version 0.91 10-Sep-97
5533 ----------------------
5534
5535 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5536 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5537
5538 2. PCRE was looping on encountering an indefinitely repeated back reference to
5539 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5540 Perl does - treats the match as successful.
5541
5542 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5