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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1419 - (show annotations)
Sun Dec 29 04:42:14 2013 UTC (5 years, 9 months ago) by zherczeg
File size: 259044 byte(s)
Error occurred while calculating annotation data.
Improve fast forward search in JIT.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.35-RC1 xx-xxxx-201x
5 -----------------------------
6
7 1. A new flag is set, when property checks are present in an XCLASS.
8 When this flag is not set, PCRE can perform certain optimizations
9 such as studying these XCLASS-es.
10
11 2. The auto-possessification of character sets were improved: a normal
12 and an extended character set can be compared now. Furthermore
13 the JIT compiler optimizes more character set checks.
14
15 3. Got rid of some compiler warnings for potentially uninitialized variables
16 that show up only when compiled with -O2.
17
18 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
19 of a match later then the end of the match. The pcretest program was not
20 handling the case sensibly - it was outputting from the start to the next
21 binary zero. It now reports this situation in a message, and outputs the
22 text from the end to the start.
23
24 5. Fast forward search is improved in JIT. Instead of the first three
25 characters, any three characters with fixed position can be searched.
26 Search order: first, last, middle.
27
28
29 Version 8.34 15-December-2013
30 -----------------------------
31
32 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
33 executable memory. Patch inspired by Carsten Klein.
34
35 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
36 this macro is never tested and has no effect, because the work to support
37 coverage involves only compiling and linking options and special targets in
38 the Makefile. The comment in config.h implied that defining the macro would
39 enable coverage support, which is totally false. There was also support for
40 setting this macro in the CMake files (my fault, I just copied it from
41 configure). SUPPORT_GCOV has now been removed.
42
43 3. Make a small performance improvement in strlen16() and strlen32() in
44 pcretest.
45
46 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
47 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
48 these cases). There was also one in pcretest.c.
49
50 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
51
52 6. In UTF mode, the code for checking whether a group could match an empty
53 string (which is used for indefinitely repeated groups to allow for
54 breaking an infinite loop) was broken when the group contained a repeated
55 negated single-character class with a character that occupied more than one
56 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
57 UTF-8 mode). The effect was undefined: the group might or might not be
58 deemed as matching an empty string, or the program might have crashed.
59
60 7. The code for checking whether a group could match an empty string was not
61 recognizing that \h, \H, \v, \V, and \R must match a character.
62
63 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
64 an empty string. If it can, pcretest shows this in its information output.
65
66 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
67 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
68 matched by pcre_exec() without using JIT:
69
70 (a) If the rest of the pattern did not match after a maximal run of
71 grapheme clusters, the code for backing up to try with fewer of them
72 did not always back up over a full grapheme when characters that do not
73 have the modifier quality were involved, e.g. Hangul syllables.
74
75 (b) If the match point in a subject started with modifier character, and
76 there was no match, the code could incorrectly back up beyond the match
77 point, and potentially beyond the first character in the subject,
78 leading to a segfault or an incorrect match result.
79
80 10. A conditional group with an assertion condition could lead to PCRE
81 recording an incorrect first data item for a match if no other first data
82 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
83 first data item, and therefore matched "ca" after "c" instead of at the
84 start.
85
86 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
87 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
88
89 12. The source of pcregrep now includes z/OS-specific code so that it can be
90 compiled for z/OS as part of the special z/OS distribution.
91
92 13. Added the -T and -TM options to pcretest.
93
94 14. The code in pcre_compile.c for creating the table of named capturing groups
95 has been refactored. Instead of creating the table dynamically during the
96 actual compiling pass, the information is remembered during the pre-compile
97 pass (on the stack unless there are more than 20 named groups, in which
98 case malloc() is used) and the whole table is created before the actual
99 compile happens. This has simplified the code (it is now nearly 150 lines
100 shorter) and prepared the way for better handling of references to groups
101 with duplicate names.
102
103 15. A back reference to a named subpattern when there is more than one of the
104 same name now checks them in the order in which they appear in the pattern.
105 The first one that is set is used for the reference. Previously only the
106 first one was inspected. This change makes PCRE more compatible with Perl.
107
108 16. Unicode character properties were updated from Unicode 6.3.0.
109
110 17. The compile-time code for auto-possessification has been refactored, based
111 on a patch by Zoltan Herczeg. It now happens after instead of during
112 compilation. The code is cleaner, and more cases are handled. The option
113 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
114 options in pcretest are provided to set it. It can also be set by
115 (*NO_AUTO_POSSESS) at the start of a pattern.
116
117 18. The character VT has been added to the default ("C" locale) set of
118 characters that match \s and are generally treated as white space,
119 following this same change in Perl 5.18. There is now no difference between
120 "Perl space" and "POSIX space". Whether VT is treated as white space in
121 other locales depends on the locale.
122
123 19. The code for checking named groups as conditions, either for being set or
124 for being recursed, has been refactored (this is related to 14 and 15
125 above). Processing unduplicated named groups should now be as fast at
126 numerical groups, and processing duplicated groups should be faster than
127 before.
128
129 20. Two patches to the CMake build system, by Alexander Barkov:
130
131 (1) Replace the "source" command by "." in CMakeLists.txt because
132 "source" is a bash-ism.
133
134 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
135 without these the CMake build does not work on Solaris.
136
137 21. Perl has changed its handling of \8 and \9. If there is no previously
138 encountered capturing group of those numbers, they are treated as the
139 literal characters 8 and 9 instead of a binary zero followed by the
140 literals. PCRE now does the same.
141
142 22. Following Perl, added \o{} to specify codepoints in octal, making it
143 possible to specify values greater than 0777 and also making them
144 unambiguous.
145
146 23. Perl now gives an error for missing closing braces after \x{... instead of
147 treating the string as literal. PCRE now does the same.
148
149 24. RunTest used to grumble if an inappropriate test was selected explicitly,
150 but just skip it when running all tests. This make it awkward to run ranges
151 of tests when one of them was inappropriate. Now it just skips any
152 inappropriate tests, as it always did when running all tests.
153
154 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
155 character types such as \d or \w, too many callouts were inserted, and the
156 data that they returned was rubbish.
157
158 26. In UCP mode, \s was not matching two of the characters that Perl matches,
159 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
160 were matched by \h. The code has now been refactored so that the lists of
161 the horizontal and vertical whitespace characters used for \h and \v (which
162 are defined only in one place) are now also used for \s.
163
164 27. Add JIT support for the 64 bit TileGX architecture.
165 Patch by Jiong Wang (Tilera Corporation).
166
167 28. Possessive quantifiers for classes (both explicit and automatically
168 generated) now use special opcodes instead of wrapping in ONCE brackets.
169
170 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
171 (because it's meaningless), this was not happening when PCRE_CASELESS was
172 set. Not wrong, but inefficient.
173
174 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
175 properties for \w, \d, etc) is present in a test regex. Otherwise if the
176 test contains no characters greater than 255, Perl doesn't realise it
177 should be using Unicode semantics.
178
179 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
180 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
181 does in Unicode mode.
182
183 32. Added the "forbid" facility to pcretest so that putting tests into the
184 wrong test files can sometimes be quickly detected.
185
186 33. There is now a limit (default 250) on the depth of nesting of parentheses.
187 This limit is imposed to control the amount of system stack used at compile
188 time. It can be changed at build time by --with-parens-nest-limit=xxx or
189 the equivalent in CMake.
190
191 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
192 errors. Perl warns for these when in warning mode, but PCRE has no facility
193 for giving warnings.
194
195 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
196 However, this was not working for (?!) because it is optimized to (*FAIL),
197 for which PCRE does not allow quantifiers. The optimization is now disabled
198 when a quantifier follows (?!). I can't see any use for this, but it makes
199 things uniform.
200
201 36. Perl no longer allows group names to start with digits, so I have made this
202 change also in PCRE. It simplifies the code a bit.
203
204 37. In extended mode, Perl ignores spaces before a + that indicates a
205 possessive quantifier. PCRE allowed a space before the quantifier, but not
206 before the possessive +. It now does.
207
208 38. The use of \K (reset reported match start) within a repeated possessive
209 group such as (a\Kb)*+ was not working.
210
211 40. Document that the same character tables must be used at compile time and
212 run time, and that the facility to pass tables to pcre_exec() and
213 pcre_dfa_exec() is for use only with saved/restored patterns.
214
215 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
216 features for Builds with MSVC:
217
218 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
219 builds). The .rc files can be used to set FileDescription and many other
220 attributes.
221
222 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
223 This allows higher-level build scripts which want .pdb files to avoid
224 hard-coding the exact files needed."
225
226 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
227 mean "start of word" and "end of word", respectively, as a transition aid.
228
229 43. A minimizing repeat of a class containing codepoints greater than 255 in
230 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
231 compiled to use the heap for recursion.
232
233 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
234 is configured.
235
236
237 Version 8.33 28-May-2013
238 ------------------------
239
240 1. Added 'U' to some constants that are compared to unsigned integers, to
241 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
242 variables that are added to signed variables, to ensure the result is
243 signed and can be negated.
244
245 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
246 CMake config files.
247
248 3. Revise the creation of config.h.generic so that all boolean macros are
249 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
250 overriding via -D on the command line possible.
251
252 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
253 to unsigned int is reported to make a quite noticeable speed difference in
254 a specific Windows environment. Testing on Linux did also appear to show
255 some benefit (and it is clearly not harmful). Also fixed the definition of
256 Xop which should be unsigned.
257
258 5. Related to (4), changing the definition of the intermediate variable cc
259 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
260 improvements.
261
262 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
263 unnecessary spaces.
264
265 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
266 and later.
267
268 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
269
270 9. Optimizing fast_forward_start_bits in JIT.
271
272 10. Adding support for callouts in JIT, and fixing some issues revealed
273 during this work. Namely:
274
275 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
276
277 (b) Minimum length was not checked before the matching is started.
278
279 11. The value of capture_last that is passed to callouts was incorrect in some
280 cases when there was a capture on one path that was subsequently abandoned
281 after a backtrack. Also, the capture_last value is now reset after a
282 recursion, since all captures are also reset in this case.
283
284 12. The interpreter no longer returns the "too many substrings" error in the
285 case when an overflowing capture is in a branch that is subsequently
286 abandoned after a backtrack.
287
288 13. In the pathological case when an offset vector of size 2 is used, pcretest
289 now prints out the matched string after a yield of 0 or 1.
290
291 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
292 Only supported by the JIT compiler at the moment.
293
294 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
295
296 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
297 offset of the starting point of the matching process, provided the offsets
298 vector is large enough.
299
300 17. The \A escape now records a lookbehind value of 1, though its execution
301 does not actually inspect the previous character. This is to ensure that,
302 in partial multi-segment matching, at least one character from the old
303 segment is retained when a new segment is processed. Otherwise, if there
304 are no lookbehinds in the pattern, \A might match incorrectly at the start
305 of a new segment.
306
307 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
308
309 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
310 gives some modest performance improvement in 8-bit mode.
311
312 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
313 be expressed in certain programming languages using Universal Character
314 Names.
315
316 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
317 which points out that "non characters" are not "characters that may not
318 appear in Unicode strings" but rather "characters that are reserved for
319 internal use and have only local meaning".
320
321 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
322 there was a conditional group that depended on an assertion, if the
323 assertion was false, the callout that immediately followed the alternation
324 in the condition was skipped when pcre_exec() was used for matching.
325
326 23. Allow an explicit callout to be inserted before an assertion that is the
327 condition for a conditional group, for compatibility with automatic
328 callouts, which always insert a callout at this point.
329
330 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
331 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
332
333 25. (*PRUNE) is now supported by the JIT compiler.
334
335 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
336
337 27. Fix the case where there are two or more SKIPs with arguments that may be
338 ignored.
339
340 28. (*SKIP) is now supported by the JIT compiler.
341
342 29. (*THEN) is now supported by the JIT compiler.
343
344 30. Update RunTest with additional test selector options.
345
346 31. The way PCRE handles backtracking verbs has been changed in two ways.
347
348 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
349 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
350 backtracking. In some cases this makes it more Perl-compatible, but Perl's
351 rather obscure rules do not always do the same thing.
352
353 (2) Previously, backtracking verbs were confined within assertions. This is
354 no longer the case for positive assertions, except for (*ACCEPT). Again,
355 this sometimes improves Perl compatibility, and sometimes does not.
356
357 32. A number of tests that were in test 2 because Perl did things differently
358 have been moved to test 1, because either Perl or PCRE has changed, and
359 these tests are now compatible.
360
361 32. Backtracking control verbs are now handled in the same way in JIT and
362 interpreter.
363
364 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
365 contained a forward subroutine reference caused a compile error.
366
367 34. Auto-detect and optimize limited repetitions in JIT.
368
369 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
370 blocking (*UTF) etc.
371
372 36. In the interpreter, maximizing pattern repetitions for characters and
373 character types now use tail recursion, which reduces stack usage.
374
375 37. The value of the max lookbehind was not correctly preserved if a compiled
376 and saved regex was reloaded on a host of different endianness.
377
378 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
379 of the compiled pattern block, expand the flags field from 16 to 32 bits
380 because it was almost full.
381
382 39. Try madvise first before posix_madvise.
383
384 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
385 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
386 original change locked out all matches of empty strings. This has been
387 changed so that one match of an empty string per line is recognized.
388 Subsequent searches on the same line (for colouring or for --only-matching,
389 for example) do not recognize empty strings.
390
391 41. Applied a user patch to fix a number of spelling mistakes in comments.
392
393 42. Data lines longer than 65536 caused pcretest to crash.
394
395 43. Clarified the data type for length and startoffset arguments for pcre_exec
396 and pcre_dfa_exec in the function-specific man pages, where they were
397 explicitly stated to be in bytes, never having been updated. I also added
398 some clarification to the pcreapi man page.
399
400 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
401 a segmentation fault.
402
403
404 Version 8.32 30-November-2012
405 -----------------------------
406
407 1. Improved JIT compiler optimizations for first character search and single
408 character iterators.
409
410 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
411 Patch by Daniel Richard G.
412
413 3. Single character iterator optimizations in the JIT compiler.
414
415 4. Improved JIT compiler optimizations for character ranges.
416
417 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
418 Reported by Giuseppe D'Angelo.
419
420 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
421 of a line, was being set incorrectly in cases where .* appeared inside
422 atomic brackets at the start of a pattern, or where there was a subsequent
423 *PRUNE or *SKIP.
424
425 7. Improved instruction cache flush for POWER/PowerPC.
426 Patch by Daniel Richard G.
427
428 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
429 grep:
430
431 (a) There is now no limit to the number of patterns to be matched.
432
433 (b) An error is given if a pattern is too long.
434
435 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
436 are now supported.
437
438 (d) --exclude-from and --include-from (multiple use) have been added.
439
440 (e) Exclusions and inclusions now apply to all files and directories, not
441 just to those obtained from scanning a directory recursively.
442
443 (f) Multiple uses of -f and --file-list are now supported.
444
445 (g) In a Windows environment, the default for -d has been changed from
446 "read" (the GNU grep default) to "skip", because otherwise the presence
447 of a directory in the file list provokes an error.
448
449 (h) The documentation has been revised and clarified in places.
450
451 9. Improve the matching speed of capturing brackets.
452
453 10. Changed the meaning of \X so that it now matches a Unicode extended
454 grapheme cluster.
455
456 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
457 out POSIX threads when JIT support is configured.
458
459 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
460
461 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
462 structure could go wrong in environments where size_t is not the same size
463 as int.
464
465 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
466 set.
467
468 15. The EBCDIC support had decayed; later updates to the code had included
469 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
470 general tidy up of EBCDIC-related issues, and the documentation was also
471 not quite right. There is now a test that can be run on ASCII systems to
472 check some of the EBCDIC-related things (but is it not a full test).
473
474 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
475 in a small tidy to the code.
476
477 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
478
479 18. If the --only-matching (-o) option in pcregrep is specified multiple
480 times, each one causes appropriate output. For example, -o1 -o2 outputs the
481 substrings matched by the 1st and 2nd capturing parentheses. A separating
482 string can be specified by --om-separator (default empty).
483
484 19. Improving the first n character searches.
485
486 20. Turn case lists for horizontal and vertical white space into macros so that
487 they are defined only once.
488
489 21. This set of changes together give more compatible Unicode case-folding
490 behaviour for characters that have more than one other case when UCP
491 support is available.
492
493 (a) The Unicode property table now has offsets into a new table of sets of
494 three or more characters that are case-equivalent. The MultiStage2.py
495 script that generates these tables (the pcre_ucd.c file) now scans
496 CaseFolding.txt instead of UnicodeData.txt for character case
497 information.
498
499 (b) The code for adding characters or ranges of characters to a character
500 class has been abstracted into a generalized function that also handles
501 case-independence. In UTF-mode with UCP support, this uses the new data
502 to handle characters with more than one other case.
503
504 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
505 whose other case is greater than 256 are now correctly matched
506 caselessly. Previously, the high codepoint matched the low one, but not
507 vice versa.
508
509 (d) The processing of \h, \H, \v, and \ in character classes now makes use
510 of the new class addition function, using character lists defined as
511 macros alongside the case definitions of 20 above.
512
513 (e) Caseless back references now work with characters that have more than
514 one other case.
515
516 (f) General caseless matching of characters with more than one other case
517 is supported.
518
519 22. Unicode character properties were updated from Unicode 6.2.0
520
521 23. Improved CMake support under Windows. Patch by Daniel Richard G.
522
523 24. Add support for 32-bit character strings, and UTF-32
524
525 25. Major JIT compiler update (code refactoring and bugfixing).
526 Experimental Sparc 32 support is added.
527
528 26. Applied a modified version of Daniel Richard G's patch to create
529 pcre.h.generic and config.h.generic by "make" instead of in the
530 PrepareRelease script.
531
532 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
533 pcre_compile.c when checking for a zero character.
534
535 28. Introducing a native interface for JIT. Through this interface, the compiled
536 machine code can be directly executed. The purpose of this interface is to
537 provide fast pattern matching, so several sanity checks are not performed.
538 However, feature tests are still performed. The new interface provides
539 1.4x speedup compared to the old one.
540
541 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
542 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
543 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
544
545 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
546 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
547 the "old" RFC 2279). Instead, it ended up passing a negative length to
548 pcre_exec().
549
550 31. Add support for GCC's visibility feature to hide internal functions.
551
552 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
553 "unknown -C option" after outputting 0 or 1.
554
555 33. There is now support for generating a code coverage report for the test
556 suite in environments where gcc is the compiler and lcov is installed. This
557 is mainly for the benefit of the developers.
558
559 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
560 unaddressable using valgrind annotations, allowing valgrind to detect
561 invalid memory accesses. This is mainly for the benefit of the developers.
562
563 25. (*UTF) can now be used to start a pattern in any of the three libraries.
564
565 26. Give configure error if --enable-cpp but no C++ compiler found.
566
567
568 Version 8.31 06-July-2012
569 -------------------------
570
571 1. Fixing a wrong JIT test case and some compiler warnings.
572
573 2. Removed a bashism from the RunTest script.
574
575 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
576 to unsigned type, result still unsigned" that was given by an MS compiler
577 on encountering the code "-sizeof(xxx)".
578
579 4. Partial matching support is added to the JIT compiler.
580
581 5. Fixed several bugs concerned with partial matching of items that consist
582 of more than one character:
583
584 (a) /^(..)\1/ did not partially match "aba" because checking references was
585 done on an "all or nothing" basis. This also applied to repeated
586 references.
587
588 (b) \R did not give a hard partial match if \r was found at the end of the
589 subject.
590
591 (c) \X did not give a hard partial match after matching one or more
592 characters at the end of the subject.
593
594 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
595 a partial match for the string "\r".
596
597 (e) When newline was set to CRLF, the metacharacter "." did not recognize
598 a partial match for a CR character at the end of the subject string.
599
600 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
601 running pcretest, the text "(JIT)" added to the output whenever JIT is
602 actually used to run the match.
603
604 7. Individual JIT compile options can be set in pcretest by following -s+[+]
605 or /S+[+] with a digit between 1 and 7.
606
607 8. OP_NOT now supports any UTF character not just single-byte ones.
608
609 9. (*MARK) control verb is now supported by the JIT compiler.
610
611 10. The command "./RunTest list" lists the available tests without actually
612 running any of them. (Because I keep forgetting what they all are.)
613
614 11. Add PCRE_INFO_MAXLOOKBEHIND.
615
616 12. Applied a (slightly modified) user-supplied patch that improves performance
617 when the heap is used for recursion (compiled with --disable-stack-for-
618 recursion). Instead of malloc and free for each heap frame each time a
619 logical recursion happens, frames are retained on a chain and re-used where
620 possible. This sometimes gives as much as 30% improvement.
621
622 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
623 call.
624
625 14. As documented, (*COMMIT) is now confined to within a positive assertion.
626
627 15. It is now possible to link pcretest with libedit as an alternative to
628 libreadline.
629
630 16. (*COMMIT) control verb is now supported by the JIT compiler.
631
632 17. The Unicode data tables have been updated to Unicode 6.1.0.
633
634 18. Added --file-list option to pcregrep.
635
636 19. Added binary file support to pcregrep, including the -a, --binary-files,
637 -I, and --text options.
638
639 20. The madvise function is renamed for posix_madvise for QNX compatibility
640 reasons. Fixed by Giuseppe D'Angelo.
641
642 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
643
644 22. Changed the option for creating symbolic links for 16-bit man pages from
645 -s to -sf so that re-installing does not cause issues.
646
647 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
648
649 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
650 matches in certain environments (the workspace was not being correctly
651 retained). Also added to pcre_dfa_exec() a simple plausibility check on
652 some of the workspace data at the beginning of a restart.
653
654 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
655 was not doing so when it should - probably a typo introduced by SVN 528
656 (change 8.10/14).
657
658 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
659 \w+ when the character tables indicated that \x{c4} was a word character.
660 There were several related cases, all because the tests for doing a table
661 lookup were testing for characters less than 127 instead of 255.
662
663 27. If a pattern contains capturing parentheses that are not used in a match,
664 their slots in the ovector are set to -1. For those that are higher than
665 any matched groups, this happens at the end of processing. In the case when
666 there were back references that the ovector was too small to contain
667 (causing temporary malloc'd memory to be used during matching), and the
668 highest capturing number was not used, memory off the end of the ovector
669 was incorrectly being set to -1. (It was using the size of the temporary
670 memory instead of the true size.)
671
672 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
673 ovector size, it uses memory at the end of the block that it has got.
674
675 29. Check for an overlong MARK name and give an error at compile time. The
676 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
677
678 30. JIT compiler update.
679
680 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
681 Rill for the patch.
682
683 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
684
685 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
686
687 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
688 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
689 was enabled.
690
691 35. Improve JIT code generation for greedy plus quantifier.
692
693 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
694 1 to "aa" instead of to an empty string. The bug affected repeated groups
695 that could potentially match an empty string.
696
697 37. Optimizing single character iterators in JIT.
698
699 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
700 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
701 codepoints that are too big for the mode are faulted, and in a UTF mode,
702 disallowed codepoints are also faulted.
703
704 39. If PCRE was compiled with UTF support, in three places in the DFA
705 matcher there was code that should only have been obeyed in UTF mode, but
706 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
707 processing when bytes with values greater than 127 were present. In 16-bit
708 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
709 both cases the values are those that cannot be the first data item in a UTF
710 character. The three items that might have provoked this were recursions,
711 possessively repeated groups, and atomic groups.
712
713 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
714 and pcregrep, because some OS require shared objects to be explicitly
715 passed to ld, causing the link step to fail if they are not.
716
717 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
718 mode, patterns that started with \h* or \R* might be incorrectly matched.
719
720
721 Version 8.30 04-February-2012
722 -----------------------------
723
724 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
725 name is defined in ctype.h.
726
727 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
728 only in quite long subpatterns.
729
730 3. Removed the function pcre_info(), which has been obsolete and deprecated
731 since it was replaced by pcre_fullinfo() in February 2000.
732
733 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
734 match a (*MARK), and the match failed at the start of the subject, a
735 reference to memory before the start of the subject could occur. This bug
736 was introduced by fix 17 of release 8.21.
737
738 5. A reference to an unset group with zero minimum repetition was giving
739 totally wrong answers (in non-JavaScript-compatibility mode). For example,
740 /(another)?(\1?)test/ matched against "hello world test". This bug was
741 introduced in release 8.13.
742
743 6. Add support for 16-bit character strings (a large amount of work involving
744 many changes and refactorings).
745
746 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
747 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
748 from a file.
749
750 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
751 rounding is not applied in this particular case).
752
753 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
754 if they appear, or are escaped, in patterns.
755
756 10. Get rid of a number of -Wunused-but-set-variable warnings.
757
758 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
759 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
760 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
761 also returns the mark "x". This bug applied to capturing parentheses,
762 non-capturing parentheses, and atomic parentheses. It also applied to some
763 assertions.
764
765 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
766 information out of configure.ac instead of relying on pcre.h.generic, which
767 is not stored in the repository.
768
769 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
770 -lreadline.
771
772 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
773
774 15. Applied Graycode's patch to put the top-level frame on the stack rather
775 than the heap when not using the stack for recursion. This gives a
776 performance improvement in many cases when recursion is not deep.
777
778 16. Experimental code added to "pcretest -C" to output the stack frame size.
779
780
781 Version 8.21 12-Dec-2011
782 ------------------------
783
784 1. Updating the JIT compiler.
785
786 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
787 are added as well.
788
789 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
790 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
791 calling _pcre_jit_exec. Some extra comments are added.
792
793 4. (*MARK) settings inside atomic groups that do not contain any capturing
794 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
795 was introduced by change 18 for 8.20.
796
797 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
798 ECMA-262 standard.
799
800 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
801 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
802 This bug was probably introduced by change 9 of 8.13.
803
804 7. While fixing 6 above, I noticed that a number of other items were being
805 incorrectly rejected as "not fixed length". This arose partly because newer
806 opcodes had not been added to the fixed-length checking code. I have (a)
807 corrected the bug and added tests for these items, and (b) arranged for an
808 error to occur if an unknown opcode is encountered while checking for fixed
809 length instead of just assuming "not fixed length". The items that were
810 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
811 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
812 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
813
814 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
815 being incorrectly compiled and would have given unpredicatble results.
816
817 9. A possessively repeated subpattern with minimum repeat count greater than
818 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
819 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
820 the first (A) could occur when it should not.
821
822 10. Add a cast and remove a redundant test from the code.
823
824 11. JIT should use pcre_malloc/pcre_free for allocation.
825
826 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
827 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
828 is anything other than /usr, -L is still shown).
829
830 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
831
832 14. Perl does not support \N without a following name in a [] class; PCRE now
833 also gives an error.
834
835 15. If a forward reference was repeated with an upper limit of around 2000,
836 it caused the error "internal error: overran compiling workspace". The
837 maximum number of forward references (including repeats) was limited by the
838 internal workspace, and dependent on the LINK_SIZE. The code has been
839 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
840 the default depends on LINK_SIZE. There is a new upper limit (for safety)
841 of around 200,000 forward references. While doing this, I also speeded up
842 the filling in of repeated forward references.
843
844 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
845 incorrectly expecting the subject to contain another "a" after the start.
846
847 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
848 in the match, the SKIP should be ignored. This was not happening; instead
849 the SKIP was being treated as NOMATCH. For patterns such as
850 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
851 tested.
852
853 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
854 now much more compatible with Perl, in particular in cases where the result
855 is a non-match for a non-anchored pattern. For example, if
856 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
857 "m", where previously it did not return a name. A side effect of this
858 change is that for partial matches, the last encountered mark name is
859 returned, as for non matches. A number of tests that were previously not
860 Perl-compatible have been moved into the Perl-compatible test files. The
861 refactoring has had the pleasing side effect of removing one argument from
862 the match() function, thus reducing its stack requirements.
863
864 19. If the /S+ option was used in pcretest to study a pattern using JIT,
865 subsequent uses of /S (without +) incorrectly behaved like /S+.
866
867 21. Retrieve executable code size support for the JIT compiler and fixing
868 some warnings.
869
870 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
871 not work when the shorter character appeared right at the end of the
872 subject string.
873
874 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
875 systems.
876
877 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
878 output it when the /M option is used in pcretest.
879
880 25. The CheckMan script was not being included in the distribution. Also, added
881 an explicit "perl" to run Perl scripts from the PrepareRelease script
882 because this is reportedly needed in Windows.
883
884 26. If study data was being save in a file and studying had not found a set of
885 "starts with" bytes for the pattern, the data written to the file (though
886 never used) was taken from uninitialized memory and so caused valgrind to
887 complain.
888
889 27. Updated RunTest.bat as provided by Sheri Pierce.
890
891 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
892
893 29. Computation of memory usage for the table of capturing group names was
894 giving an unnecessarily large value.
895
896
897 Version 8.20 21-Oct-2011
898 ------------------------
899
900 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
901 a POSIX class. After further experiments with Perl, which convinced me that
902 Perl has bugs and confusions, a closing square bracket is no longer allowed
903 in a POSIX name. This bug also affected patterns with classes that started
904 with full stops.
905
906 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
907 captured substring, but while checking the failing first alternative,
908 substring 1 is temporarily captured. If the output vector supplied to
909 pcre_exec() was not big enough for this capture, the yield of the function
910 was still zero ("insufficient space for captured substrings"). This cannot
911 be totally fixed without adding another stack variable, which seems a lot
912 of expense for a edge case. However, I have improved the situation in cases
913 such as /(a)(b)x|abc/ matched against "abc", where the return code
914 indicates that fewer than the maximum number of slots in the ovector have
915 been set.
916
917 3. Related to (2) above: when there are more back references in a pattern than
918 slots in the output vector, pcre_exec() uses temporary memory during
919 matching, and copies in the captures as far as possible afterwards. It was
920 using the entire output vector, but this conflicts with the specification
921 that only 2/3 is used for passing back captured substrings. Now it uses
922 only the first 2/3, for compatibility. This is, of course, another edge
923 case.
924
925 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
926 main code base, and can be used by building with --enable-jit. When this is
927 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
928 runtime --no-jit option is given.
929
930 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
931 ovector, the return from the function was zero, implying that there were
932 other matches that did not fit. The correct "exactly full" value is now
933 returned.
934
935 6. If a subpattern that was called recursively or as a subroutine contained
936 (*PRUNE) or any other control that caused it to give a non-standard return,
937 invalid errors such as "Error -26 (nested recursion at the same subject
938 position)" or even infinite loops could occur.
939
940 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
941 computing the minimum length on reaching *ACCEPT, and so ended up with the
942 wrong value of 1 rather than 0. Further investigation indicates that
943 computing a minimum subject length in the presence of *ACCEPT is difficult
944 (think back references, subroutine calls), and so I have changed the code
945 so that no minimum is registered for a pattern that contains *ACCEPT.
946
947 8. If (*THEN) was present in the first (true) branch of a conditional group,
948 it was not handled as intended. [But see 16 below.]
949
950 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
951 Sheri Pierce.
952
953 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
954 the first byte in a match must be "a".
955
956 11. Change 17 for 8.13 increased the recursion depth for patterns like
957 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
958 pattern contains any instances of (*THEN). If it does not, the old
959 optimizations are restored. It would be nice to do this on a per-group
960 basis, but at the moment that is not feasible.
961
962 12. In some environments, the output of pcretest -C is CRLF terminated. This
963 broke RunTest's code that checks for the link size. A single white space
964 character after the value is now allowed for.
965
966 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
967 For "fr", it uses the Windows-specific input and output files.
968
969 14. If (*THEN) appeared in a group that was called recursively or as a
970 subroutine, it did not work as intended. [But see next item.]
971
972 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
973 pattern fragments (but not containing any | characters). If A and B are
974 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
975 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
976 D. In other words, Perl considers parentheses that do not contain any |
977 characters to be part of a surrounding alternative, whereas PCRE was
978 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
979 differently. PCRE now behaves in the same way as Perl, except in the case
980 of subroutine/recursion calls such as (?1) which have in any case always
981 been different (but PCRE had them first :-).
982
983 16. Related to 15 above: Perl does not treat the | in a conditional group as
984 creating alternatives. Such a group is treated in the same way as an
985 ordinary group without any | characters when processing (*THEN). PCRE has
986 been changed to match Perl's behaviour.
987
988 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
989 RunGrepTest script failed.
990
991 18. Change 22 for version 13 caused atomic groups to use more stack. This is
992 inevitable for groups that contain captures, but it can lead to a lot of
993 stack use in large patterns. The old behaviour has been restored for atomic
994 groups that do not contain any capturing parentheses.
995
996 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
997 suppress the check for a minimum subject length at run time. (If it was
998 given to pcre_exec() or pcre_dfa_exec() it did work.)
999
1000 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1001 fail to work when decoding hex characters in data strings in EBCDIC
1002 environments.
1003
1004 21. It appears that in at least one Mac OS environment, the isxdigit() function
1005 is implemented as a macro that evaluates to its argument more than once,
1006 contravening the C 90 Standard (I haven't checked a later standard). There
1007 was an instance in pcretest which caused it to go wrong when processing
1008 \x{...} escapes in subject strings. The has been rewritten to avoid using
1009 things like p++ in the argument of isxdigit().
1010
1011
1012 Version 8.13 16-Aug-2011
1013 ------------------------
1014
1015 1. The Unicode data tables have been updated to Unicode 6.0.0.
1016
1017 2. Two minor typos in pcre_internal.h have been fixed.
1018
1019 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1020 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1021 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1022
1023 4. There were a number of related bugs in the code for matching backrefences
1024 caselessly in UTF-8 mode when codes for the characters concerned were
1025 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1026 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1027 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1028 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1029 2-byte code at the end of the subject (it thought there wasn't enough data
1030 left).
1031
1032 5. Comprehensive information about what went wrong is now returned by
1033 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1034 as the output vector has at least 2 elements. The offset of the start of
1035 the failing character and a reason code are placed in the vector.
1036
1037 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1038 now returned is for the first byte of the failing character, instead of the
1039 last byte inspected. This is an incompatible change, but I hope it is small
1040 enough not to be a problem. It makes the returned offset consistent with
1041 pcre_exec() and pcre_dfa_exec().
1042
1043 7. pcretest now gives a text phrase as well as the error number when
1044 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1045 failure, the offset and reason code are output.
1046
1047 8. When \R was used with a maximizing quantifier it failed to skip backwards
1048 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1049 back over a single character (\n). This seems wrong (because it treated the
1050 two characters as a single entity when going forwards), conflicts with the
1051 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1052 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1053 has been changed.
1054
1055 9. Some internal refactoring has changed the processing so that the handling
1056 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1057 time (the PCRE_DOTALL option was changed this way some time ago: version
1058 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1059 which was always a bit of a fudge. It also means that there is one less
1060 argument for the match() function, which reduces its stack requirements
1061 slightly. This change also fixes an incompatibility with Perl: the pattern
1062 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1063
1064 10. More internal refactoring has drastically reduced the number of recursive
1065 calls to match() for possessively repeated groups such as (abc)++ when
1066 using pcre_exec().
1067
1068 11. While implementing 10, a number of bugs in the handling of groups were
1069 discovered and fixed:
1070
1071 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1072 (a|)*(?1) gave a compile-time internal error.
1073 ((a|)+)+ did not notice that the outer group could match an empty string.
1074 (^a|^)+ was not marked as anchored.
1075 (.*a|.*)+ was not marked as matching at start or after a newline.
1076
1077 12. Yet more internal refactoring has removed another argument from the match()
1078 function. Special calls to this function are now indicated by setting a
1079 value in a variable in the "match data" data block.
1080
1081 13. Be more explicit in pcre_study() instead of relying on "default" for
1082 opcodes that mean there is no starting character; this means that when new
1083 ones are added and accidentally left out of pcre_study(), testing should
1084 pick them up.
1085
1086 14. The -s option of pcretest has been documented for ages as being an old
1087 synonym of -m (show memory usage). I have changed it to mean "force study
1088 for every regex", that is, assume /S for every regex. This is similar to -i
1089 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1090 using it. It makes it easier to run collections of tests with and without
1091 study enabled, and thereby test pcre_study() more easily. All the standard
1092 tests are now run with and without -s (but some patterns can be marked as
1093 "never study" - see 20 below).
1094
1095 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1096 restoration of the capturing data to the outer values was not happening
1097 correctly.
1098
1099 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1100 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1101 pattern had matched an empty string, and so incorrectly returned a no
1102 match.
1103
1104 17. There was optimizing code for the last branch of non-capturing parentheses,
1105 and also for the obeyed branch of a conditional subexpression, which used
1106 tail recursion to cut down on stack usage. Unfortunately, now that there is
1107 the possibility of (*THEN) occurring in these branches, tail recursion is
1108 no longer possible because the return has to be checked for (*THEN). These
1109 two optimizations have therefore been removed. [But see 8.20/11 above.]
1110
1111 18. If a pattern containing \R was studied, it was assumed that \R always
1112 matched two bytes, thus causing the minimum subject length to be
1113 incorrectly computed because \R can also match just one byte.
1114
1115 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1116 was incorrectly computed.
1117
1118 20. If /S is present twice on a test pattern in pcretest input, it now
1119 *disables* studying, thereby overriding the use of -s on the command line
1120 (see 14 above). This is necessary for one or two tests to keep the output
1121 identical in both cases.
1122
1123 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1124 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1125
1126 22. When an atomic group that contained a capturing parenthesis was
1127 successfully matched, but the branch in which it appeared failed, the
1128 capturing was not being forgotten if a higher numbered group was later
1129 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1130 group 1 to "a", when in fact it should be unset. This applied to multi-
1131 branched capturing and non-capturing groups, repeated or not, and also to
1132 positive assertions (capturing in negative assertions does not happen
1133 in PCRE) and also to nested atomic groups.
1134
1135 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1136 subject after a captured substring, to make it easier to tell which of a
1137 number of identical substrings has been captured.
1138
1139 24. The way atomic groups are processed by pcre_exec() has been changed so that
1140 if they are repeated, backtracking one repetition now resets captured
1141 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1142 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1143 "aaa". Previously, it would have been "a". As part of this code
1144 refactoring, the way recursive calls are handled has also been changed.
1145
1146 25. If an assertion condition captured any substrings, they were not passed
1147 back unless some other capturing happened later. For example, if
1148 (?(?=(a))a) was matched against "a", no capturing was returned.
1149
1150 26. When studying a pattern that contained subroutine calls or assertions,
1151 the code for finding the minimum length of a possible match was handling
1152 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1153 group 1 called group 2 while simultaneously a separate group 2 called group
1154 1). A stack overflow occurred in this case. I have fixed this by limiting
1155 the recursion depth to 10.
1156
1157 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1158 Fortmann. This supports explicit test numbers on the command line, and has
1159 argument validation and error reporting.
1160
1161 28. An instance of \X with an unlimited repeat could fail if at any point the
1162 first character it looked at was a mark character.
1163
1164 29. Some minor code refactoring concerning Unicode properties and scripts
1165 should reduce the stack requirement of match() slightly.
1166
1167 30. Added the '=' option to pcretest to check the setting of unused capturing
1168 slots at the end of the pattern, which are documented as being -1, but are
1169 not included in the return count.
1170
1171 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1172 compiled something random. Now it gives a compile-time error (as does
1173 Perl).
1174
1175 32. A *MARK encountered during the processing of a positive assertion is now
1176 recorded and passed back (compatible with Perl).
1177
1178 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1179 had alternative anchored branches, the search for a second match in a line
1180 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1181 matched the line "0102" twice. The same bug affected patterns that started
1182 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1183 twice.
1184
1185 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1186 does, and because of capturing effects, quantifying parenthesized
1187 assertions may at times be useful. Quantifiers are now allowed for
1188 parenthesized assertions.
1189
1190 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1191
1192 36. \g was being checked for fancy things in a character class, when it should
1193 just be a literal "g".
1194
1195 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1196 appearance of a nested POSIX class supersedes an apparent external class.
1197 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1198 unescaped square brackets may also appear as part of class names. For
1199 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1200 more like Perl. (But see 8.20/1 above.)
1201
1202 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1203 was because it thought it was \N{name}, which is not supported).
1204
1205 39. Add minix to OS list not supporting the -S option in pcretest.
1206
1207 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1208 cannot analyze patterns in sufficient detail to catch mutual recursions
1209 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1210 subgroup is called recursively as a subpattern for a second time at the
1211 same position in the subject string. In previous releases this might have
1212 been caught by the recursion limit, or it might have run out of stack.
1213
1214 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1215 happen only once. PCRE was, however incorrectly giving a compile time error
1216 "recursive call could loop indefinitely" because it cannot analyze the
1217 pattern in sufficient detail. The compile time test no longer happens when
1218 PCRE is compiling a conditional subpattern, but actual runaway loops are
1219 now caught at runtime (see 40 above).
1220
1221 42. It seems that Perl allows any characters other than a closing parenthesis
1222 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1223 has been changed to be the same.
1224
1225 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1226 as not to get warnings when autogen.sh is called. Also changed
1227 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1228
1229 44. To help people who use pcregrep to scan files containing exceedingly long
1230 lines, the following changes have been made:
1231
1232 (a) The default value of the buffer size parameter has been increased from
1233 8K to 20K. (The actual buffer used is three times this size.)
1234
1235 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1236 PCRE is built.
1237
1238 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1239 to be set at run time.
1240
1241 (d) Numerical values in pcregrep options can be followed by K or M, for
1242 example --buffer-size=50K.
1243
1244 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1245 given and the return code is set to 2.
1246
1247 45. Add a pointer to the latest mark to the callout data block.
1248
1249 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1250 partial match of an empty string instead of no match. This was specific to
1251 the use of ".".
1252
1253 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1254 complete match instead of a partial match. This bug was dependent on both
1255 the PCRE_UTF8 and PCRE_DOTALL options being set.
1256
1257 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1258 starting byte set, because \b was not being ignored.
1259
1260
1261 Version 8.12 15-Jan-2011
1262 ------------------------
1263
1264 1. Fixed some typos in the markup of the man pages, and wrote a script that
1265 checks for such things as part of the documentation building process.
1266
1267 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1268 --match-limit and --recursion-limit options (added for 8.11). In
1269 particular, this made one of the standard tests fail. (The integer value
1270 went into the wrong half of a long int.)
1271
1272 3. If the --colour option was given to pcregrep with -v (invert match), it
1273 did strange things, either producing crazy output, or crashing. It should,
1274 of course, ignore a request for colour when reporting lines that do not
1275 match.
1276
1277 4. Another pcregrep bug caused similar problems if --colour was specified with
1278 -M (multiline) and the pattern match finished with a line ending.
1279
1280 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1281 matched in multiline mode, the following line was shown as part of the
1282 match. This seems wrong, so I have changed it.
1283
1284 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1285 the check for further matches in the same line (so they could be coloured)
1286 to overrun the end of the current line. If another match was found, it was
1287 incorrectly shown (and then shown again when found in the next line).
1288
1289 7. If pcregrep was compiled under Windows, there was a reference to the
1290 function pcregrep_exit() before it was defined. I am assuming this was
1291 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1292 reported by a user. I've moved the definition above the reference.
1293
1294
1295 Version 8.11 10-Dec-2010
1296 ------------------------
1297
1298 1. (*THEN) was not working properly if there were untried alternatives prior
1299 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1300 backtracked to try for "b" instead of moving to the next alternative branch
1301 at the same level (in this case, to look for "c"). The Perl documentation
1302 is clear that when (*THEN) is backtracked onto, it goes to the "next
1303 alternative in the innermost enclosing group".
1304
1305 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1306 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1307 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1308 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1309 (*THEN).
1310
1311 3. If \s appeared in a character class, it removed the VT character from
1312 the class, even if it had been included by some previous item, for example
1313 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1314 of \s, but is part of the POSIX "space" class.)
1315
1316 4. A partial match never returns an empty string (because you can always
1317 match an empty string at the end of the subject); however the checking for
1318 an empty string was starting at the "start of match" point. This has been
1319 changed to the "earliest inspected character" point, because the returned
1320 data for a partial match starts at this character. This means that, for
1321 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1322 (previously it gave "no match").
1323
1324 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1325 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1326 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1327 has an implication that the given string is incomplete (because a partial
1328 match is preferred over a full match). For this reason, these items now
1329 give a partial match in this situation. [Aside: previously, the one case
1330 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1331 match rather than a full match, which was wrong by the old rules, but is
1332 now correct.]
1333
1334 6. There was a bug in the handling of #-introduced comments, recognized when
1335 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1336 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1337 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1338 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1339 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1340 places in pcre_compile().
1341
1342 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1343 comments when looking ahead for named forward references to subpatterns,
1344 the only newline sequence it recognized was NL. It now handles newlines
1345 according to the set newline convention.
1346
1347 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1348 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1349 cater for a lack of strerror(). These oversights have been fixed.
1350
1351 9. Added --match-limit and --recursion-limit to pcregrep.
1352
1353 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1354
1355 11. When the -o option was used, pcregrep was setting a return code of 1, even
1356 when matches were found, and --line-buffered was not being honoured.
1357
1358 12. Added an optional parentheses number to the -o and --only-matching options
1359 of pcregrep.
1360
1361 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1362 can match an empty string. The code to do it in pcretest and pcredemo
1363 needed fixing:
1364
1365 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1366 only one byte after an empty string match just before CRLF (this case
1367 just got forgotten; "any" and "anycrlf" were OK).
1368
1369 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1370 mode when an empty string match preceded an ASCII character followed by
1371 a non-ASCII character. (The code for advancing by one character rather
1372 than one byte was nonsense.)
1373
1374 (c) The pcredemo.c sample program did not have any code at all to handle
1375 the cases when CRLF is a valid newline sequence.
1376
1377 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1378 as a starting offset was within the subject string. There is now a new
1379 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1380 negative or greater than the length of the string. In order to test this,
1381 pcretest is extended to allow the setting of negative starting offsets.
1382
1383 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1384 starting offset points to the beginning of a UTF-8 character was
1385 unnecessarily clumsy. I tidied it up.
1386
1387 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1388 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1389
1390 17. Nobody had reported that the --include_dir option, which was added in
1391 release 7.7 should have been called --include-dir (hyphen, not underscore)
1392 for compatibility with GNU grep. I have changed it to --include-dir, but
1393 left --include_dir as an undocumented synonym, and the same for
1394 --exclude-dir, though that is not available in GNU grep, at least as of
1395 release 2.5.4.
1396
1397 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1398 characters from a string of bytes) have been redefined so as not to use
1399 loops, in order to improve performance in some environments. At the same
1400 time, I abstracted some of the common code into auxiliary macros to save
1401 repetition (this should not affect the compiled code).
1402
1403 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1404 compile-time error is now given if \c is not followed by an ASCII
1405 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1406 different, and any byte value is allowed.)
1407
1408 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1409 START_OPTIMIZE option, which is now allowed at compile time - but just
1410 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1411 to pcregrep and other applications that have no direct access to PCRE
1412 options. The new /Y option in pcretest sets this option when calling
1413 pcre_compile().
1414
1415 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1416 back references. Groups containing recursive back references were forced to
1417 be atomic by that change, but in the case of named groups, the amount of
1418 memory required was incorrectly computed, leading to "Failed: internal
1419 error: code overflow". This has been fixed.
1420
1421 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1422 pcretest.c, to avoid build problems in some Borland environments.
1423
1424
1425 Version 8.10 25-Jun-2010
1426 ------------------------
1427
1428 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1429 THEN.
1430
1431 2. (*ACCEPT) was not working when inside an atomic group.
1432
1433 3. Inside a character class, \B is treated as a literal by default, but
1434 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1435 causes the error). The code is unchanged, but I tidied the documentation.
1436
1437 4. Inside a character class, PCRE always treated \R and \X as literals,
1438 whereas Perl faults them if its -w option is set. I have changed PCRE so
1439 that it faults them when PCRE_EXTRA is set.
1440
1441 5. Added support for \N, which always matches any character other than
1442 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1443
1444 6. When compiling pcregrep with newer versions of gcc which may have
1445 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1446 declared with attribute warn_unused_result" were given. Just casting the
1447 result to (void) does not stop the warnings; a more elaborate fudge is
1448 needed. I've used a macro to implement this.
1449
1450 7. Minor change to pcretest.c to avoid a compiler warning.
1451
1452 8. Added four artifical Unicode properties to help with an option to make
1453 \s etc use properties (see next item). The new properties are: Xan
1454 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1455
1456 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1457 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1458 this option. Modified pcretest to add /W to test this facility. Added
1459 REG_UCP to make it available via the POSIX interface.
1460
1461 10. Added --line-buffered to pcregrep.
1462
1463 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1464 studied, and the match started with a letter with a code point greater than
1465 127 whose first byte was different to the first byte of the other case of
1466 the letter, the other case of this starting letter was not recognized
1467 (#976).
1468
1469 12. If a pattern that was studied started with a repeated Unicode property
1470 test, for example, \p{Nd}+, there was the theoretical possibility of
1471 setting up an incorrect bitmap of starting bytes, but fortunately it could
1472 not have actually happened in practice until change 8 above was made (it
1473 added property types that matched character-matching opcodes).
1474
1475 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1476 possible starting bytes for non-anchored patterns.
1477
1478 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1479 \R, and also a number of cases that involve Unicode properties, both
1480 explicit and implicit when PCRE_UCP is set.
1481
1482 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1483 input, it could crash or give wrong results if characters with values
1484 greater than 0xc0 were present in the subject string. (Detail: it assumed
1485 UTF-8 input when processing these items.)
1486
1487 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1488 size_t is 64-bit (#991).
1489
1490 17. Added a check for running out of memory when PCRE is compiled with
1491 --disable-stack-for-recursion (#990).
1492
1493 18. If the last data line in a file for pcretest does not have a newline on
1494 the end, a newline was missing in the output.
1495
1496 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1497 less than 128) in its various bitmaps. However, there is a facility for
1498 generating tables according to the current locale when PCRE is compiled. It
1499 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1500 characters, are recognized by isspace() and therefore were getting set in
1501 these tables, and indeed these tables seem to approximate to ISO 8859. This
1502 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1503 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1504 which of course cannot start UTF-8 characters. I have changed the code so
1505 that only real ASCII characters (less than 128) and the correct starting
1506 bytes for UTF-8 encodings are set for characters greater than 127 when in
1507 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1508 altogether.)
1509
1510 20. Added the /T option to pcretest so as to be able to run tests with non-
1511 standard character tables, thus making it possible to include the tests
1512 used for 19 above in the standard set of tests.
1513
1514 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1515 reference to a subpattern the other side of a comment that contains an
1516 opening parenthesis caused either an internal compiling error, or a
1517 reference to the wrong subpattern.
1518
1519
1520 Version 8.02 19-Mar-2010
1521 ------------------------
1522
1523 1. The Unicode data tables have been updated to Unicode 5.2.0.
1524
1525 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1526 configured.
1527
1528 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1529 original author of that file, following a query about its status.
1530
1531 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1532 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1533
1534 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1535 quantifier applied to a forward-referencing subroutine call, could compile
1536 incorrect code or give the error "internal error: previously-checked
1537 referenced subpattern not found".
1538
1539 6. Both MS Visual Studio and Symbian OS have problems with initializing
1540 variables to point to external functions. For these systems, therefore,
1541 pcre_malloc etc. are now initialized to local functions that call the
1542 relevant global functions.
1543
1544 7. There were two entries missing in the vectors called coptable and poptable
1545 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1546 I've fixed the data, and added a kludgy way of testing at compile time that
1547 the lengths are correct (equal to the number of opcodes).
1548
1549 8. Following on from 7, I added a similar kludge to check the length of the
1550 eint vector in pcreposix.c.
1551
1552 9. Error texts for pcre_compile() are held as one long string to avoid too
1553 much relocation at load time. To find a text, the string is searched,
1554 counting zeros. There was no check for running off the end of the string,
1555 which could happen if a new error number was added without updating the
1556 string.
1557
1558 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1559
1560 11. \K was not working if it appeared in an atomic group or in a group that
1561 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1562 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1563 the assertion is positive, but not if it is negative.
1564
1565 12. Change 11 fortuitously reduced the size of the stack frame used in the
1566 "match()" function of pcre_exec.c by one pointer. Forthcoming
1567 implementation of support for (*MARK) will need an extra pointer on the
1568 stack; I have reserved it now, so that the stack frame size does not
1569 decrease.
1570
1571 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1572 item in branch that calls a recursion is a subroutine call - as in the
1573 second branch in the above example - was incorrectly given the compile-
1574 time error "recursive call could loop indefinitely" because pcre_compile()
1575 was not correctly checking the subroutine for matching a non-empty string.
1576
1577 14. The checks for overrunning compiling workspace could trigger after an
1578 overrun had occurred. This is a "should never occur" error, but it can be
1579 triggered by pathological patterns such as hundreds of nested parentheses.
1580 The checks now trigger 100 bytes before the end of the workspace.
1581
1582 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1583
1584
1585 Version 8.01 19-Jan-2010
1586 ------------------------
1587
1588 1. If a pattern contained a conditional subpattern with only one branch (in
1589 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1590 computed the wrong minimum data length (which is of course zero for such
1591 subpatterns). This could cause incorrect "no match" results.
1592
1593 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1594 the pattern is reset in the first branch, pcre_compile() failed with
1595 "internal error: code overflow at offset...". This happened only when
1596 the reset was to the original external option setting. (An optimization
1597 abstracts leading options settings into an external setting, which was the
1598 cause of this.)
1599
1600 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1601 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1602 assertion pattern did not match (meaning that the assertion was true), it
1603 was incorrectly treated as false if the SKIP had been reached during the
1604 matching. This also applied to assertions used as conditions.
1605
1606 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1607 assertion subpattern, including such a pattern used as a condition,
1608 unpredictable results occurred, instead of the error return
1609 PCRE_ERROR_DFA_UITEM.
1610
1611 5. The C++ GlobalReplace function was not working like Perl for the special
1612 situation when an empty string is matched. It now does the fancy magic
1613 stuff that is necessary.
1614
1615 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1616 removed. (These were left over from very, very early versions of PCRE.)
1617
1618 7. Some cosmetic changes to the code to make life easier when compiling it
1619 as part of something else:
1620
1621 (a) Change DEBUG to PCRE_DEBUG.
1622
1623 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1624 called "current" as "current_branch", to prevent a collision with the
1625 Linux macro when compiled as a kernel module.
1626
1627 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1628 prevent a collision with the Linux macro when compiled as a kernel
1629 module.
1630
1631 8. In pcre_compile() there are some checks for integer overflows that used to
1632 cast potentially large values to (double). This has been changed to that
1633 when building, a check for int64_t is made, and if it is found, it is used
1634 instead, thus avoiding the use of floating point arithmetic. (There is no
1635 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1636 double.
1637
1638 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1639 2005 (difference between two addresses compared to an unsigned value).
1640
1641 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1642 custom one, because of the following reported problem in Windows:
1643
1644 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1645 under Win32.
1646 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1647 therefore missing the function definition.
1648 - The compiler thus generates a "C" signature for the test function.
1649 - The linker fails to find the "C" function.
1650 - PCRE fails to configure if asked to do so against libbz2.
1651
1652 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1653 messages were output:
1654
1655 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1656 rerunning libtoolize, to keep the correct libtool macros in-tree.
1657 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1658
1659 I have done both of these things.
1660
1661 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1662 most of the time, it *can* run out if it is given a pattern that contains a
1663 runaway infinite recursion. I updated the discussion in the pcrestack man
1664 page.
1665
1666 13. Now that we have gone to the x.xx style of version numbers, the minor
1667 version may start with zero. Using 08 or 09 is a bad idea because users
1668 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1669 interpreted as invalid octal numbers. I've updated the previous comment in
1670 configure.ac, and also added a check that gives an error if 08 or 09 are
1671 used.
1672
1673 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1674 causing partial matching to fail when the end of the subject matched \W
1675 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1676
1677 15. There were some discrepancies between the declarations in pcre_internal.h
1678 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1679 their definitions. The declarations used "const uschar *" and the
1680 definitions used USPTR. Even though USPTR is normally defined as "const
1681 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1682 reported that: "This difference in casting confuses some C++ compilers, for
1683 example, SunCC recognizes above declarations as different functions and
1684 generates broken code for hbpcre." I have changed the declarations to use
1685 USPTR.
1686
1687 16. GNU libtool is named differently on some systems. The autogen.sh script now
1688 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1689 (FreeBSD).
1690
1691 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1692 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1693 comment: "Figure out how to create a longlong from a string: strtoll and
1694 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1695 instance, but it only takes 2 args instead of 3!"
1696
1697 18. A subtle bug concerned with back references has been fixed by a change of
1698 specification, with a corresponding code fix. A pattern such as
1699 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1700 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1701 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1702 same bug. Such groups have to be quantified to be useful, or contained
1703 inside another quantified group. (If there's no repetition, the reference
1704 can never match.) The problem arises because, having left the group and
1705 moved on to the rest of the pattern, a later failure that backtracks into
1706 the group uses the captured value from the final iteration of the group
1707 rather than the correct earlier one. I have fixed this in PCRE by forcing
1708 any group that contains a reference to itself to be an atomic group; that
1709 is, there cannot be any backtracking into it once it has completed. This is
1710 similar to recursive and subroutine calls.
1711
1712
1713 Version 8.00 19-Oct-09
1714 ----------------------
1715
1716 1. The table for translating pcre_compile() error codes into POSIX error codes
1717 was out-of-date, and there was no check on the pcre_compile() error code
1718 being within the table. This could lead to an OK return being given in
1719 error.
1720
1721 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1722 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1723 in a Windows environment.
1724
1725 3. The pcregrep --count option prints the count for each file even when it is
1726 zero, as does GNU grep. However, pcregrep was also printing all files when
1727 --files-with-matches was added. Now, when both options are given, it prints
1728 counts only for those files that have at least one match. (GNU grep just
1729 prints the file name in this circumstance, but including the count seems
1730 more useful - otherwise, why use --count?) Also ensured that the
1731 combination -clh just lists non-zero counts, with no names.
1732
1733 4. The long form of the pcregrep -F option was incorrectly implemented as
1734 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1735 but it seems right to fix it, and I didn't think it was worth preserving
1736 the old behaviour.
1737
1738 5. The command line items --regex=pattern and --regexp=pattern were not
1739 recognized by pcregrep, which required --regex pattern or --regexp pattern
1740 (with a space rather than an '='). The man page documented the '=' forms,
1741 which are compatible with GNU grep; these now work.
1742
1743 6. No libpcreposix.pc file was created for pkg-config; there was just
1744 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1745
1746 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1747 when UCP support is not needed, by modifying the Python script that
1748 generates it from Unicode data files. This should not matter if the module
1749 is correctly used as a library, but I received one complaint about 50K of
1750 unwanted data. My guess is that the person linked everything into his
1751 program rather than using a library. Anyway, it does no harm.
1752
1753 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1754 was a minimum greater than 1 for a wide character in a possessive
1755 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1756 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1757 character. Chaos in the form of incorrect output or a compiling loop could
1758 result.
1759
1760 9. The restrictions on what a pattern can contain when partial matching is
1761 requested for pcre_exec() have been removed. All patterns can now be
1762 partially matched by this function. In addition, if there are at least two
1763 slots in the offset vector, the offset of the earliest inspected character
1764 for the match and the offset of the end of the subject are set in them when
1765 PCRE_ERROR_PARTIAL is returned.
1766
1767 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1768 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1769 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1770 and may be more useful for multi-segment matching.
1771
1772 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1773 used to be given if ever the end of the subject was reached; now it is
1774 given only if matching could not proceed because another character was
1775 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1776 string "Z", which now yields "no match" instead of "partial match". In the
1777 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1778 final character ended with (*FAIL).
1779
1780 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1781 if the pattern had a "must contain" character that was already found in the
1782 earlier partial match, unless partial matching was again requested. For
1783 example, with the pattern /dog.(body)?/, the "must contain" character is
1784 "g". If the first part-match was for the string "dog", restarting with
1785 "sbody" failed. This bug has been fixed.
1786
1787 13. The string returned by pcre_dfa_exec() after a partial match has been
1788 changed so that it starts at the first inspected character rather than the
1789 first character of the match. This makes a difference only if the pattern
1790 starts with a lookbehind assertion or \b or \B (\K is not supported by
1791 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1792 matching functions compatible, and I think it's the right thing to do.
1793
1794 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1795 so that the demonstration program is easily available in environments where
1796 PCRE has not been installed from source.
1797
1798 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1799 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1800 library.
1801
1802 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1803 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1804 is not the first non-POSIX option to be added. Clearly some people find
1805 these options useful.
1806
1807 17. If a caller to the POSIX matching function regexec() passes a non-zero
1808 value for nmatch with a NULL value for pmatch, the value of
1809 nmatch is forced to zero.
1810
1811 18. RunGrepTest did not have a test for the availability of the -u option of
1812 the diff command, as RunTest does. It now checks in the same way as
1813 RunTest, and also checks for the -b option.
1814
1815 19. If an odd number of negated classes containing just a single character
1816 interposed, within parentheses, between a forward reference to a named
1817 subpattern and the definition of the subpattern, compilation crashed with
1818 an internal error, complaining that it could not find the referenced
1819 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1820 [The bug was that it was starting one character too far in when skipping
1821 over the character class, thus treating the ] as data rather than
1822 terminating the class. This meant it could skip too much.]
1823
1824 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1825 /g option in pcretest when the pattern contains \K, which makes it possible
1826 to have an empty string match not at the start, even when the pattern is
1827 anchored. Updated pcretest and pcredemo to use this option.
1828
1829 21. If the maximum number of capturing subpatterns in a recursion was greater
1830 than the maximum at the outer level, the higher number was returned, but
1831 with unset values at the outer level. The correct (outer level) value is
1832 now given.
1833
1834 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1835 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1836 make it do so. The string so far is captured, making this feature
1837 compatible with Perl.
1838
1839 23. The tests have been re-organized, adding tests 11 and 12, to make it
1840 possible to check the Perl 5.10 features against Perl 5.10.
1841
1842 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1843 pattern matches a fixed length string. PCRE did not allow this; now it
1844 does. Neither allows recursion.
1845
1846 25. I finally figured out how to implement a request to provide the minimum
1847 length of subject string that was needed in order to match a given pattern.
1848 (It was back references and recursion that I had previously got hung up
1849 on.) This code has now been added to pcre_study(); it finds a lower bound
1850 to the length of subject needed. It is not necessarily the greatest lower
1851 bound, but using it to avoid searching strings that are too short does give
1852 some useful speed-ups. The value is available to calling programs via
1853 pcre_fullinfo().
1854
1855 26. While implementing 25, I discovered to my embarrassment that pcretest had
1856 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1857 study optimizations had never been tested with that matching function.
1858 Oops. What is worse, even when it was passed study data, there was a bug in
1859 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1860 were also very few tests of studied patterns with pcre_dfa_exec().
1861
1862 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1863 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1864 on the other side of the coin, they are no longer allowed to have different
1865 names, because these cannot be distinguished in PCRE, and this has caused
1866 confusion. (This is a difference from Perl.)
1867
1868 28. When duplicate subpattern names are present (necessarily with different
1869 numbers, as required by 27 above), and a test is made by name in a
1870 conditional pattern, either for a subpattern having been matched, or for
1871 recursion in such a pattern, all the associated numbered subpatterns are
1872 tested, and the overall condition is true if the condition is true for any
1873 one of them. This is the way Perl works, and is also more like the way
1874 testing by number works.
1875
1876
1877 Version 7.9 11-Apr-09
1878 ---------------------
1879
1880 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1881 (pcretest), all targets were linked against these libraries. This included
1882 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1883 libraries. This caused unwanted dependencies to be created. This problem
1884 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1885 pcretest is linked with readline.
1886
1887 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1888 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1889 moved outside it again, because FALSE and TRUE are already defined in AIX,
1890 but BOOL is not.
1891
1892 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1893 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1894
1895 4. The pcregrep documentation said spaces were inserted as well as colons (or
1896 hyphens) following file names and line numbers when outputting matching
1897 lines. This is not true; no spaces are inserted. I have also clarified the
1898 wording for the --colour (or --color) option.
1899
1900 5. In pcregrep, when --colour was used with -o, the list of matching strings
1901 was not coloured; this is different to GNU grep, so I have changed it to be
1902 the same.
1903
1904 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1905 each matching line was coloured. Now it goes on to look for further matches
1906 of any of the test patterns, which is the same behaviour as GNU grep.
1907
1908 7. A pattern that could match an empty string could cause pcregrep to loop; it
1909 doesn't make sense to accept an empty string match in pcregrep, so I have
1910 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1911 seems to be how GNU grep behaves. [But see later change 40 for release
1912 8.33.]
1913
1914 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1915 start or after a newline", because the conditional assertion was not being
1916 correctly handled. The rule now is that both the assertion and what follows
1917 in the first alternative must satisfy the test.
1918
1919 9. If auto-callout was enabled in a pattern with a conditional group whose
1920 condition was an assertion, PCRE could crash during matching, both with
1921 pcre_exec() and pcre_dfa_exec().
1922
1923 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1924 used for matching.
1925
1926 11. Unicode property support in character classes was not working for
1927 characters (bytes) greater than 127 when not in UTF-8 mode.
1928
1929 12. Added the -M command line option to pcretest.
1930
1931 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1932
1933 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1934
1935 16. Added comments and documentation about mis-use of no_arg in the C++
1936 wrapper.
1937
1938 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1939 from Martin Jerabek that uses macro names for all relevant character and
1940 string constants.
1941
1942 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1943 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1944 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1945 these, but not everybody uses configure.
1946
1947 19. A conditional group that had only one branch was not being correctly
1948 recognized as an item that could match an empty string. This meant that an
1949 enclosing group might also not be so recognized, causing infinite looping
1950 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1951 with the subject "ab", where knowledge that the repeated group can match
1952 nothing is needed in order to break the loop.
1953
1954 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1955 exec(), but without supplying a callout function, matching went wrong.
1956
1957 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1958 leak if the size of the offset vector was greater than 30. When the vector
1959 is smaller, the saved offsets during recursion go onto a local stack
1960 vector, but for larger vectors malloc() is used. It was failing to free
1961 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1962 error, in fact).
1963
1964 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1965 heapframe that is used only when UTF-8 support is enabled. This caused no
1966 problem, but was untidy.
1967
1968 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1969 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1970 included within another project.
1971
1972 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1973 slightly modified by me:
1974
1975 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1976 not building pcregrep.
1977
1978 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1979 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1980
1981 25. Forward references, both numeric and by name, in patterns that made use of
1982 duplicate group numbers, could behave incorrectly or give incorrect errors,
1983 because when scanning forward to find the reference group, PCRE was not
1984 taking into account the duplicate group numbers. A pattern such as
1985 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1986
1987 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1988 the feature of a custom pointer more persuasive (as requested by a user).
1989
1990 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1991 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1992 pre-defined.
1993
1994 28. Added support for (*UTF8) at the start of a pattern.
1995
1996 29. Arrange for flags added by the "release type" setting in CMake to be shown
1997 in the configuration summary.
1998
1999
2000 Version 7.8 05-Sep-08
2001 ---------------------
2002
2003 1. Replaced UCP searching code with optimized version as implemented for Ad
2004 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2005 stage table and inline lookup instead of a function, giving speed ups of 2
2006 to 5 times on some simple patterns that I tested. Permission was given to
2007 distribute the MultiStage2.py script that generates the tables (it's not in
2008 the tarball, but is in the Subversion repository).
2009
2010 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2011 scripts.
2012
2013 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2014 a group with a zero qualifier. The result of the study could be incorrect,
2015 or the function might crash, depending on the pattern.
2016
2017 4. Caseless matching was not working for non-ASCII characters in back
2018 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2019 It now works when Unicode Property Support is available.
2020
2021 5. In pcretest, an escape such as \x{de} in the data was always generating
2022 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2023 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2024 truncation.
2025
2026 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2027
2028 7. Added two (int) casts to pcregrep when printing the difference of two
2029 pointers, in case they are 64-bit values.
2030
2031 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2032 test 2 if it fails.
2033
2034 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2035 and a #define of that name to empty if it is not externally set. This is to
2036 allow users of MSVC to set it if necessary.
2037
2038 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2039 the convenience functions in the pcre_get.c source file.
2040
2041 11. An option change at the start of a pattern that had top-level alternatives
2042 could cause overwriting and/or a crash. This command provoked a crash in
2043 some environments:
2044
2045 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2046
2047 This potential security problem was recorded as CVE-2008-2371.
2048
2049 12. For a pattern where the match had to start at the beginning or immediately
2050 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2051 pcre_dfa_exec() could read past the end of the passed subject if there was
2052 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2053 pcretest so that it places the subject at the end of its malloc-ed buffer.
2054
2055 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2056 exec() might read past the end of the data buffer in UTF-8 mode.
2057
2058 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2059 the data contained the byte 0x85 as part of a UTF-8 character within its
2060 first line. This applied both to normal and DFA matching.
2061
2062 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2063 /^[^d]*?$/8 failed to match "abc".
2064
2065 16. Added a missing copyright notice to pcrecpp_internal.h.
2066
2067 17. Make it more clear in the documentation that values returned from
2068 pcre_exec() in ovector are byte offsets, not character counts.
2069
2070 18. Tidied a few places to stop certain compilers from issuing warnings.
2071
2072 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2073 supplied by Stefan Weber. I made a further small update for 7.8 because
2074 there is a change of source arrangements: the pcre_searchfuncs.c module is
2075 replaced by pcre_ucd.c.
2076
2077
2078 Version 7.7 07-May-08
2079 ---------------------
2080
2081 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2082 a string to a long long, pretend we don't even have a long long." This is
2083 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2084
2085 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2086 pre-7.6 versions, which defined a global no_arg variable instead of putting
2087 it in the RE class. (See also #8 below.)
2088
2089 3. Remove a line of dead code, identified by coverity and reported by Nuno
2090 Lopes.
2091
2092 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2093
2094 (1) The include/exclude patterns were being applied to the whole pathnames
2095 of files, instead of just to the final components.
2096
2097 (2) If there was more than one level of directory, the subdirectories were
2098 skipped unless they satisfied the include/exclude conditions. This is
2099 inconsistent with GNU grep (and could even be seen as contrary to the
2100 pcregrep specification - which I improved to make it absolutely clear).
2101 The action now is always to scan all levels of directory, and just
2102 apply the include/exclude patterns to regular files.
2103
2104 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2105 --exclude_dir in the tests to avoid scanning .svn directories.
2106
2107 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2108 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2109 doesn't support NULs in patterns.
2110
2111 7. Added some missing "const"s to declarations of static tables in
2112 pcre_compile.c and pcre_dfa_exec.c.
2113
2114 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2115 caused by fix #2 above. (Subsequently also a second patch to fix the
2116 first patch. And a third patch - this was a messy problem.)
2117
2118 9. Applied Craig's patch to remove the use of push_back().
2119
2120 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2121 matching function regexec().
2122
2123 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2124 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2125 references. PCRE supports relative numbers with this syntax (I don't think
2126 Oniguruma does).
2127
2128 12. Previously, a group with a zero repeat such as (...){0} was completely
2129 omitted from the compiled regex. However, this means that if the group
2130 was called as a subroutine from elsewhere in the pattern, things went wrong
2131 (an internal error was given). Such groups are now left in the compiled
2132 pattern, with a new opcode that causes them to be skipped at execution
2133 time.
2134
2135 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2136 to the way PCRE behaves:
2137
2138 (a) A lone ] character is dis-allowed (Perl treats it as data).
2139
2140 (b) A back reference to an unmatched subpattern matches an empty string
2141 (Perl fails the current match path).
2142
2143 (c) A data ] in a character class must be notated as \] because if the
2144 first data character in a class is ], it defines an empty class. (In
2145 Perl it is not possible to have an empty class.) The empty class []
2146 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2147 The negative empty class [^] matches any one character, independently
2148 of the DOTALL setting.
2149
2150 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2151 non-existent subpattern following a character class starting with ']' and
2152 containing () gave an internal compiling error instead of "reference to
2153 non-existent subpattern". Fortunately, when the pattern did exist, the
2154 compiled code was correct. (When scanning forwards to check for the
2155 existence of the subpattern, it was treating the data ']' as terminating
2156 the class, so got the count wrong. When actually compiling, the reference
2157 was subsequently set up correctly.)
2158
2159 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2160 it was being rejected as not supported by pcre_dfa_exec(), even though
2161 other assertions are supported. I have made pcre_dfa_exec() support
2162 (*FAIL).
2163
2164 16. The implementation of 13c above involved the invention of a new opcode,
2165 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2166 cannot be changed at match time, I realized I could make a small
2167 improvement to matching performance by compiling OP_ALLANY instead of
2168 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2169 on the OP_ANY path.
2170
2171 17. Compiling pcretest on Windows with readline support failed without the
2172 following two fixes: (1) Make the unistd.h include conditional on
2173 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2174
2175 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2176 ncurses library to be included for pcretest when ReadLine support is
2177 requested, but also to allow for it to be overridden. This patch came from
2178 Daniel Bergström.
2179
2180 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2181 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2182 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2183 spotting this.
2184
2185
2186 Version 7.6 28-Jan-08
2187 ---------------------
2188
2189 1. A character class containing a very large number of characters with
2190 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2191 overflow.
2192
2193 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2194 HAVE_LONG_LONG is not defined.
2195
2196 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2197 bring them up to date and include new features. This patch includes:
2198
2199 - Fixed PH's badly added libz and libbz2 support.
2200 - Fixed a problem with static linking.
2201 - Added pcredemo. [But later removed - see 7 below.]
2202 - Fixed dftables problem and added an option.
2203 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2204 HAVE_LONG_LONG.
2205 - Added readline support for pcretest.
2206 - Added an listing of the option settings after cmake has run.
2207
2208 4. A user submitted a patch to Makefile that makes it easy to create
2209 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2210 Makefile.am that cause it to include this special target, without
2211 affecting anything else. Note that the same mingw target plus all
2212 the other distribution libraries and programs are now supported
2213 when configuring with CMake (see 6 below) instead of with
2214 Configure/Make.
2215
2216 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2217 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2218 exported in the Windows port". It has not yet been confirmed that the patch
2219 solves the problem, but it does no harm.
2220
2221 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2222 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2223 with CMake, and also correct the comment about stack recursion.
2224
2225 7. Remove the automatic building of pcredemo from the ./configure system and
2226 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2227 of a program that users should build themselves after PCRE is installed, so
2228 building it automatically is not really right. What is more, it gave
2229 trouble in some build environments.
2230
2231 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2232
2233
2234 Version 7.5 10-Jan-08
2235 ---------------------
2236
2237 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2238 values in parens when parsing an RE using the C++ wrapper."
2239
2240 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2241 Characters greater than 255 were excluded from the class instead of being
2242 included.
2243
2244 3. The same bug as (2) above applied to negated POSIX classes such as
2245 [:^space:].
2246
2247 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2248 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2249 I have changed it.
2250
2251 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2252 first named subpattern) and a construct such as (?&a) would reference the
2253 first named subpattern whose name started with "a" (in other words, the
2254 length check was missing). Both these problems are fixed. "Subpattern name
2255 expected" is now given for (?&) (a zero-length name), and this patch also
2256 makes it give the same error for \k'' (previously it complained that that
2257 was a reference to a non-existent subpattern).
2258
2259 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2260 this is right because (?- can be followed by option settings as well as by
2261 digits. I have, however, made the messages clearer.
2262
2263 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2264 than the number used in the conditional) now cause a compile-time error.
2265 This is actually not compatible with Perl, which accepts such patterns, but
2266 treats the conditional as always being FALSE (as PCRE used to), but it
2267 seems to me that giving a diagnostic is better.
2268
2269 8. Change "alphameric" to the more common word "alphanumeric" in comments
2270 and messages.
2271
2272 9. Fix two occurrences of "backslash" in comments that should have been
2273 "backspace".
2274
2275 10. Remove two redundant lines of code that can never be obeyed (their function
2276 was moved elsewhere).
2277
2278 11. The program that makes PCRE's Unicode character property table had a bug
2279 which caused it to generate incorrect table entries for sequences of
2280 characters that have the same character type, but are in different scripts.
2281 It amalgamated them into a single range, with the script of the first of
2282 them. In other words, some characters were in the wrong script. There were
2283 thirteen such cases, affecting characters in the following ranges:
2284
2285 U+002b0 - U+002c1
2286 U+0060c - U+0060d
2287 U+0061e - U+00612
2288 U+0064b - U+0065e
2289 U+0074d - U+0076d
2290 U+01800 - U+01805
2291 U+01d00 - U+01d77
2292 U+01d9b - U+01dbf
2293 U+0200b - U+0200f
2294 U+030fc - U+030fe
2295 U+03260 - U+0327f
2296 U+0fb46 - U+0fbb1
2297 U+10450 - U+1049d
2298
2299 12. The -o option (show only the matching part of a line) for pcregrep was not
2300 compatible with GNU grep in that, if there was more than one match in a
2301 line, it showed only the first of them. It now behaves in the same way as
2302 GNU grep.
2303
2304 13. If the -o and -v options were combined for pcregrep, it printed a blank
2305 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2306 does the same. The return code can be used to tell if there were any
2307 non-matching lines.
2308
2309 14. Added --file-offsets and --line-offsets to pcregrep.
2310
2311 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2312 infinitely looping recursion. The bug was that positive lookaheads were not
2313 being skipped when checking for a possible empty match (negative lookaheads
2314 and both kinds of lookbehind were skipped).
2315
2316 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2317 inclusion of <windows.h> to before rather than after the definition of
2318 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2319
2320 17. Specifying a possessive quantifier with a specific limit for a Unicode
2321 character property caused pcre_compile() to compile bad code, which led at
2322 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2323 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2324 caused the error; without that there was no problem.
2325
2326 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2327
2328 19. Added --enable-pcretest-libreadline.
2329
2330 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2331 RE::GlobalReplace(). As a result, the number of replacements returned was
2332 double what it should be. I removed one of the increments, but Craig sent a
2333 later patch that removed the other one (the right fix) and added unit tests
2334 that check the return values (which was not done before).
2335
2336 21. Several CMake things:
2337
2338 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2339 the names libpcre and libpcreposix, not just pcre and pcreposix.
2340
2341 (2) The above change means that pcretest and pcregrep are now correctly
2342 linked with the newly-built libraries, not previously installed ones.
2343
2344 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2345
2346 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2347 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2348 UTF-8 newline character). The key issue is that the pattern starts .*;
2349 this means that the match must be either at the beginning, or after a
2350 newline. The bug was in the code for advancing after a failed match and
2351 checking that the new position followed a newline. It was not taking
2352 account of UTF-8 characters correctly.
2353
2354 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2355 character classes. PCRE was not treating the sequence [:...:] as a
2356 character class unless the ... were all letters. Perl, however, seems to
2357 allow any characters between [: and :], though of course it rejects as
2358 unknown any "names" that contain non-letters, because all the known class
2359 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2360 for example, whereas PCRE did not - it did not recognize a POSIX character
2361 class. This seemed a bit dangerous, so the code has been changed to be
2362 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2363 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2364 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2365 Perl does, and where it didn't before.
2366
2367 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2368 Windows environments %n is disabled by default.
2369
2370
2371 Version 7.4 21-Sep-07
2372 ---------------------
2373
2374 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2375 means that a class such as [\s] counted as "explicit reference to CR or
2376 LF". That isn't really right - the whole point of the change was to try to
2377 help when there was an actual mention of one of the two characters. So now
2378 the change happens only if \r or \n (or a literal CR or LF) character is
2379 encountered.
2380
2381 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2382 of both had grown to the point where there were only 3 bits left.
2383 Fortunately, there was spare space in the data structure, and so I have
2384 moved the internal flags into a new 16-bit field to free up more option
2385 bits.
2386
2387 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2388 but did not set the internal JCHANGED flag - either of these is enough to
2389 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2390 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2391 start sets both bits.
2392
2393 4. Added options (at build time, compile time, exec time) to change \R from
2394 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2395
2396 5. doc/pcresyntax.html was missing from the distribution.
2397
2398 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2399 compatibility, even though it is no longer used.
2400
2401 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2402 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2403 windows.h file is present (where different names are used). [This was
2404 reversed later after testing - see 16 below.]
2405
2406 8. Changed all #include <config.h> to #include "config.h". There were also
2407 some further <pcre.h> cases that I changed to "pcre.h".
2408
2409 9. When pcregrep was used with the --colour option, it missed the line ending
2410 sequence off the lines that it output.
2411
2412 10. It was pointed out to me that arrays of string pointers cause lots of
2413 relocations when a shared library is dynamically loaded. A technique of
2414 using a single long string with a table of offsets can drastically reduce
2415 these. I have refactored PCRE in four places to do this. The result is
2416 dramatic:
2417
2418 Originally: 290
2419 After changing UCP table: 187
2420 After changing error message table: 43
2421 After changing table of "verbs" 36
2422 After changing table of Posix names 22
2423
2424 Thanks to the folks working on Gregex for glib for this insight.
2425
2426 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2427 unicode-properties was also set.
2428
2429 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2430
2431 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2432 checked only for CRLF.
2433
2434 14. Added casts to pcretest.c to avoid compiler warnings.
2435
2436 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2437
2438 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2439 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2440 entirely. This removes changes made in 7 above.
2441
2442 17. The CMake files have been updated, and there is now more information about
2443 building with CMake in the NON-UNIX-USE document.
2444
2445
2446 Version 7.3 28-Aug-07
2447 ---------------------
2448
2449 1. In the rejigging of the build system that eventually resulted in 7.1, the
2450 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2451 brackets there is not right, since it causes compilers to look for an
2452 installed pcre.h, not the version that is in the source that is being
2453 compiled (which of course may be different). I have changed it back to:
2454
2455 #include "pcre.h"
2456
2457 I have a vague recollection that the change was concerned with compiling in
2458 different directories, but in the new build system, that is taken care of
2459 by the VPATH setting the Makefile.
2460
2461 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2462 when the subject happened to end in the byte 0x85 (e.g. if the last
2463 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2464 characters but of course it shouldn't be taken as a newline when it is part
2465 of another character. The bug was that, for an unlimited repeat of . in
2466 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2467 characters when looking for a newline.
2468
2469 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2470
2471 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2472 in debug output.
2473
2474 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2475 long printing in the pcrecpp unittest when running under MinGW.
2476
2477 6. ESC_K was left out of the EBCDIC table.
2478
2479 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2480 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2481 limit also applies to "virtual nesting" when a pattern is recursive, and in
2482 this case 1000 isn't so big. I have been able to remove this limit at the
2483 expense of backing off one optimization in certain circumstances. Normally,
2484 when pcre_exec() would call its internal match() function recursively and
2485 immediately return the result unconditionally, it uses a "tail recursion"
2486 feature to save stack. However, when a subpattern that can match an empty
2487 string has an unlimited repetition quantifier, it no longer makes this
2488 optimization. That gives it a stack frame in which to save the data for
2489 checking that an empty string has been matched. Previously this was taken
2490 from the 1000-entry workspace that had been reserved. So now there is no
2491 explicit limit, but more stack is used.
2492
2493 8. Applied Daniel's patches to solve problems with the import/export magic
2494 syntax that is required for Windows, and which was going wrong for the
2495 pcreposix and pcrecpp parts of the library. These were overlooked when this
2496 problem was solved for the main library.
2497
2498 9. There were some crude static tests to avoid integer overflow when computing
2499 the size of patterns that contain repeated groups with explicit upper
2500 limits. As the maximum quantifier is 65535, the maximum group length was
2501 set at 30,000 so that the product of these two numbers did not overflow a
2502 32-bit integer. However, it turns out that people want to use groups that
2503 are longer than 30,000 bytes (though not repeat them that many times).
2504 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2505 made it possible to implement the integer overflow checks in a much more
2506 dynamic way, which I have now done. The artificial limitation on group
2507 length has been removed - we now have only the limit on the total length of
2508 the compiled pattern, which depends on the LINK_SIZE setting.
2509
2510 10. Fixed a bug in the documentation for get/copy named substring when
2511 duplicate names are permitted. If none of the named substrings are set, the
2512 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2513 empty string.
2514
2515 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2516 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2517 because the ] is interpreted as the first data character and the
2518 terminating ] is not found. PCRE has been made compatible with Perl in this
2519 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2520 cause memory overwriting.
2521
2522 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2523 string has been matched (to stop an infinite loop). It was not recognizing
2524 a conditional subpattern that could match an empty string if that
2525 subpattern was within another subpattern. For example, it looped when
2526 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2527 condition was not nested. This bug has been fixed.
2528
2529 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2530 past the start of the subject in the presence of bytes with the top bit
2531 set, for example "\x8aBCD".
2532
2533 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2534 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2535
2536 14. Optimized (?!) to (*FAIL).
2537
2538 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2539 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2540 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2541 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2542 does: it's just the validity check that is more restrictive.
2543
2544 16. Inserted checks for integer overflows during escape sequence (backslash)
2545 processing, and also fixed erroneous offset values for syntax errors during
2546 backslash processing.
2547
2548 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2549 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2550
2551 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2552 caused an overrun.
2553
2554 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2555 something other than just ASCII characters) inside a group that had an
2556 unlimited repeat caused a loop at compile time (while checking to see
2557 whether the group could match an empty string).
2558
2559 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2560 [\P{Any}] did so. (Error in the code for printing property names.)
2561
2562 21. An orphan \E inside a character class could cause a crash.
2563
2564 22. A repeated capturing bracket such as (A)? could cause a wild memory
2565 reference during compilation.
2566
2567 23. There are several functions in pcre_compile() that scan along a compiled
2568 expression for various reasons (e.g. to see if it's fixed length for look
2569 behind). There were bugs in these functions when a repeated \p or \P was
2570 present in the pattern. These operators have additional parameters compared
2571 with \d, etc, and these were not being taken into account when moving along
2572 the compiled data. Specifically:
2573
2574 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2575 length.
2576
2577 (b) An item such as \pL+ within a repeated group could cause crashes or
2578 loops.
2579
2580 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2581 "reference to non-existent subpattern" error.
2582
2583 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2584
2585 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2586 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2587
2588 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2589
2590 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2591 character were causing crashes (broken optimization).
2592
2593 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2594 \p or \P) caused a compile-time loop.
2595
2596 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2597 break. For example, the unstudied pattern [\r\n]A does not match the string
2598 "\r\nA" because change 7.0/46 below moves the current point on by two
2599 characters after failing to match at the start. However, the pattern \nA
2600 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2601 the same is true. There doesn't seem any very clean way out of this, but
2602 what I have chosen to do makes the common cases work: PCRE now takes note
2603 of whether there can be an explicit match for \r or \n anywhere in the
2604 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2605 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2606 pattern has explicit CR or LF references.
2607
2608 29. Added (*CR) etc for changing newline setting at start of pattern.
2609
2610
2611 Version 7.2 19-Jun-07
2612 ---------------------
2613
2614 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2615 which is apparently normally available under Windows.
2616
2617 2. Re-jig the pcregrep tests with different newline settings in an attempt
2618 to make them independent of the local environment's newline setting.
2619
2620 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2621
2622 4. Some of the "internals" tests were previously cut out when the link size
2623 was not 2, because the output contained actual offsets. The recent new
2624 "Z" feature of pcretest means that these can be cut out, making the tests
2625 usable with all link sizes.
2626
2627 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2628 stack recursion. This gives a massive performance boost under BSD, but just
2629 a small improvement under Linux. However, it saves one field in the frame
2630 in all cases.
2631
2632 6. Added more features from the forthcoming Perl 5.10:
2633
2634 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2635 recursion call. It refers to the nth most recently opened parentheses.
2636
2637 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2638 to be opened parentheses.
2639
2640 (c) Conditions that refer to capturing parentheses can be specified
2641 relatively, for example, (?(-2)... or (?(+3)...
2642
2643 (d) \K resets the start of the current match so that everything before
2644 is not part of it.
2645
2646 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2647
2648 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2649 reference syntax.
2650
2651 (g) (?| introduces a group in which the numbering of parentheses in each
2652 alternative starts with the same number.
2653
2654 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2655
2656 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2657 PCRE_INFO_JCHANGED.
2658
2659 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2660 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2661 for detecting groups that can match an empty string.
2662
2663 9. A pattern with a very large number of alternatives (more than several
2664 hundred) was running out of internal workspace during the pre-compile
2665 phase, where pcre_compile() figures out how much memory will be needed. A
2666 bit of new cunning has reduced the workspace needed for groups with
2667 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2668 workspace instead of running out of the 4096 that are available.
2669
2670 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2671
2672 11. Applied patch from Google to remove an optimization that didn't quite work.
2673 The report of the bug said:
2674
2675 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2676 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2677 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2678
2679 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2680 it matched the wrong number of bytes.
2681
2682
2683 Version 7.1 24-Apr-07
2684 ---------------------
2685
2686 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2687 that is more "standard", making use of automake and other Autotools. There
2688 is some re-arrangement of the files and adjustment of comments consequent
2689 on this.
2690
2691 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2692 for recursive directory scanning broke on some systems because the files
2693 are not scanned in any specific order and on different systems the order
2694 was different. A call to "sort" has been inserted into RunGrepTest for the
2695 approprate test as a short-term fix. In the longer term there may be an
2696 alternative.
2697
2698 3. I had an email from Eric Raymond about problems translating some of PCRE's
2699 man pages to HTML (despite the fact that I distribute HTML pages, some
2700 people do their own conversions for various reasons). The problems
2701 concerned the use of low-level troff macros .br and .in. I have therefore
2702 removed all such uses from the man pages (some were redundant, some could
2703 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2704 HTML has been updated to handle .nf/.fi and to complain if it encounters
2705 .br or .in.
2706
2707 4. Updated comments in configure.ac that get placed in config.h.in and also
2708 arranged for config.h to be included in the distribution, with the name
2709 config.h.generic, for the benefit of those who have to compile without
2710 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2711
2712 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2713 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2714 makevp.bat for the current PCRE, using the additional files
2715 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2716
2717 6. A Windows user reported a minor discrepancy with test 2, which turned out
2718 to be caused by a trailing space on an input line that had got lost in his
2719 copy. The trailing space was an accident, so I've just removed it.
2720
2721 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2722 that is needed.
2723
2724 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2725 as "const" (a) because they are and (b) because it helps the PHP
2726 maintainers who have recently made a script to detect big data structures
2727 in the php code that should be moved to the .rodata section. I remembered
2728 to update Builducptable as well, so it won't revert if ucptable.h is ever
2729 re-created.
2730
2731 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2732 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2733 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2734 support is not required. This saves 1.5-2K of code, which is important in
2735 some applications.
2736
2737 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2738 so as not to refer to the tables, even though these functions will never be
2739 called when UTF-8 support is disabled. Otherwise there are problems with a
2740 shared library.
2741
2742 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2743
2744 (a) It was defining its arguments as char * instead of void *.
2745
2746 (b) It was assuming that all moves were upwards in memory; this was true
2747 a long time ago when I wrote it, but is no longer the case.
2748
2749 The emulated memove() is provided for those environments that have neither
2750 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2751 is clearly not the case, as these two bugs were recently reported.
2752
2753 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2754 and Detrail to create the HTML documentation, the .txt form of the man
2755 pages, and it removes trailing spaces from listed files. It also creates
2756 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2757 case, it wraps all the #defines with #ifndefs. This script should be run
2758 before "make dist".
2759
2760 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2761 with Unicode property support.
2762
2763 (a) For a maximizing quantifier, if the two different cases of the
2764 character were of different lengths in their UTF-8 codings (there are
2765 some cases like this - I found 11), and the matching function had to
2766 back up over a mixture of the two cases, it incorrectly assumed they
2767 were both the same length.
2768
2769 (b) When PCRE was configured to use the heap rather than the stack for
2770 recursion during matching, it was not correctly preserving the data for
2771 the other case of a UTF-8 character when checking ahead for a match
2772 while processing a minimizing repeat. If the check also involved
2773 matching a wide character, but failed, corruption could cause an
2774 erroneous result when trying to check for a repeat of the original
2775 character.
2776
2777 13. Some tidying changes to the testing mechanism:
2778
2779 (a) The RunTest script now detects the internal link size and whether there
2780 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2781 values substituted by "configure". (The RunGrepTest script already did
2782 this for UTF-8.) The configure.ac script no longer substitutes the
2783 relevant variables.
2784
2785 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2786 with length and offset values. This means that the output is different
2787 for different internal link sizes. Test 2 is skipped for link sizes
2788 other than 2 because of this, bypassing the problem. Unfortunately,
2789 there was also a test in test 3 (the locale tests) that used /B and
2790 failed for link sizes other than 2. Rather than cut the whole test out,
2791 I have added a new /Z option to pcretest that replaces the length and
2792 offset values with spaces. This is now used to make test 3 independent
2793 of link size. (Test 2 will be tidied up later.)
2794
2795 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2796 segmentation fault instead of returning the appropriate error message.
2797
2798 15. In multiline mode when the newline sequence was set to "any", the pattern
2799 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2800 This doesn't seem right; it now treats the CRLF combination as the line
2801 ending, and so does not match in that case. It's only a pattern such as ^$
2802 that would hit this one: something like ^ABC$ would have failed after \r
2803 and then tried again after \r\n.
2804
2805 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2806 in an attempt to make files that differ only in their line terminators
2807 compare equal. This works on Linux.
2808
2809 17. Under certain error circumstances pcregrep might try to free random memory
2810 as it exited. This is now fixed, thanks to valgrind.
2811
2812 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2813 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2814 was because its rules for how to advance for /g after matching an empty
2815 string at the end of a line did not allow for this case. They now check for
2816 it specially.
2817
2818 20. pcretest is supposed to handle patterns and data of any length, by
2819 extending its buffers when necessary. It was getting this wrong when the
2820 buffer for a data line had to be extended.
2821
2822 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2823 CRLF as a newline sequence.
2824
2825 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2826 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2827 I have nevertheless tidied it up.
2828
2829 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2830
2831 24. Added a man page for pcre-config.
2832
2833
2834 Version 7.0 19-Dec-06
2835 ---------------------
2836
2837 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2838 moving to gcc 4.1.1.
2839
2840 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2841 sys/time.h, which is documented as needed for this function. It doesn't
2842 seem to matter on Linux, but it showed up on some releases of OS X.
2843
2844 3. It seems that there are systems where bytes whose values are greater than
2845 127 match isprint() in the "C" locale. The "C" locale should be the
2846 default when a C program starts up. In most systems, only ASCII printing
2847 characters match isprint(). This difference caused the output from pcretest
2848 to vary, making some of the tests fail. I have changed pcretest so that:
2849
2850 (a) When it is outputting text in the compiled version of a pattern, bytes
2851 other than 32-126 are always shown as hex escapes.
2852
2853 (b) When it is outputting text that is a matched part of a subject string,
2854 it does the same, unless a different locale has been set for the match
2855 (using the /L modifier). In this case, it uses isprint() to decide.
2856
2857 4. Fixed a major bug that caused incorrect computation of the amount of memory
2858 required for a compiled pattern when options that changed within the
2859 pattern affected the logic of the preliminary scan that determines the
2860 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2861 that the computed length was too small. The symptoms of this bug were
2862 either the PCRE error "internal error: code overflow" from pcre_compile(),
2863 or a glibc crash with a message such as "pcretest: free(): invalid next
2864 size (fast)". Examples of patterns that provoked this bug (shown in
2865 pcretest format) are:
2866
2867 /(?-x: )/x
2868 /(?x)(?-x: \s*#\s*)/
2869 /((?i)[\x{c0}])/8
2870 /(?i:[\x{c0}])/8
2871
2872 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2873 is now done differently.
2874
2875 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2876 wrapper classes; (b) implement a new function in the C++ scanner that is
2877 more efficient than the old way of doing things because it avoids levels of
2878 recursion in the regex matching; (c) add a paragraph to the documentation
2879 for the FullMatch() function.
2880
2881 6. The escape sequence \n was being treated as whatever was defined as
2882 "newline". Not only was this contrary to the documentation, which states
2883 that \n is character 10 (hex 0A), but it also went horribly wrong when
2884 "newline" was defined as CRLF. This has been fixed.
2885
2886 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2887 was being set to -1 for the "end of line" case (supposedly a value that no
2888 character can have). Though this value is never used (the check for end of
2889 line is "zero bytes in current character"), it caused compiler complaints.
2890 I've changed it to 0xffffffff.
2891
2892 8. In pcre_version.c, the version string was being built by a sequence of
2893 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2894 string (as it is for production releases) called a macro with an empty
2895 argument. The C standard says the result of this is undefined. The gcc
2896 compiler treats it as an empty string (which was what was wanted) but it is
2897 reported that Visual C gives an error. The source has been hacked around to
2898 avoid this problem.
2899
2900 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2901 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2902 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2903 of them did).
2904
2905 10. Originally, pcretest opened its input and output without "b"; then I was
2906 told that "b" was needed in some environments, so it was added for release
2907 5.0 to both the input and output. (It makes no difference on Unix-like
2908 systems.) Later I was told that it is wrong for the input on Windows. I've
2909 now abstracted the modes into two macros, to make it easier to fiddle with
2910 them, and removed "b" from the input mode under Windows.
2911
2912 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2913
2914 12. Added -help and --help to pcretest as an official way of being reminded
2915 of the options.
2916
2917 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2918 and pcrecpp.cc because they annoy compilers at high warning levels.
2919
2920 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2921
2922 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2923 scripts are not C programs :-) and which was not noticed because it works
2924 on Linux.
2925
2926 16. pcretest is supposed to handle any length of pattern and data line (as one
2927 line or as a continued sequence of lines) by extending its input buffer if
2928 necessary. This feature was broken for very long pattern lines, leading to
2929 a string of junk being passed to pcre_compile() if the pattern was longer
2930 than about 50K.
2931
2932 17. I have done a major re-factoring of the way pcre_compile() computes the
2933 amount of memory needed for a compiled pattern. Previously, there was code
2934 that made a preliminary scan of the pattern in order to do this. That was
2935 OK when PCRE was new, but as the facilities have expanded, it has become
2936 harder and harder to keep it in step with the real compile phase, and there
2937 have been a number of bugs (see for example, 4 above). I have now found a
2938 cunning way of running the real compile function in a "fake" mode that
2939 enables it to compute how much memory it would need, while actually only
2940 ever using a few hundred bytes of working memory and without too many
2941 tests of the mode. This should make future maintenance and development
2942 easier. A side effect of this work is that the limit of 200 on the nesting
2943 depth of parentheses has been removed (though this was never a serious
2944 limitation, I suspect). However, there is a downside: pcre_compile() now
2945 runs more slowly than before (30% or more, depending on the pattern). I
2946 hope this isn't a big issue. There is no effect on runtime performance.
2947
2948 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2949 newline (only possible for the last line of a file) and it was a
2950 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2951
2952 19. Added additional timing features to pcretest. (1) The -tm option now times
2953 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2954 separate command line item, by a number that specifies the number of
2955 repeats to use when timing. The default is 50000; this gives better
2956 precision, but takes uncomfortably long for very large patterns.
2957
2958 20. Extended pcre_study() to be more clever in cases where a branch of a
2959 subpattern has no definite first character. For example, (a*|b*)[cd] would
2960 previously give no result from pcre_study(). Now it recognizes that the
2961 first character must be a, b, c, or d.
2962
2963 21. There was an incorrect error "recursive call could loop indefinitely" if
2964 a subpattern (or the entire pattern) that was being tested for matching an
2965 empty string contained only one non-empty item after a nested subpattern.
2966 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2967 incorrectly, because the \d was being skipped in the check.
2968
2969 22. The pcretest program now has a new pattern option /B and a command line
2970 option -b, which is equivalent to adding /B to every pattern. This causes
2971 it to show the compiled bytecode, without the additional information that
2972 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2973 is the same as /B/I).
2974
2975 23. A new optimization is now able automatically to treat some sequences such
2976 as a*b as a*+b. More specifically, if something simple (such as a character
2977 or a simple class like \d) has an unlimited quantifier, and is followed by
2978 something that cannot possibly match the quantified thing, the quantifier
2979 is automatically "possessified".
2980
2981 24. A recursive reference to a subpattern whose number was greater than 39
2982 went wrong under certain circumstances in UTF-8 mode. This bug could also
2983 have affected the operation of pcre_study().
2984
2985 25. Realized that a little bit of performance could be had by replacing
2986 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2987
2988 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2989
2990 27. Possessive quantifiers such as a++ were previously implemented by turning
2991 them into atomic groups such as ($>a+). Now they have their own opcodes,
2992 which improves performance. This includes the automatically created ones
2993 from 23 above.
2994
2995 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2996 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2997 the first matched character to be a colon. This applied both to named and
2998 numbered groups.
2999
3000 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3001
3002 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3003 building PCRE on Windows easier, so I have included it in the distribution.
3004
3005 31. There is now a check in pcretest against a ridiculously large number being
3006 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3007 loop, the loop is abandoned.
3008
3009 32. Forward references to subpatterns in conditions such as (?(2)...) where
3010 subpattern 2 is defined later cause pcre_compile() to search forwards in
3011 the pattern for the relevant set of parentheses. This search went wrong
3012 when there were unescaped parentheses in a character class, parentheses
3013 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3014
3015 33. "Subroutine" calls and backreferences were previously restricted to
3016 referencing subpatterns earlier in the regex. This restriction has now
3017 been removed.
3018
3019 34. Added a number of extra features that are going to be in Perl 5.10. On the
3020 whole, these are just syntactic alternatives for features that PCRE had
3021 previously implemented using the Python syntax or my own invention. The
3022 other formats are all retained for compatibility.
3023
3024 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3025 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3026 also .NET compatible.
3027
3028 (b) A recursion or subroutine call to a named group can now be defined as
3029 (?&name) as well as (?P>name).
3030
3031 (c) A backreference to a named group can now be defined as \k<name> or
3032 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3033 5.10, are also .NET compatible.
3034
3035 (d) A conditional reference to a named group can now use the syntax
3036 (?(<name>) or (?('name') as well as (?(name).
3037
3038 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3039 groups (named and numbered) that are never evaluated inline, but can be
3040 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3041 is always false. There may be only one alternative in such a group.
3042
3043 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3044 as the simple (?(R). The condition is true only if the most recent
3045 recursion is that of the given number or name. It does not search out
3046 through the entire recursion stack.
3047
3048 (g) The escape \gN or \g{N} has been added, where N is a positive or
3049 negative number, specifying an absolute or relative reference.
3050
3051 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3052 some "unreachable code" warnings.
3053
3054 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3055 things, this adds five new scripts.
3056
3057 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3058 There were also incompatibilities regarding the handling of \Q..\E inside
3059 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3060 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3061
3062 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3063 matches an empty string, and forcibly breaks the loop. There were bugs in
3064 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3065 against aaaa the result was just "a" rather than "aaaa", for example. Two
3066 separate and independent bugs (that affected different cases) have been
3067 fixed.
3068
3069 39. Refactored the code to abolish the use of different opcodes for small
3070 capturing bracket numbers. This is a tidy that I avoided doing when I
3071 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3072 The new approach is not only tidier, it makes it possible to reduce the
3073 memory needed to fix the previous bug (38).
3074
3075 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3076 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3077 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3078 mode.
3079
3080 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3081 report.
3082
3083 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3084 copy construction and assignment in the C++ wrapper.
3085
3086 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3087 couple of bugs that could have given wrong results in the "--newline=crlf"
3088 case.
3089
3090 44. Added a number of casts and did some reorganization of signed/unsigned int
3091 variables following suggestions from Dair Grant. Also renamed the variable
3092 "this" as "item" because it is a C++ keyword.
3093
3094 45. Arranged for dftables to add
3095
3096 #include "pcre_internal.h"
3097
3098 to pcre_chartables.c because without it, gcc 4.x may remove the array
3099 definition from the final binary if PCRE is built into a static library and
3100 dead code stripping is activated.
3101
3102 46. For an unanchored pattern, if a match attempt fails at the start of a
3103 newline sequence, and the newline setting is CRLF or ANY, and the next two
3104 characters are CRLF, advance by two characters instead of one.
3105
3106
3107 Version 6.7 04-Jul-06
3108 ---------------------
3109
3110 1. In order to handle tests when input lines are enormously long, pcretest has
3111 been re-factored so that it automatically extends its buffers when
3112 necessary. The code is crude, but this _is_ just a test program. The
3113 default size has been increased from 32K to 50K.
3114
3115 2. The code in pcre_study() was using the value of the re argument before
3116 testing it for NULL. (Of course, in any sensible call of the function, it
3117 won't be NULL.)
3118
3119 3. The memmove() emulation function in pcre_internal.h, which is used on
3120 systems that lack both memmove() and bcopy() - that is, hardly ever -
3121 was missing a "static" storage class specifier.
3122
3123 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3124 containing an extended class (one that cannot be represented by a bitmap
3125 because it contains high-valued characters or Unicode property items, e.g.
3126 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3127 pattern, but PCRE should not loop if you do not (it no longer does).
3128 [Detail: two cases were found: (a) a repeated subpattern containing an
3129 extended class; (b) a recursive reference to a subpattern that followed a
3130 previous extended class. It wasn't skipping over the extended class
3131 correctly when UTF-8 mode was not set.]
3132
3133 5. A negated single-character class was not being recognized as fixed-length
3134 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3135 compile error "lookbehind assertion is not fixed length".
3136
3137 6. The RunPerlTest auxiliary script was showing an unexpected difference
3138 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3139 write a Perl script that can interpret lines of an input file either as
3140 byte characters or as UTF-8, which is what "perltest" was being required to
3141 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3142 can't do is switch easily at run time between having the "use utf8;" pragma
3143 or not. In the end, I fudged it by using the RunPerlTest script to insert
3144 "use utf8;" explicitly for the UTF-8 tests.
3145
3146 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3147 the end of the subject string, contrary to the documentation and to what
3148 Perl does. This was true of both matching functions. Now it matches only at
3149 the start of the subject and immediately after *internal* newlines.
3150
3151 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3152 a pointer to an int instead of a pointer to an unsigned long int. This
3153 caused problems on 64-bit systems.
3154
3155 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3156 instance of the 'standard' template library not being so standard".
3157
3158 10. There was no check on the number of named subpatterns nor the maximum
3159 length of a subpattern name. The product of these values is used to compute
3160 the size of the memory block for a compiled pattern. By supplying a very
3161 long subpattern name and a large number of named subpatterns, the size
3162 computation could be caused to overflow. This is now prevented by limiting
3163 the length of names to 32 characters, and the number of named subpatterns
3164 to 10,000.
3165
3166 11. Subpatterns that are repeated with specific counts have to be replicated in
3167 the compiled pattern. The size of memory for this was computed from the
3168 length of the subpattern and the repeat count. The latter is limited to
3169 65535, but there was no limit on the former, meaning that integer overflow
3170 could in principle occur. The compiled length of a repeated subpattern is
3171 now limited to 30,000 bytes in order to prevent this.
3172
3173 12. Added the optional facility to have named substrings with the same name.
3174
3175 13. Added the ability to use a named substring as a condition, using the
3176 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3177 are numbers (not recommended). Forward references are permitted.
3178
3179 14. Added forward references in named backreferences (if you see what I mean).
3180
3181 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3182 pattern could run off the end of the subject. For example, the pattern
3183 "(?s)(.{1,5})"8 did this with the subject "ab".
3184
3185 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3186 PCRE_CASELESS was set when matching characters that were quantified with ?
3187 or *.
3188
3189 17. A character class other than a single negated character that had a minimum
3190 but no maximum quantifier - for example [ab]{6,} - was not handled
3191 correctly by pce_dfa_exec(). It would match only one character.
3192
3193 18. A valid (though odd) pattern that looked like a POSIX character
3194 class but used an invalid character after [ (for example [[,abc,]]) caused
3195 pcre_compile() to give the error "Failed: internal error: code overflow" or
3196 in some cases to crash with a glibc free() error. This could even happen if
3197 the pattern terminated after [[ but there just happened to be a sequence of
3198 letters, a binary zero, and a closing ] in the memory that followed.
3199
3200 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3201 over the years. Originally (before any Unicode support), just the bottom 8
3202 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3203 output from "man perlunicode" includes this:
3204
3205 The regular expression compiler produces polymorphic opcodes. That
3206 is, the pattern adapts to the data and automatically switches to
3207 the Unicode character scheme when presented with Unicode data--or
3208 instead uses a traditional byte scheme when presented with byte
3209 data.
3210
3211 Sadly, a wide octal escape does not cause a switch, and in a string with
3212 no other multibyte characters, these octal escapes are treated as before.
3213 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3214 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3215 Unicode string.
3216
3217 I have not perpetrated such confusion in PCRE. Up till now, it took just
3218 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3219 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3220 translate to the appropriate multibyte character.
3221
3222 29. Applied some refactoring to reduce the number of warnings from Microsoft
3223 and Borland compilers. This has included removing the fudge introduced
3224 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3225 a warning about an unused variable.
3226
3227 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3228 characters since release 4.0, because Perl (from release 5.004) does not.
3229 [Or at least, is documented not to: some releases seem to be in conflict
3230 with the documentation.] However, when a pattern was studied with
3231 pcre_study() and all its branches started with \s, PCRE still included VT
3232 as a possible starting character. Of course, this did no harm; it just
3233 caused an unnecessary match attempt.
3234
3235 22. Removed a now-redundant internal flag bit that recorded the fact that case
3236 dependency changed within the pattern. This was once needed for "required
3237 byte" processing, but is no longer used. This recovers a now-scarce options
3238 bit. Also moved the least significant internal flag bit to the most-
3239 significant bit of the word, which was not previously used (hangover from
3240 the days when it was an int rather than a uint) to free up another bit for
3241 the future.
3242
3243 23. Added support for CRLF line endings as well as CR and LF. As well as the
3244 default being selectable at build time, it can now be changed at runtime
3245 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3246 specify that it is scanning data with non-default line endings.
3247
3248 24. Changed the definition of CXXLINK to make it agree with the definition of
3249 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3250
3251 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3252 recursions. This makes a big different to stack usage for some patterns.
3253
3254 26. If a subpattern containing a named recursion or subroutine reference such
3255 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3256 the space required for the compiled pattern went wrong and gave too small a
3257 value. Depending on the environment, this could lead to "Failed: internal
3258 error: code overflow at offset 49" or "glibc detected double free or
3259 corruption" errors.
3260
3261 27. Applied patches from Google (a) to support the new newline modes and (b) to
3262 advance over multibyte UTF-8 characters in GlobalReplace.
3263
3264 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3265 difference for some implementation of PCRE in some Windows version.
3266
3267 29. Added some extra testing facilities to pcretest:
3268
3269 \q<number> in a data line sets the "match limit" value
3270 \Q<number> in a data line sets the "match recursion limt" value
3271 -S <number> sets the stack size, where <number> is in megabytes
3272
3273 The -S option isn't available for Windows.
3274
3275
3276 Version 6.6 06-Feb-06
3277 ---------------------
3278
3279 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3280 in pcreposix.h. I have copied the definition from pcre.h.
3281
3282 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3283 because pcre.h is no longer a built file.
3284
3285 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3286 not normally included in the compiled code.
3287
3288
3289 Version 6.5 01-Feb-06
3290 ---------------------
3291
3292 1. When using the partial match feature with pcre_dfa_exec(), it was not
3293 anchoring the second and subsequent partial matches at the new starting
3294 point. This could lead to incorrect results. For example, with the pattern
3295 /1234/, partially matching against "123" and then "a4" gave a match.
3296
3297 2. Changes to pcregrep:
3298
3299 (a) All non-match returns from pcre_exec() were being treated as failures
3300 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3301 error message is output. Some extra information is given for the
3302 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3303 probably the only errors that are likely to be caused by users (by
3304 specifying a regex that has nested indefinite repeats, for instance).
3305 If there are more than 20 of these errors, pcregrep is abandoned.
3306
3307 (b) A binary zero was treated as data while matching, but terminated the
3308 output line if it was written out. This has been fixed: binary zeroes
3309 are now no different to any other data bytes.
3310
3311 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3312 used to set a locale for matching. The --locale=xxxx long option has
3313 been added (no short equivalent) to specify a locale explicitly on the
3314 pcregrep command, overriding the environment variables.
3315
3316 (d) When -B was used with -n, some line numbers in the output were one less
3317 than they should have been.
3318
3319 (e) Added the -o (--only-matching) option.
3320
3321 (f) If -A or -C was used with -c (count only), some lines of context were
3322 accidentally printed for the final match.
3323
3324 (g) Added the -H (--with-filename) option.
3325
3326 (h) The combination of options -rh failed to suppress file names for files
3327 that were found from directory arguments.
3328
3329 (i) Added the -D (--devices) and -d (--directories) options.
3330
3331 (j) Added the -F (--fixed-strings) option.
3332
3333 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3334
3335 (l) Added the --colo(u)r option.
3336
3337 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3338 is not present by default.
3339
3340 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3341 items such as (?R) or (?1), when the recursion could match a number of
3342 alternatives. If it matched one of the alternatives, but subsequently,
3343 outside the recursion, there was a failure, the code tried to back up into
3344 the recursion. However, because of the way PCRE is implemented, this is not
3345 possible, and the result was an incorrect result from the match.
3346
3347 In order to prevent this happening, the specification of recursion has
3348 been changed so that all such subpatterns are automatically treated as
3349 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3350
3351 4. I had overlooked the fact that, in some locales, there are characters for
3352 which isalpha() is true but neither isupper() nor islower() are true. In
3353 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3354 and ordfeminine) are like this. This affected the treatment of \w and \W
3355 when they appeared in character classes, but not when they appeared outside
3356 a character class. The bit map for "word" characters is now created
3357 separately from the results of isalnum() instead of just taking it from the
3358 upper, lower, and digit maps. (Plus the underscore character, of course.)
3359
3360 5. The above bug also affected the handling of POSIX character classes such as
3361 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3362 permanent tables. Instead, the bit maps for such a class were previously
3363 created as the appropriate unions of the upper, lower, and digit bitmaps.
3364 Now they are created by subtraction from the [[:word:]] class, which has
3365 its own bitmap.
3366
3367 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3368 It is created by subtracting the vertical space characters (\x09, \x0a,
3369 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3370 subtraction was done in the overall bitmap for a character class, meaning
3371 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3372 be recognized. This bug has been fixed.
3373
3374 7. Patches from the folks at Google:
3375
3376 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3377 real life, but is still worth protecting against".
3378
3379 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3380 regular expressions".
3381
3382 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3383 have it.
3384
3385 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3386 "configure" and the latter not, in order to fix a problem somebody had
3387 with compiling the Arg class on HP-UX.
3388
3389 (e) Improve the error-handling of the C++ wrapper a little bit.
3390
3391 (f) New tests for checking recursion limiting.
3392
3393 8. The pcre_memmove() function, which is used only if the environment does not
3394 have a standard memmove() function (and is therefore rarely compiled),
3395 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3396 returning a result (though PCRE never actually uses the result).
3397
3398 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3399 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3400 returned instead of calling malloc() with an overflowing number that would
3401 most likely cause subsequent chaos.
3402
3403 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3404
3405 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3406 with this option is matched, the nmatch and pmatch options of regexec() are
3407 ignored.
3408
3409 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3410 provided in case anyone wants to the the POSIX interface with UTF-8
3411 strings.
3412
3413 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3414 C++ linking (needed for some HP-UX environments).
3415
3416 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3417 (unused parameter) and in the pcre_printint() function (omitted "default"
3418 switch label when the default is to do nothing).
3419
3420 15. Added some code to make it possible, when PCRE is compiled as a C++
3421 library, to replace subject pointers for pcre_exec() with a smart pointer
3422 class, thus making it possible to process discontinuous strings.
3423
3424 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3425 much the same function. They were added by different people who were trying
3426 to make PCRE easy to compile on non-Unix systems. It has been suggested
3427 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3428 for compiling on Windows systems. I have therefore replaced it with
3429 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3430 defaults to "extern" for C or "extern C" for C++, which works fine on
3431 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3432 SCOPE with something explicit in config.h. In addition:
3433
3434 (a) pcreposix.h still had just "extern" instead of either of these macros;
3435 I have replaced it with PCRE_DATA_SCOPE.
3436
3437 (b) Functions such as _pcre_xclass(), which are internal to the library,
3438 but external in the C sense, all had PCRE_EXPORT in their definitions.
3439 This is apparently wrong for the Windows case, so I have removed it.
3440 (It makes no difference on Unix-like systems.)
3441
3442 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3443 of recursive calls to match(). This is different to MATCH_LIMIT because
3444 that limits the total number of calls to match(), not all of which increase
3445 the depth of recursion. Limiting the recursion depth limits the amount of
3446 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3447 when PCRE is compiled, and changed at run time. A patch from Google adds
3448 this functionality to the C++ interface.
3449
3450 18. Changes to the handling of Unicode character properties:
3451
3452 (a) Updated the table to Unicode 4.1.0.
3453
3454 (b) Recognize characters that are not in the table as "Cn" (undefined).
3455
3456 (c) I revised the way the table is implemented to a much improved format
3457 which includes recognition of ranges. It now supports the ranges that
3458 are defined in UnicodeData.txt, and it also amalgamates other
3459 characters into ranges. This has reduced the number of entries in the
3460 table from around 16,000 to around 3,000, thus reducing its size
3461 considerably. I realized I did not need to use a tree structure after
3462 all - a binary chop search is just as efficient. Having reduced the
3463 number of entries, I extended their size from 6 bytes to 8 bytes to
3464 allow for more data.
3465
3466 (d) Added support for Unicode script names via properties such as \p{Han}.
3467
3468 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3469 matching that character.
3470
3471 20. When matching a repeated Unicode property with a minimum greater than zero,
3472 (for example \pL{2,}), PCRE could look past the end of the subject if it
3473 reached it while seeking the minimum number of characters. This could
3474 happen only if some of the characters were more than one byte long, because
3475 there is a check for at least the minimum number of bytes.
3476
3477 21. Refactored the implementation of \p and \P so as to be more general, to
3478 allow for more different types of property in future. This has changed the
3479 compiled form incompatibly. Anybody with saved compiled patterns that use
3480 \p or \P will have to recompile them.
3481
3482 22. Added "Any" and "L&" to the supported property types.
3483
3484 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3485 but give a compile time error if the value is greater than 0xff.
3486
3487 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3488 accidentally not being installed or uninstalled.
3489
3490 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3491 made were to insert the current release number. This seemed silly, because
3492 it made things harder for people building PCRE on systems that don't run
3493 "configure". I have turned pcre.h into a distributed file, no longer built
3494 by "configure", with the version identification directly included. There is
3495 no longer a pcre.h.in file.
3496
3497 However, this change necessitated a change to the pcre-config script as
3498 well. It is built from pcre-config.in, and one of the substitutions was the
3499 release number. I have updated configure.ac so that ./configure now finds
3500 the release number by grepping pcre.h.
3501
3502 26. Added the ability to run the tests under valgrind.
3503
3504
3505 Version 6.4 05-Sep-05
3506 ---------------------
3507
3508 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3509 "--" to be printed when multiple files were scanned, even when none of the
3510 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3511 consider it to be a bug, and have restored the previous behaviour.
3512
3513 2. A couple of code tidies to get rid of compiler warnings.
3514
3515 3. The pcretest program used to cheat by referring to symbols in the library
3516 whose names begin with _pcre_. These are internal symbols that are not
3517 really supposed to be visible externally, and in some environments it is
3518 possible to suppress them. The cheating is now confined to including
3519 certain files from the library's source, which is a bit cleaner.
3520
3521 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3522 file's purpose clearer.
3523
3524 5. Reorganized pcre_ucp_findchar().
3525
3526
3527 Version 6.3 15-Aug-05
3528 ---------------------
3529
3530 1. The file libpcre.pc.in did not have general read permission in the tarball.
3531
3532 2. There were some problems when building without C++ support:
3533
3534 (a) If C++ support was not built, "make install" and "make test" still
3535 tried to test it.
3536
3537 (b) There were problems when the value of CXX was explicitly set. Some
3538 changes have been made to try to fix these, and ...
3539
3540 (c) --disable-cpp can now be used to explicitly disable C++ support.
3541
3542 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3543 backslash in a target when C++ was disabled. This confuses some
3544 versions of "make", apparently. Using an intermediate variable solves
3545 this. (Same for CPP_LOBJ.)
3546
3547 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3548 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3549 necessary on certain architectures.
3550
3551 4. Added a setting of -export-symbols-regex to the link command to remove
3552 those symbols that are exported in the C sense, but actually are local
3553 within the library, and not documented. Their names all begin with
3554 "_pcre_". This is not a perfect job, because (a) we have to except some
3555 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3556 available (and never for static libraries). I have made a note to try to
3557 find a way round (a) in the future.
3558
3559
3560 Version 6.2 01-Aug-05
3561 ---------------------
3562
3563 1. There was no test for integer overflow of quantifier values. A construction
3564 such as {1111111111111111} would give undefined results. What is worse, if
3565 a minimum quantifier for a parenthesized subpattern overflowed and became
3566 negative, the calculation of the memory size went wrong. This could have
3567 led to memory overwriting.
3568
3569 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3570
3571 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3572 operating environments where this matters.
3573
3574 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3575 PCRE options from within the C++ wrapper.
3576
3577 5. Named capturing subpatterns were not being correctly counted when a pattern
3578 was compiled. This caused two problems: (a) If there were more than 100
3579 such subpatterns, the calculation of the memory needed for the whole
3580 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3581 back references of the form \12, where the number was greater than 9, were
3582 not recognized as back references, even though there were sufficient
3583 previous subpatterns.
3584
3585 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3586 versions of gcc, e.g. 2.95.4.
3587
3588
3589 Version 6.1 21-Jun-05
3590 ---------------------
3591
3592 1. There was one reference to the variable "posix" in pcretest.c that was not
3593 surrounded by "#if !defined NOPOSIX".
3594
3595 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3596 the cross-check on the old pcre_info() function, for the benefit of the
3597 cut-down version of PCRE that is currently imported into Exim.
3598
3599 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3600 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3601 patterns that start (?i) but I don't think that matters. The use of (?i) is
3602 just an example; this all applies to the other options as well.
3603
3604 4. Since libtool seems to echo the compile commands it is issuing, the output
3605 from "make" can be reduced a bit by putting "@" in front of each libtool
3606 compile command.
3607
3608 5. Patch from the folks at Google for configure.in to be a bit more thorough
3609 in checking for a suitable C++ installation before trying to compile the
3610 C++ stuff. This should fix a reported problem when a compiler was present,
3611 but no suitable headers.
3612
3613 6. The man pages all had just "PCRE" as their title. I have changed them to
3614 be the relevant file name. I have also arranged that these names are
3615 retained in the file doc/pcre.txt, which is a concatenation in text format
3616 of all the man pages except the little individual ones for each function.
3617
3618 7. The NON-UNIX-USE file had not been updated for the different set of source
3619 files that come with release 6. I also added a few comments about the C++
3620 wrapper.
3621
3622
3623 Version 6.0 07-Jun-05
3624 ---------------------
3625
3626 1. Some minor internal re-organization to help with my DFA experiments.
3627
3628 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3629 didn't matter for the library itself when fully configured, but did matter
3630 when compiling without UCP support, or within Exim, where the ucp files are
3631 not imported.
3632
3633 3. Refactoring of the library code to split up the various functions into
3634 different source modules. The addition of the new DFA matching code (see
3635 below) to a single monolithic source would have made it really too
3636 unwieldy, quite apart from causing all the code to be include in a
3637 statically linked application, when only some functions are used. This is
3638 relevant even without the DFA addition now that patterns can be compiled in
3639 one application and matched in another.
3640
3641 The downside of splitting up is that there have to be some external
3642 functions and data tables that are used internally in different modules of
3643 the library but which are not part of the API. These have all had their
3644 names changed to start with "_pcre_" so that they are unlikely to clash
3645 with other external names.
3646
3647 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3648 a different (DFA) algorithm. Although it is slower than the original
3649 function, it does have some advantages for certain types of matching
3650 problem.
3651
3652 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3653 including restarting after a partial match.
3654
3655 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3656 defined when compiling for Windows was sent to me. I have put it into the
3657 code, though I have no means of testing or verifying it.
3658
3659 7. Added the pcre_refcount() auxiliary function.
3660
3661 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3662 match before or at the first newline in the subject string. In pcretest,
3663 the /f option on a pattern can be used to set this.
3664
3665 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3666 would behave wrongly. This has been present in PCRE since release 4.0.
3667
3668 10. A number of changes to the pcregrep command:
3669
3670 (a) Refactored how -x works; insert ^(...)$ instead of setting
3671 PCRE_ANCHORED and checking the length, in preparation for adding
3672 something similar for -w.
3673
3674 (b) Added the -w (match as a word) option.
3675
3676 (c) Refactored the way lines are read and buffered so as to have more
3677 than one at a time available.
3678
3679 (d) Implemented a pcregrep test script.
3680
3681 (e) Added the -M (multiline match) option. This allows patterns to match
3682 over several lines of the subject. The buffering ensures that at least
3683 8K, or the rest of the document (whichever is the shorter) is available
3684 for matching (and similarly the previous 8K for lookbehind assertions).
3685
3686 (f) Changed the --help output so that it now says
3687
3688 -w, --word-regex(p)
3689
3690 instead of two lines, one with "regex" and the other with "regexp"
3691 because that confused at least one person since the short forms are the
3692 same. (This required a bit of code, as the output is generated
3693 automatically from a table. It wasn't just a text change.)
3694
3695 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3696 option but starts with a hyphen. Could be a pattern or a path name
3697 starting with a hyphen, for instance.
3698
3699 (h) "-" can be given as a file name to represent stdin.
3700
3701 (i) When file names are being printed, "(standard input)" is used for
3702 the standard input, for compatibility with GNU grep. Previously
3703 "<stdin>" was used.
3704
3705 (j) The option --label=xxx can be used to supply a name to be used for
3706 stdin when file names are being printed. There is no short form.
3707
3708 (k) Re-factored the options decoding logic because we are going to add
3709 two more options that take data. Such options can now be given in four
3710 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3711
3712 (l) Added the -A, -B, and -C options for requesting that lines of context
3713 around matches be printed.
3714
3715 (m) Added the -L option to print the names of files that do not contain
3716 any matching lines, that is, the complement of -l.
3717
3718 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3719 continue to scan other files.
3720
3721 (o) The -s option was incorrectly implemented. For compatibility with other
3722 greps, it now suppresses the error message for a non-existent or non-
3723 accessible file (but not the return code). There is a new option called
3724 -q that suppresses the output of matching lines, which was what -s was
3725 previously doing.
3726
3727 (p) Added --include and --exclude options to specify files for inclusion
3728 and exclusion when recursing.
3729
3730 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3731 Hopefully, it now does.
3732
3733 12. Missing cast in pcre_study().
3734
3735 13. Added an "uninstall" target to the makefile.
3736
3737 14. Replaced "extern" in the function prototypes in Makefile.in with
3738 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3739 world, but is set differently for Windows.
3740
3741 15. Added a second compiling function called pcre_compile2(). The only
3742 difference is that it has an extra argument, which is a pointer to an
3743 integer error code. When there is a compile-time failure, this is set
3744 non-zero, in addition to the error test pointer being set to point to an
3745 error message. The new argument may be NULL if no error number is required
3746 (but then you may as well call pcre_compile(), which is now just a
3747 wrapper). This facility is provided because some applications need a
3748 numeric error indication, but it has also enabled me to tidy up the way
3749 compile-time errors are handled in the POSIX wrapper.
3750
3751 16. Added VPATH=.libs to the makefile; this should help when building with one
3752 prefix path and installing with another. (Or so I'm told by someone who
3753 knows more about this stuff than I do.)
3754
3755 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3756 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3757 match everything, including newlines. This is not POSIX-compatible, but
3758 somebody wanted the feature. From pcretest it can be activated by using
3759 both the P and the s flags.
3760
3761 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3762
3763 19. libpcre.pc was being incorrectly installed as executable.
3764
3765 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3766 it now also looks for '\r' so that it will work unmodified on Windows.
3767
3768 21. Added Google's contributed C++ wrapper to the distribution.
3769
3770 22. Added some untidy missing memory free() calls in pcretest, to keep
3771 Electric Fence happy when testing.
3772
3773
3774
3775 Version 5.0 13-Sep-04
3776 ---------------------
3777
3778 1. Internal change: literal characters are no longer packed up into items
3779 containing multiple characters in a single byte-string. Each character
3780 is now matched using a separate opcode. However, there may be more than one
3781 byte in the character in UTF-8 mode.
3782
3783 2. The pcre_callout_block structure has two new fields: pattern_position and
3784 next_item_length. These contain the offset in the pattern to the next match
3785 item, and its length, respectively.
3786
3787 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3788 insertion of callouts before each pattern item. Added the /C option to
3789 pcretest to make use of this.
3790
3791 4. On the advice of a Windows user, the lines
3792
3793 #if defined(_WIN32) || defined(WIN32)
3794 _setmode( _fileno( stdout ), 0x8000 );
3795 #endif /* defined(_WIN32) || defined(WIN32) */
3796
3797 have been added to the source of pcretest. This apparently does useful
3798 magic in relation to line terminators.
3799
3800 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3801 for the benefit of those environments where the "b" makes a difference.
3802
3803 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3804 to know about it. I have put a hack into configure.in that adds in code
3805 to set GCC=yes if CC=icc. This seems to end up at a point in the
3806 generated configure script that is early enough to affect the setting of
3807 compiler options, which is what is needed, but I have no means of testing
3808 whether it really works. (The user who reported this had patched the
3809 generated configure script, which of course I cannot do.)
3810
3811 LATER: After change 22 below (new libtool files), the configure script
3812 seems to know about icc (and also ecc). Therefore, I have commented out
3813 this hack in configure.in.
3814
3815 7. Added support for pkg-config (2 patches were sent in).
3816
3817 8. Negated POSIX character classes that used a combination of internal tables
3818 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3819 [[:^ascii]]. Typically, they would match almost any characters. The other
3820 POSIX classes were not broken in this way.
3821
3822 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3823 to find the match, as PCRE was deluded into thinking that the match had to
3824 start at the start point or following a newline. The same bug applied to
3825 patterns with negative forward assertions or any backward assertions
3826 preceding ".*" at the start, unless the pattern required a fixed first
3827 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3828
3829 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3830 starting at the last subject character, bytes beyond the end of the subject
3831 string were read.
3832
3833 11. Renamed the variable "class" as "classbits" to make life easier for C++
3834 users. (Previously there was a macro definition, but it apparently wasn't
3835 enough.)
3836
3837 12. Added the new field "tables" to the extra data so that tables can be passed
3838 in at exec time, or the internal tables can be re-selected. This allows
3839 a compiled regex to be saved and re-used at a later time by a different
3840 program that might have everything at different addresses.
3841
3842 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3843 -R library as well as a -L library.
3844
3845 14. The debugging options of pcretest (-d on the command line or D on a
3846 pattern) showed incorrect output for anything following an extended class
3847 that contained multibyte characters and which was followed by a quantifier.
3848
3849 15. Added optional support for general category Unicode character properties
3850 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3851 support. It adds about 90K to the size of the library. The meanings of the
3852 inbuilt class escapes such as \d and \s have NOT been changed.
3853
3854 16. Updated pcredemo.c to include calls to free() to release the memory for the
3855 compiled pattern.
3856
3857 17. The generated file chartables.c was being created in the source directory
3858 instead of in the building directory. This caused the build to fail if the
3859 source directory was different from the building directory, and was
3860 read-only.
3861
3862 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3863 file. No doubt somebody will tell me if they don't make sense... Also added
3864 Dan Mooney's comments about building on OpenVMS.
3865
3866 19. Added support for partial matching via the PCRE_PARTIAL option for
3867 pcre_exec() and the \P data escape in pcretest.
3868
3869 20. Extended pcretest with 3 new pattern features:
3870
3871 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3872 write the compiled pattern to the file whose name is "rest-of-line".
3873 This is a straight binary dump of the data, with the saved pointer to
3874 the character tables forced to be NULL. The study data, if any, is
3875 written too. After writing, pcretest reads a new pattern.
3876
3877 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3878 compiled pattern from the given file. There must not be any
3879 occurrences of "<" in the file name (pretty unlikely); if there are,
3880 pcretest will instead treat the initial "<" as a pattern delimiter.
3881 After reading in the pattern, pcretest goes on to read data lines as
3882 usual.
3883
3884 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3885 and 16-bit fields in a compiled pattern, to simulate a pattern that
3886 was compiled on a host of opposite endianness.
3887
3888 21. The pcre-exec() function can now cope with patterns that were compiled on
3889 hosts of opposite endianness, with this restriction:
3890
3891 As for any compiled expression that is saved and used later, the tables
3892 pointer field cannot be preserved; the extra_data field in the arguments
3893 to pcre_exec() should be used to pass in a tables address if a value
3894 other than the default internal tables were used at compile time.
3895
3896 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3897 now diagnosed as an error. Previously, most of the time, a negative number
3898 would have been treated as zero, but if in addition "ovector" was passed as
3899 NULL, a crash could occur.
3900
3901 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3902 new versions from the libtool 1.5 distribution (the last one is a copy of
3903 a file called libtool.m4). This seems to have fixed the need to patch
3904 "configure" to support Darwin 1.3 (which I used to do). However, I still
3905 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3906 workstation).
3907
3908 24. Changed the PCRE licence to be the more standard "BSD" licence.
3909
3910
3911 Version 4.5 01-Dec-03
3912 ---------------------
3913
3914 1. There has been some re-arrangement of the code for the match() function so
3915 that it can be compiled in a version that does not call itself recursively.
3916 Instead, it keeps those local variables that need separate instances for
3917 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3918 needs to "recurse". Keeping track of where control must go is done by means
3919 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3920 hide most of the details from the main code, and operates only if
3921 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3922 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3923 operating.
3924
3925 To make it easier for callers to provide specially tailored get/free
3926 functions for this usage, two new functions, pcre_stack_malloc, and
3927 pcre_stack_free, are used. They are always called in strict stacking order,
3928 and the size of block requested is always the same.
3929
3930 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3931 PCRE has been compiled to use the stack or the heap for recursion. The
3932 -C option of pcretest uses this to show which version is compiled.
3933
3934 A new data escape \S, is added to pcretest; it causes the amounts of store
3935 obtained and freed by both kinds of malloc/free at match time to be added
3936 to the output.
3937
3938 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3939 what's available on my current Linux desktop machine.
3940
3941 3. When matching a UTF-8 string, the test for a valid string at the start has
3942 been extended. If start_offset is not zero, PCRE now checks that it points
3943 to a byte that is the start of a UTF-8 character. If not, it returns
3944 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3945 this is necessary because there may be backward assertions in the pattern.
3946 When matching the same subject several times, it may save resources to use
3947 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3948
3949 4. The code for checking the validity of UTF-8 strings has been tightened so
3950 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3951 containing "overlong sequences".
3952
3953 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3954 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3955 should have been "&", but it just so happened that all the cases this let
3956 through by mistake were picked up later in the function.
3957
3958 6. I had used a variable called "isblank" - this is a C99 function, causing
3959 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3960
3961 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3962 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3963 so the version is shown; (c) stop "make test" echoing "./RunTest".
3964
3965 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3966
3967 9. The prototype for memmove() for systems that don't have it was using
3968 size_t, but the inclusion of the header that defines size_t was later. I've
3969 moved the #includes for the C headers earlier to avoid this.
3970
3971 10. Added some adjustments to the code to make it easier to compiler on certain
3972 special systems:
3973
3974 (a) Some "const" qualifiers were missing.
3975 (b) Added the macro EXPORT before all exported functions; by default this
3976 is defined to be empty.
3977 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3978 that it reads its output file name as an argument instead of writing
3979 to the standard output and assuming this can be redirected.
3980
3981 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3982 class containing characters with values greater than 255, PCRE compilation
3983 went into a loop.
3984
3985 12. A recursive reference to a subpattern that was within another subpattern
3986 that had a minimum quantifier of zero caused PCRE to crash. For example,
3987 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3988 recursion. If the recursively-called subpattern itself had a zero repeat,
3989 that was OK.
3990
3991 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3992 buffer into which it was copied (for escape processing) was still set at
3993 1024, so long lines caused crashes.
3994
3995 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3996 "internal error: code overflow...". This applied to any character class
3997 that was followed by a possessive quantifier.
3998
3999 15. Modified the Makefile to add libpcre.la as a prerequisite for
4000 libpcreposix.la because I was told this is needed for a parallel build to
4001 work.
4002
4003 16. If a pattern that contained .* following optional items at the start was
4004 studied, the wrong optimizing data was generated, leading to matching
4005 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4006 matching string must start with a or b or c. The correct conclusion for
4007 this pattern is that a match can start with any character.
4008
4009
4010 Version 4.4 13-Aug-03
4011 ---------------------
4012
4013 1. In UTF-8 mode, a character class containing characters with values between
4014 127 and 255 was not handled correctly if the compiled pattern was studied.
4015 In fixing this, I have also improved the studying algorithm for such
4016 classes (slightly).
4017
4018 2. Three internal functions had redundant arguments passed to them. Removal
4019 might give a very teeny performance improvement.
4020
4021 3. Documentation bug: the value of the capture_top field in a callout is *one
4022 more than* the number of the hightest numbered captured substring.
4023
4024 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4025 in incorrectly linking with a previously installed version. They now link
4026 explicitly with libpcre.la.
4027
4028 5. configure.in no longer needs to recognize Cygwin specially.
4029
4030 6. A problem in pcre.in for Windows platforms is fixed.
4031
4032 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4033 pcretest, it used to include the size of the study block as part of its
4034 output. Unfortunately, the structure contains a field that has a different
4035 size on different hardware architectures. This meant that the tests that
4036 showed this size failed. As the block is currently always of a fixed size,
4037 this information isn't actually particularly useful in pcretest output, so
4038 I have just removed it.
4039
4040 8. Three pre-processor statements accidentally did not start in column 1.
4041 Sadly, there are *still* compilers around that complain, even though
4042 standard C has not required this for well over a decade. Sigh.
4043
4044 9. In pcretest, the code for checking callouts passed small integers in the
4045 callout_data field, which is a void * field. However, some picky compilers
4046 complained about the casts involved for this on 64-bit systems. Now
4047 pcretest passes the address of the small integer instead, which should get
4048 rid of the warnings.
4049
4050 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4051 both compile and run time, and gives an error if an invalid UTF-8 sequence
4052 is found. There is a option for disabling this check in cases where the
4053 string is known to be correct and/or the maximum performance is wanted.
4054
4055 11. In response to a bug report, I changed one line in Makefile.in from
4056
4057 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4058 to
4059 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4060
4061 to look similar to other lines, but I have no way of telling whether this
4062 is the right thing to do, as I do not use Windows. No doubt I'll get told
4063 if it's wrong...
4064
4065
4066 Version 4.3 21-May-03
4067 ---------------------
4068
4069 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4070 Makefile.
4071
4072 2. Some refactoring to improve the quality of the code:
4073
4074 (i) The utf8_table... variables are now declared "const".
4075
4076 (ii) The code for \cx, which used the "case flipping" table to upper case
4077 lower case letters, now just substracts 32. This is ASCII-specific,
4078 but the whole concept of \cx is ASCII-specific, so it seems
4079 reasonable.
4080
4081 (iii) PCRE was using its character types table to recognize decimal and
4082 hexadecimal digits in the pattern. This is silly, because it handles
4083 only 0-9, a-f, and A-F, but the character types table is locale-
4084 specific, which means strange things might happen. A private
4085 table is now used for this - though it costs 256 bytes, a table is
4086 much faster than multiple explicit tests. Of course, the standard
4087 character types table is still used for matching digits in subject
4088 strings against \d.
4089
4090 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4091 ending in _t are). So I've renamed it as ESC_tee.
4092
4093 3. The first argument for regexec() in the POSIX wrapper should have been
4094 defined as "const".
4095
4096 4. Changed pcretest to use malloc() for its buffers so that they can be
4097 Electric Fenced for debugging.
4098
4099 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4100 to read one or more bytes before the start of the subject string. Often this
4101 had no effect on PCRE's behaviour, but in some circumstances it could
4102 provoke a segmentation fault.
4103
4104 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4105 to try to read one or more bytes before the start of the subject string.
4106
4107 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4108 UTF-8 support could misbehave in various ways if the subject string
4109 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4110 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4111 back over UTF-8 characters.)
4112
4113
4114 Version 4.2 14-Apr-03
4115 ---------------------
4116
4117 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4118
4119 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4120 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4121 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4122 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4123 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4124 and BUILD_EXEEXT
4125 Note: automatic setting of the BUILD variables is not yet working
4126 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4127 compile-time but not at link-time
4128 [LINK]: use for linking executables only
4129 make different versions for Windows and non-Windows
4130 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4131 libraries
4132 [LINK_FOR_BUILD]: new variable
4133 [OBJEXT]: use throughout
4134 [EXEEXT]: use throughout
4135 <winshared>: new target
4136 <wininstall>: new target
4137 <dftables.o>: use native compiler
4138 <dftables>: use native linker
4139 <install>: handle Windows platform correctly
4140 <clean>: ditto
4141 <check>: ditto
4142 copy DLL to top builddir before testing
4143
4144 As part of these changes, -no-undefined was removed again. This was reported
4145 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4146 in any case.
4147
4148 3. Some tidies to get rid of compiler warnings:
4149
4150 . In the match_data structure, match_limit was an unsigned long int, whereas
4151 match_call_count was an int. I've made them both unsigned long ints.
4152
4153 . In pcretest the fact that a const uschar * doesn't automatically cast to
4154 a void * provoked a warning.
4155
4156 . Turning on some more compiler warnings threw up some "shadow" variables
4157 and a few more missing casts.
4158
4159 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4160 option, a class that contained a single character with a value between 128
4161 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4162
4163 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4164 option, a class that contained several characters, but with at least one
4165 whose value was between 128 and 255 caused PCRE to crash.
4166
4167
4168 Version 4.1 12-Mar-03
4169 ---------------------
4170
4171 1. Compiling with gcc -pedantic found a couple of places where casts were
4172 needed, and a string in dftables.c that was longer than standard compilers are
4173 required to support.
4174
4175 2. Compiling with Sun's compiler found a few more places where the code could
4176 be tidied up in order to avoid warnings.
4177
4178 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4179 first of these names is deprecated in the latest Autoconf in favour of the name
4180 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4181 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4182 analogy I have changed it to CFLAGS_FOR_BUILD.
4183
4184 4. Added -no-undefined to the linking command in the Makefile, because this is
4185 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4186 linking step for the pcreposix library.
4187
4188 5. PCRE was failing to diagnose the case of two named groups with the same
4189 name.
4190
4191 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4192 literal character that is needed in the subject for a match, and scans along to
4193 ensure that it is present before embarking on the full matching process. This
4194 saves time in cases of nested unlimited repeats that are never going to match.
4195 Problem: the scan can take a lot of time if the subject is very long (e.g.
4196 megabytes), thus penalizing straightforward matches. It is now done only if the
4197 amount of subject to be scanned is less than 1000 bytes.
4198
4199 7. A lesser problem with the same optimization is that it was recording the
4200 first character of an anchored pattern as "needed", thus provoking a search
4201 right along the subject, even when the first match of the pattern was going to
4202 fail. The "needed" character is now not set for anchored patterns, unless it
4203 follows something in the pattern that is of non-fixed length. Thus, it still
4204 fulfils its original purpose of finding quick non-matches in cases of nested
4205 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4206
4207
4208 Version 4.0 17-Feb-03
4209 ---------------------
4210
4211 1. If a comment in an extended regex that started immediately after a meta-item
4212 extended to the end of string, PCRE compiled incorrect data. This could lead to
4213 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4214
4215 2. Moved to autoconf 2.53 and libtool 1.4.2.
4216
4217 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4218 the special perltest8 script is no longer needed - all the tests can be run
4219 from a single perltest script.
4220
4221 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4222 by \s. It has now been removed in PCRE. This means it isn't recognized as
4223 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4224 class [:space:] *does* include VT, thereby creating a mess.
4225
4226 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4227 space and tab.
4228
4229 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4230 its new features into the main test script, reducing the number of scripts.
4231
4232 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4233 were backward compatible, and made the (?i) apply to the whole pattern, as if
4234 /i were given. Now it behaves more logically, and applies the option setting
4235 only to what follows. PCRE has been changed to follow suit. However, if it
4236 finds options settings right at the start of the pattern, it extracts them into
4237 the global options, as before. Thus, they show up in the info data.
4238
4239 8. Added support for the \Q...\E escape sequence. Characters in between are
4240 treated as literals. This is slightly different from Perl in that $ and @ are
4241 also handled as literals inside the quotes. In Perl, they will cause variable
4242 interpolation. Note the following examples:
4243
4244 Pattern PCRE matches Perl matches
4245
4246 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4247 \Qabc\$xyz\E abc\$xyz abc\$xyz
4248 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4249
4250 For compatibility with Perl, \Q...\E sequences are recognized inside character
4251 classes as well as outside them.
4252
4253 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4254 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4255 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4256 signed/unsigned warnings.
4257
4258 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4259 option for pcretest, so I've replaced it by a simple function that does just
4260 that job.
4261
4262 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4263 "pcregrep -".
4264
4265 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4266 Java package. This provides some syntactic sugar for simple cases of what my
4267 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4268 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4269 item, you can use this simplified notation. Note that only makes sense with
4270 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4271 greediness, whatever the setting of the PCRE_UNGREEDY option.
4272
4273 13. A change of greediness default within a pattern was not taking effect at
4274 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4275 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4276 was abstracted outside.
4277
4278 14. PCRE now supports the \G assertion. It is true when the current matching
4279 position is at the start point of the match. This differs from \A when the
4280 starting offset is non-zero. Used with the /g option of pcretest (or similar
4281 code), it works in the same way as it does for Perl's /g option. If all
4282 alternatives of a regex begin with \G, the expression is anchored to the start
4283 match position, and the "anchored" flag is set in the compiled expression.
4284
4285 15. Some bugs concerning the handling of certain option changes within patterns
4286 have been fixed. These applied to options other than (?ims). For example,
4287 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4288 been the other way round. Some of this was related to change 7 above.
4289
4290 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4291 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4292 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4293 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4294
4295 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4296 mode. Unlike ".", it always matches newline, whatever the setting of
4297 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4298 assertions. Perl allows it, but it doesn't (in general) work because it can't
4299 calculate the length of the lookbehind. At least, that's the case for Perl
4300 5.8.0 - I've been told they are going to document that it doesn't work in
4301 future.
4302
4303 18. Added an error diagnosis for escapes that PCRE does not support: these are
4304 \L, \l, \N, \P, \p, \U, \u, and \X.
4305
4306 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4307 reading past the end of the pattern in cases such as /[abcd/.
4308
4309 20. PCRE was getting more memory than necessary for patterns with classes that
4310 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4311
4312 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4313 compiling PCRE for use with Virtual Pascal.
4314
4315 22. Small fix to the Makefile to make it work properly if the build is done
4316 outside the source tree.
4317
4318 23. Added a new extension: a condition to go with recursion. If a conditional
4319 subpattern starts with (?(R) the "true" branch is used if recursion has
4320 happened, whereas the "false" branch is used only at the top level.
4321
4322 24. When there was a very long string of literal characters (over 255 bytes
4323 without UTF support, over 250 bytes with UTF support), the computation of how
4324 much memory was required could be incorrect, leading to segfaults or other
4325 strange effects.
4326
4327 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4328 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4329 there was a subsequent back reference to those brackets. This meant that, for
4330 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4331 possible to check for precisely this case. All we can do is abandon the
4332 optimization if .* occurs inside capturing brackets when there are any back
4333 references whatsoever. (See below for a better fix that came later.)
4334
4335 26. The handling of the optimization for finding the first character of a
4336 non-anchored pattern, and for finding a character that is required later in the
4337 match were failing in some cases. This didn't break the matching; it just
4338 failed to optimize when it could. The way this is done has been re-implemented.
4339
4340 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4341
4342 28. Added a new feature that provides some of the functionality that Perl
4343 provides with (?{...}). The facility is termed a "callout". The way it is done
4344 in PCRE is for the caller to provide an optional function, by setting
4345 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4346 global variable. By default it is unset, which disables all calling out. To get
4347 the function called, the regex must include (?C) at appropriate points. This
4348 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4349 This provides a means of identifying different callout points. When PCRE
4350 reaches such a point in the regex, if pcre_callout has been set, the external
4351 function is called. It is provided with data in a structure called
4352 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4353 matching continues; if it returns a non-zero value, the match at the current
4354 point fails. However, backtracking will occur if possible. [This was changed
4355 later and other features added - see item 49 below.]
4356
4357 29. pcretest is upgraded to test the callout functionality. It provides a
4358 callout function that displays information. By default, it shows the start of
4359 the match and the current position in the text. There are some new data escapes
4360 to vary what happens:
4361
4362 \C+ in addition, show current contents of captured substrings
4363 \C- do not supply a callout function
4364 \C!n return 1 when callout number n is reached
4365 \C!n!m return 1 when callout number n is reached for the mth time
4366
4367 30. If pcregrep was called with the -l option and just a single file name, it
4368 output "<stdin>" if a match was found, instead of the file name.
4369
4370 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4371 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4372 pcre_exec(). This saves a malloc/free per call. The default value of
4373 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4374 when configuring.
4375
4376 32. The default maximum size of a compiled pattern is 64K. There have been a
4377 few cases of people hitting this limit. The code now uses macros to handle the
4378 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4379 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4380 configuring. Tests 2 and 5 work only with 2-byte links because they output
4381 debugging information about compiled patterns.
4382
4383 33. Internal code re-arrangements:
4384
4385 (a) Moved the debugging function for printing out a compiled regex into
4386 its own source file (printint.c) and used #include to pull it into
4387 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4388 separate copies.
4389
4390 (b) Defined the list of op-code names for debugging as a macro in
4391 internal.h so that it is next to the definition of the opcodes.
4392
4393 (c) Defined a table of op-code lengths for simpler skipping along compiled
4394 code. This is again a macro in internal.h so that it is next to the
4395 definition of the opcodes.
4396
4397 34. Added support for recursive calls to individual subpatterns, along the
4398 lines of Robin Houston's patch (but implemented somewhat differently).
4399
4400 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4401 allow it to read and process whole directories in Win32. This code was
4402 contributed by Lionel Fourquaux; it has not been tested by me.
4403
4404 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4405 used to name a group. Names consist of alphanumerics and underscores, and must
4406 be unique. Back references use the syntax (?P=name) and recursive calls use
4407 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4408 numbers. The function pcre_fullinfo() can be used after compilation to extract
4409 a name/number map. There are three relevant calls:
4410
4411 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4412 PCRE_INFO_NAMECOUNT yields the number of entries
4413 PCRE_INFO_NAMETABLE yields a pointer to the map.
4414
4415 The map is a vector of fixed-size entries. The size of each entry depends on
4416 the length of the longest name used. The first two bytes of each entry are the
4417 group number, most significant byte first. There follows the corresponding
4418 name, zero terminated. The names are in alphabetical order.
4419
4420 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4421 case instead of 255. Making it the same both with and without UTF-8 support
4422 means that the same test output works with both.
4423
4424 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4425 calling malloc() with a zero argument.
4426
4427 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4428 optimization. I've improved things by keeping a bitmap of backreferences with
4429 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4430 fact referenced, the optimization can be applied. It is unlikely that a
4431 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4432 the match to follow \n) will appear inside brackets with a number greater than
4433 31, but if it does, any back reference > 31 suppresses the optimization.
4434
4435 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4436 of disabling numbered capturing parentheses. Any opening parenthesis that is
4437 not followed by ? behaves as if it were followed by ?: but named parentheses
4438 can still be used for capturing (and they will acquire numbers in the usual
4439 way).
4440
4441 41. Redesigned the return codes from the match() function into yes/no/error so
4442 that errors can be passed back from deep inside the nested calls. A malloc
4443 failure while inside a recursive subpattern call now causes the
4444 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4445
4446 42. It is now possible to set a limit on the number of times the match()
4447 function is called in a call to pcre_exec(). This facility makes it possible to
4448 limit the amount of recursion and backtracking, though not in a directly
4449 obvious way, because the match() function is used in a number of different
4450 circumstances. The count starts from zero for each position in the subject
4451 string (for non-anchored patterns). The default limit is, for compatibility, a
4452 large number, namely 10 000 000. You can change this in two ways:
4453
4454 (a) When configuring PCRE before making, you can use --with-match-limit=n
4455 to set a default value for the compiled library.
4456
4457 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4458 a different value is set. See 45 below.
4459
4460 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4461
4462 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4463 of things that can be changed at compile time. The first argument specifies
4464 what is wanted and the second points to where the information is to be placed.
4465 The current list of available information is:
4466
4467 PCRE_CONFIG_UTF8
4468
4469 The output is an integer that is set to one if UTF-8 support is available;
4470 otherwise it is set to zero.
4471
4472 PCRE_CONFIG_NEWLINE
4473
4474 The output is an integer that it set to the value of the code that is used for
4475 newline. It is either LF (10) or CR (13).
4476
4477 PCRE_CONFIG_LINK_SIZE
4478
4479 The output is an integer that contains the number of bytes used for internal
4480 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4481
4482 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4483
4484 The output is an integer that contains the threshold above which the POSIX
4485 interface uses malloc() for output vectors. See item 31 above.
4486
4487 PCRE_CONFIG_MATCH_LIMIT
4488
4489 The output is an unsigned integer that contains the default limit of the number
4490 of match() calls in a pcre_exec() execution. See 42 above.
4491
4492 44. pcretest has been upgraded by the addition of the -C option. This causes it
4493 to extract all the available output from the new pcre_config() function, and to
4494 output it. The program then exits immediately.
4495
4496 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4497 order to support additional features. One way would have been to define
4498 pcre_exec2() (for example) with extra arguments, but this would not have been
4499 extensible, and would also have required all calls to the original function to
4500 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4501 is used for passing in "extra" data from pcre_study().
4502
4503 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4504 contains the following fields:
4505
4506 flags a bitmap indicating which of the following fields are set
4507 study_data opaque data from pcre_study()
4508 match_limit a way of specifying a limit on match() calls for a specific
4509 call to pcre_exec()
4510 callout_data data for callouts (see 49 below)
4511
4512 The flag bits are also defined in pcre.h, and are
4513
4514 PCRE_EXTRA_STUDY_DATA
4515 PCRE_EXTRA_MATCH_LIMIT
4516 PCRE_EXTRA_CALLOUT_DATA
4517
4518 The pcre_study() function now returns one of these new pcre_extra blocks, with
4519 the actual study data pointed to by the study_data field, and the
4520 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4521 before. That is, this change is entirely upwards-compatible and requires no
4522 change to existing code.
4523
4524 If you want to pass in additional data to pcre_exec(), you can either place it
4525 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4526 block.
4527
4528 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4529 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4530 times with different match limits, until it finds the minimum value needed for
4531 pcre_exec() to complete. The value is then output. This can be instructive; for
4532 most simple matches the number is quite small, but for pathological cases it
4533 gets very large very quickly.
4534
4535 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4536 returns the size of the data block pointed to by the study_data field in a
4537 pcre_extra block, that is, the value that was passed as the argument to
4538 pcre_malloc() when PCRE was getting memory in which to place the information
4539 created by pcre_study(). The fourth argument should point to a size_t variable.
4540 pcretest has been extended so that this information is shown after a successful
4541 pcre_study() call when information about the compiled regex is being displayed.
4542
4543 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4544 because what follows is always an absolute path. (Later: it turns out that this
4545 is more than cosmetic for MinGW, because it doesn't like empty path
4546 components.)
4547
4548 49. Some changes have been made to the callout feature (see 28 above):
4549
4550 (i) A callout function now has three choices for what it returns:
4551
4552 0 => success, carry on matching
4553 > 0 => failure at this point, but backtrack if possible
4554 < 0 => serious error, return this value from pcre_exec()
4555
4556 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4557 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4558 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4559 use by callout functions. It will never be used by PCRE itself.
4560
4561 (ii) The pcre_extra structure (see 45 above) has a void * field called
4562 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4563 pcre_callout_block structure has a field of the same name. The contents of
4564 the field passed in the pcre_extra structure are passed to the callout
4565 function in the corresponding field in the callout block. This makes it
4566 easier to use the same callout-containing regex from multiple threads. For
4567 testing, the pcretest program has a new data escape
4568
4569 \C*n pass the number n (may be negative) as callout_data
4570
4571 If the callout function in pcretest receives a non-zero value as
4572 callout_data, it returns that value.
4573
4574 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4575 there were some redundant $(CFLAGS) in commands that are now specified as
4576 $(LINK), which already includes $(CFLAGS).
4577
4578 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4579 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4580 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4581 one-byte characters throughout. Note that case-insensitive matching applies
4582 only to characters whose values are less than 256. PCRE doesn't support the
4583 notion of cases for higher-valued characters.
4584
4585 (i) A character class whose characters are all within 0-255 is handled as
4586 a bit map, and the map is inverted for negative classes. Previously, a
4587 character > 255 always failed to match such a class; however it should
4588 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4589
4590 (ii) A negated character class with a single character < 255 is coded as
4591 "not this character" (OP_NOT). This wasn't working properly when the test
4592 character was multibyte, either singly or repeated.
4593
4594 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4595 mode, for example: \x{100}{2,3}.
4596
4597 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4598 singly or repeated) now correctly test multibyte characters. However,
4599 PCRE doesn't recognize any characters with values greater than 255 as
4600 digits, spaces, or word characters. Such characters always match \D, \S,
4601 and \W, and never match \d, \s, or \w.
4602
4603 (v) Classes may now contain characters and character ranges with values
4604 greater than 255. For example: [ab\x{100}-\x{400}].
4605
4606 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4607 PCRE in UTF-8 mode.
4608
4609 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4610 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4611 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4612 value.)
4613
4614 53. The single man page has become too large. I have therefore split it up into
4615 a number of separate man pages. These also give rise to individual HTML pages;
4616 these are now put in a separate directory, and there is an index.html page that
4617 lists them all. Some hyperlinking between the pages has been installed.
4618
4619 54. Added convenience functions for handling named capturing parentheses.
4620
4621 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4622 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4623 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4624 are faulted.
4625
4626 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4627 calling configure. These values are used when compiling the dftables.c program
4628 which is run to generate the source of the default character tables. They
4629 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4630 you will need to set these values.
4631
4632 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4633
4634
4635 Version 3.9 02-Jan-02
4636 ---------------------
4637
4638 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4639
4640 2. If --disable-static was given, the building process failed when trying to
4641 build pcretest and pcregrep. (For some reason it was using libtool to compile
4642 them, which is not right, as they aren't part of the library.)
4643
4644
4645 Version 3.8 18-Dec-01
4646 ---------------------
4647
4648 1. The experimental UTF-8 code was completely screwed up. It was packing the
4649 bytes in the wrong order. How dumb can you get?
4650
4651
4652 Version 3.7 29-Oct-01
4653 ---------------------
4654
4655 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4656 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4657 this didn't happen under Solaris 8, where I normally test things.
4658
4659 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4660 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4661 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4662 this fix an hour or so after the initial 3.7 release.)
4663
4664
4665 Version 3.6 23-Oct-01
4666 ---------------------
4667
4668 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4669 offsets passed as NULL with zero offset count.
4670
4671 2. The config.guess and config.sub files had not been updated when I moved to
4672 the latest autoconf.
4673
4674
4675 Version 3.5 15-Aug-01
4676 ---------------------
4677
4678 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4679 had been forgotten.
4680
4681 2. By using declared but undefined structures, we can avoid using "void"
4682 definitions in pcre.h while keeping the internal definitions of the structures
4683 private.
4684
4685 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4686 user point of view, this means that both static and shared libraries are built
4687 by default, but this can be individually controlled. More of the work of
4688 handling this static/shared cases is now inside libtool instead of PCRE's make
4689 file.
4690
4691 4. The pcretest utility is now installed along with pcregrep because it is
4692 useful for users (to test regexs) and by doing this, it automatically gets
4693 relinked by libtool. The documentation has been turned into a man page, so
4694 there are now .1, .txt, and .html versions in /doc.
4695
4696 5. Upgrades to pcregrep:
4697 (i) Added long-form option names like gnu grep.
4698 (ii) Added --help to list all options with an explanatory phrase.
4699 (iii) Added -r, --recursive to recurse into sub-directories.
4700 (iv) Added -f, --file to read patterns from a file.
4701
4702 6. pcre_exec() was referring to its "code" argument before testing that
4703 argument for NULL (and giving an error if it was NULL).
4704
4705 7. Upgraded Makefile.in to allow for compiling in a different directory from
4706 the source directory.
4707
4708 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4709 options bits, the pointer it was passed was to an int instead of to an unsigned
4710 long int. This mattered only on 64-bit systems.
4711
4712 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4713 generated) instead of pcre.in, which it its source. Also made the same change
4714 in several of the .c files.
4715
4716 10. A new release of gcc defines printf() as a macro, which broke pcretest
4717 because it had an ifdef in the middle of a string argument for printf(). Fixed
4718 by using separate calls to printf().
4719
4720 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4721 script, to force use of CR or LF instead of \n in the source. On non-Unix
4722 systems, the value can be set in config.h.
4723
4724 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4725 absolute limit. Changed the text of the error message to make this clear, and
4726 likewise updated the man page.
4727
4728 13. The limit of 99 on the number of capturing subpatterns has been removed.
4729 The new limit is 65535, which I hope will not be a "real" limit.
4730
4731
4732 Version 3.4 22-Aug-00
4733 ---------------------
4734
4735 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4736
4737 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4738
4739
4740 Version 3.3 01-Aug-00
4741 ---------------------
4742
4743 1. If an octal character was given, but the value was greater than \377, it
4744 was not getting masked to the least significant bits, as documented. This could
4745 lead to crashes in some systems.
4746
4747 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4748 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4749
4750 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4751 These just pass their arguments on to (pcre_free)(), but they are provided
4752 because some uses of PCRE bind it to non-C systems that can call its functions,
4753 but cannot call free() or pcre_free() directly.
4754
4755 4. Add "make test" as a synonym for "make check". Corrected some comments in
4756 the Makefile.
4757
4758 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4759 Makefile.
4760
4761 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4762 command called pgrep for grepping around the active processes.
4763
4764 7. Added the beginnings of support for UTF-8 character strings.
4765
4766 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4767 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4768 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4769 out for the ar command.)
4770
4771
4772 Version 3.2 12-May-00
4773 ---------------------
4774
4775 This is purely a bug fixing release.
4776
4777 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4778 of ZA. This was just one example of several cases that could provoke this bug,
4779 which was introduced by change 9 of version 2.00. The code for breaking
4780 infinite loops after an iteration that matches an empty string was't working
4781 correctly.
4782
4783 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4784 when matched against abbab (for example). After matching an empty string, it
4785 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4786 caused it to match further down the string than it should.
4787
4788 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4789 was there because it doesn't seem to be needed, and it causes trouble on some
4790 systems, as it is not a Standard C header. It has been removed.
4791
4792 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4793 were reported on the Macintosh. The changes were from
4794
4795 while ((c = *(++ptr)) != 0 && c != '\n');
4796 to
4797 while ((c = *(++ptr)) != 0 && c != '\n') ;
4798
4799 Totally extraordinary, but if that's what it takes...
4800
4801 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4802 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4803 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4804 assumes the way PCRE uses memmove() (always moving upwards).
4805
4806 6. PCRE is being used in one environment where strchr() is not available. There
4807 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4808 faster code anyway.
4809
4810
4811 Version 3.1 09-Feb-00
4812 ---------------------
4813
4814 The only change in this release is the fixing of some bugs in Makefile.in for
4815 the "install" target:
4816
4817 (1) It was failing to install pcreposix.h.
4818
4819 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4820
4821
4822 Version 3.0 01-Feb-00
4823 ---------------------
4824
4825 1. Add support for the /+ modifier to perltest (to output $` like it does in
4826 pcretest).
4827
4828 2. Add support for the /g modifier to perltest.
4829
4830 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4831 matches null strings.
4832
4833 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4834 pattern. Perl treats empty patterns specially - it reuses the most recent
4835 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4836 effect.
4837
4838 5. The POSIX interface was broken in that it was just handing over the POSIX
4839 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4840 required a bigger vector, with some working space on the end. This means that
4841 the POSIX wrapper now has to get and free some memory, and copy the results.
4842
4843 6. Added some simple autoconf support, placing the test data and the
4844 documentation in separate directories, re-organizing some of the
4845 information files, and making it build pcre-config (a GNU standard). Also added
4846 libtool support for building PCRE as a shared library, which is now the
4847 default.
4848
4849 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4850 09 are not valid octal constants. Single digits will be used for minor values
4851 less than 10.
4852
4853 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4854 existing programs that set these in the POSIX interface can use PCRE without
4855 modification.
4856
4857 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4858 return all that pcre_info() returns, plus additional data. The pcre_info()
4859 function is retained for compatibility, but is considered to be obsolete.
4860
4861 10. Added experimental recursion feature (?R) to handle one common case that
4862 Perl 5.6 will be able to do with (?p{...}).
4863
4864 11. Added support for POSIX character classes like [:alpha:], which Perl is
4865 adopting.
4866
4867
4868 Version 2.08 31-Aug-99
4869 ----------------------
4870
4871 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4872 trying to match at the startoffset position, but instead was moving forward to
4873 the next newline as if a previous match had failed.
4874
4875 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4876 and could get into a loop if a null string was matched other than at the start
4877 of the subject.
4878
4879 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4880 be distinguished at compile time, and for completeness also added PCRE_DATE.
4881
4882 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4883 in GnuWin32 environments.
4884
4885
4886 Version 2.07 29-Jul-99
4887 ----------------------
4888
4889 1. The documentation is now supplied in plain text form and HTML as well as in
4890 the form of man page sources.
4891
4892 2. C++ compilers don't like assigning (void *) values to other pointer types.
4893 In particular this affects malloc(). Although there is no problem in Standard
4894 C, I've put in casts to keep C++ compilers happy.
4895
4896 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4897 should be (const char *).
4898
4899 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4900 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4901 However, I haven't made this a standard facility. The documentation doesn't
4902 mention it, and the Makefile doesn't support it.
4903
4904 5. The Makefile now contains an "install" target, with editable destinations at
4905 the top of the file. The pcretest program is not installed.
4906
4907 6. pgrep -V now gives the PCRE version number and date.
4908
4909 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4910 causing the entire string to be ignored, instead of just the last character.
4911
4912 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4913 non-matching string, it can take a very, very long time, even for strings of
4914 quite modest length, because of the nested recursion. PCRE now does better in
4915 some of these cases. It does this by remembering the last required literal
4916 character in the pattern, and pre-searching the subject to ensure it is present
4917 before running the real match. In other words, it applies a heuristic to detect
4918 some types of certain failure quickly, and in the above example, if presented
4919 with a string that has no trailing " it gives "no match" very quickly.
4920
4921 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4922 other alternatives are tried instead.
4923
4924
4925 Version 2.06 09-Jun-99
4926 ----------------------
4927
4928 1. Change pcretest's output for amount of store used to show just the code
4929 space, because the remainder (the data block) varies in size between 32-bit and
4930 64-bit systems.
4931
4932 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4933 start matching at. This allows lookbehinds to work when searching for multiple
4934 occurrences in a string.
4935
4936 3. Added additional options to pcretest for testing multiple occurrences:
4937
4938 /+ outputs the rest of the string that follows a match
4939 /g loops for multiple occurrences, using the new startoffset argument
4940 /G loops for multiple occurrences by passing an incremented pointer
4941
4942 4. PCRE wasn't doing the "first character" optimization for patterns starting
4943 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4944 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4945 the letter 'x'. On long subject strings, this gives a significant speed-up.
4946
4947
4948 Version 2.05 21-Apr-99
4949 ----------------------
4950
4951 1. Changed the type of magic_number from int to long int so that it works
4952 properly on 16-bit systems.
4953
4954 2. Fixed a bug which caused patterns starting with .* not to work correctly
4955 when the subject string contained newline characters. PCRE was assuming
4956 anchoring for such patterns in all cases, which is not correct because .* will
4957 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4958 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4959 must be retried after every newline in the subject.
4960
4961
4962 Version 2.04 18-Feb-99
4963 ----------------------
4964
4965 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4966 computation of the store needed to hold the pattern was incorrect (too large).
4967 If such patterns were nested a few deep, this could multiply and become a real
4968 problem.
4969
4970 2. Added /M option to pcretest to show the memory requirement of a specific
4971 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4972
4973 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4974 compiled in such a way that the backtracking after subsequent failure was
4975 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4976 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4977
4978
4979 Version 2.03 02-Feb-99
4980 ----------------------
4981
4982 1. Fixed typo and small mistake in man page.
4983
4984 2. Added 4th condition (GPL supersedes if conflict) and created separate
4985 LICENCE file containing the conditions.
4986
4987 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4988 Perl, that is the internal \ allows the delimiter to be included in the
4989 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4990 the final delimiter, add \ to the end of the pattern (to test the error).
4991
4992 4. Added the convenience functions for extracting substrings after a successful
4993 match. Updated pcretest to make it able to test these functions.
4994
4995
4996 Version 2.02 14-Jan-99
4997 ----------------------
4998
4999 1. Initialized the working variables associated with each extraction so that
5000 their saving and restoring doesn't refer to uninitialized store.
5001
5002 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5003 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5004 fix the problem.
5005
5006 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5007 calls, and wasn't printing the correct value for compiling calls. Increased the
5008 default value of LOOPREPEAT, and the number of significant figures in the
5009 times.
5010
5011 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5012
5013 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5014 a building problem on Windows NT with a FAT file system.
5015
5016
5017 Version 2.01 21-Oct-98
5018 ----------------------
5019
5020 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5021 to character tables built by pcre_maketables() in the current locale. If NULL
5022 is passed, the default tables are used.
5023
5024
5025 Version 2.00 24-Sep-98
5026 ----------------------
5027
5028 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5029 it any more.
5030
5031 2. Allow quantification of (?>) groups, and make it work correctly.
5032
5033 3. The first character computation wasn't working for (?>) groups.
5034
5035 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5036 end of the subject) and add 5.005's \z, which really does match only at the
5037 very end of the subject.
5038
5039 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5040
5041 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5042 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5043 localized options. All options to pcre_study() were also removed.
5044
5045 7. Add other new features from 5.005:
5046
5047 $(?<= positive lookbehind
5048 $(?<! negative lookbehind
5049 (?imsx-imsx) added the unsetting capability
5050 such a setting is global if at outer level; local otherwise
5051 (?imsx-imsx:) non-capturing groups with option setting
5052 (?(cond)re|re) conditional pattern matching
5053
5054 A backreference to itself in a repeated group matches the previous
5055 captured string.
5056
5057 8. General tidying up of studying (both automatic and via "study")
5058 consequential on the addition of new assertions.
5059
5060 9. As in 5.005, unlimited repeated groups that could match an empty substring
5061 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5062 runtime if any iteration does actually match an empty substring.
5063
5064 10. Include the RunTest script in the distribution.
5065
5066 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5067 discrepancies, some of which were old and were also with respect to 5.004. They
5068 have now been fixed.
5069
5070
5071 Version 1.09 28-Apr-98
5072 ----------------------
5073
5074 1. A negated single character class followed by a quantifier with a minimum
5075 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5076 program crashes, or just wrong answers. This did not apply to negated classes
5077 containing more than one character, or to minima other than one.
5078
5079
5080 Version 1.08 27-Mar-98
5081 ----------------------
5082
5083 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5084
5085 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5086 latter must appear before anything that relies on it in the pattern.
5087
5088
5089 Version 1.07 16-Feb-98
5090 ----------------------
5091
5092 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5093 repeat of a potentially empty string).
5094
5095
5096 Version 1.06 23-Jan-98
5097 ----------------------
5098
5099 1. Added Markus Oberhumer's little patches for C++.
5100
5101 2. Literal strings longer than 255 characters were broken.
5102
5103
5104 Version 1.05 23-Dec-97
5105 ----------------------
5106
5107 1. Negated character classes containing more than one character were failing if
5108 PCRE_CASELESS was set at run time.
5109
5110
5111 Version 1.04 19-Dec-97
5112 ----------------------
5113
5114 1. Corrected the man page, where some "const" qualifiers had been omitted.
5115
5116 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5117 input syntax.
5118
5119 3. Fixed memory leak which occurred when a regex with back references was
5120 matched with an offsets vector that wasn't big enough. The temporary memory
5121 that is used in this case wasn't being freed if the match failed.
5122
5123 4. Tidied pcretest to ensure it frees memory that it gets.
5124
5125 5. Temporary memory was being obtained in the case where the passed offsets
5126 vector was exactly big enough.
5127
5128 6. Corrected definition of offsetof() from change 5 below.
5129
5130 7. I had screwed up change 6 below and broken the rules for the use of
5131 setjmp(). Now fixed.
5132
5133
5134 Version 1.03 18-Dec-97
5135 ----------------------
5136
5137 1. A erroneous regex with a missing opening parenthesis was correctly
5138 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5139 on some systems.
5140
5141 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5142 it was reported that one broken compiler failed on the former because "code" is
5143 also an independent variable.
5144
5145 3. The erroneous regex a[]b caused an array overrun reference.
5146
5147 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5148 fail on data ending with that character. (It was going on too far, and checking
5149 the next character, typically a binary zero.) This was specific to the
5150 optimized code for single-character negative classes.
5151
5152 5. Added a contributed patch from the TIN world which does the following:
5153
5154 + Add an undef for memmove, in case the the system defines a macro for it.
5155
5156 + Add a definition of offsetof(), in case there isn't one. (I don't know
5157 the reason behind this - offsetof() is part of the ANSI standard - but
5158 it does no harm).
5159
5160 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5161 most of the places where whitespace preceded '#'. I have given up and
5162 allowed the remaining 2 cases to be at the margin.
5163
5164 + Rename some variables in pcre to eliminate shadowing. This seems very
5165 pedantic, but does no harm, of course.
5166
5167 6. Moved the call to setjmp() into its own function, to get rid of warnings
5168 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5169
5170 7. Constructs such as \d{8,} were compiling into the equivalent of
5171 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5172 outcome, but in this particular case used more store than had been allocated,
5173 which caused the bug to be discovered because it threw up an internal error.
5174
5175 8. The debugging code in both pcre and pcretest for outputting the compiled
5176 form of a regex was going wrong in the case of back references followed by
5177 curly-bracketed repeats.
5178
5179
5180 Version 1.02 12-Dec-97
5181 ----------------------
5182
5183 1. Typos in pcre.3 and comments in the source fixed.
5184
5185 2. Applied a contributed patch to get rid of places where it used to remove
5186 'const' from variables, and fixed some signed/unsigned and uninitialized
5187 variable warnings.
5188
5189 3. Added the "runtest" target to Makefile.
5190
5191 4. Set default compiler flag to -O2 rather than just -O.
5192
5193
5194 Version 1.01 19-Nov-97
5195 ----------------------
5196
5197 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5198 like /([ab]*)*/, that is, for classes with more than one character in them.
5199
5200 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5201 as /((?>a*))*/ (a PCRE_EXTRA facility).
5202
5203
5204 Version 1.00 18-Nov-97
5205 ----------------------
5206
5207 1. Added compile-time macros to support systems such as SunOS4 which don't have
5208 memmove() or strerror() but have other things that can be used instead.
5209
5210 2. Arranged that "make clean" removes the executables.
5211
5212
5213 Version 0.99 27-Oct-97
5214 ----------------------
5215
5216 1. Fixed bug in code for optimizing classes with only one character. It was
5217 initializing a 32-byte map regardless, which could cause it to run off the end
5218 of the memory it had got.
5219
5220 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5221
5222
5223 Version 0.98 22-Oct-97
5224 ----------------------
5225
5226 1. Fixed bug in code for handling temporary memory usage when there are more
5227 back references than supplied space in the ovector. This could cause segfaults.
5228
5229
5230 Version 0.97 21-Oct-97
5231 ----------------------
5232
5233 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5234
5235 2. Optimized negated single characters not to use a bit map.
5236
5237 3. Brought error texts together as macro definitions; clarified some of them;
5238 fixed one that was wrong - it said "range out of order" when it meant "invalid
5239 escape sequence".
5240
5241 4. Changed some char * arguments to const char *.
5242
5243 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5244
5245 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5246 pcretest.
5247
5248
5249 Version 0.96 16-Oct-97
5250 ----------------------
5251
5252 1. Added a simple "pgrep" utility to the distribution.
5253
5254 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5255 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5256 where "ddd" means "one or more decimal digits".
5257
5258 3. Fixed serious bug. If a pattern had a back reference, but the call to
5259 pcre_exec() didn't supply a large enough ovector to record the related
5260 identifying subpattern, the match always failed. PCRE now remembers the number
5261 of the largest back reference, and gets some temporary memory in which to save
5262 the offsets during matching if necessary, in order to ensure that
5263 backreferences always work.
5264
5265 4. Increased the compatibility with Perl in a number of ways:
5266
5267 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5268 to request this handling. The option can be set at compile or exec time.
5269
5270 (b) $ matches before a terminating newline by default; an option
5271 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5272 mode). The option can be set at compile or exec time.
5273
5274 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5275 the same as Perl's. If the decimal number it represents is less than 10
5276 or there aren't that many previous left capturing parentheses, an octal
5277 escape is read. Inside a character class, it's always an octal escape,
5278 even if it is a single digit.
5279
5280 (d) An escaped but undefined alphabetic character is taken as a literal,
5281 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5282 escapes.
5283
5284 (e) {0} is now permitted. (The previous item is removed from the compiled
5285 pattern).
5286
5287 5. Changed all the names of code files so that the basic parts are no longer
5288 than 10 characters, and abolished the teeny "globals.c" file.
5289
5290 6. Changed the handling of character classes; they are now done with a 32-byte
5291 bit map always.
5292
5293 7. Added the -d and /D options to pcretest to make it possible to look at the
5294 internals of compilation without having to recompile pcre.
5295
5296
5297 Version 0.95 23-Sep-97
5298 ----------------------
5299
5300 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5301 \x20 at the start of a run of normal characters. These were being treated as
5302 real characters, instead of the source characters being re-checked.
5303
5304
5305 Version 0.94 18-Sep-97
5306 ----------------------
5307
5308 1. The functions are now thread-safe, with the caveat that the global variables
5309 containing pointers to malloc() and free() or alternative functions are the
5310 same for all threads.
5311
5312 2. Get pcre_study() to generate a bitmap of initial characters for non-
5313 anchored patterns when this is possible, and use it if passed to pcre_exec().
5314
5315
5316 Version 0.93 15-Sep-97
5317 ----------------------
5318
5319 1. /(b)|(:+)/ was computing an incorrect first character.
5320
5321 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5322 but not actually doing anything yet.
5323
5324 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5325 as Perl does (e.g. [-az] or [az-]).
5326
5327 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5328 all possible positions.
5329
5330 5. Split up into different modules to avoid including unneeded functions in a
5331 compiled binary. However, compile and exec are still in one module. The "study"
5332 function is split off.
5333
5334 6. The character tables are now in a separate module whose source is generated
5335 by an auxiliary program - but can then be edited by hand if required. There are
5336 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5337 toupper() in the code.
5338
5339 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5340 make them global. Abolish the function for setting them, as the caller can now
5341 set them directly.
5342
5343
5344 Version 0.92 11-Sep-97
5345 ----------------------
5346
5347 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5348 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5349
5350 2. Caseless matching was not working in character classes if the characters in
5351 the pattern were in upper case.
5352
5353 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5354
5355 4. Make PCRE_ANCHORED public and accept as a compile option.
5356
5357 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5358 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5359 pass them.
5360
5361 6. Give an error if bad option bits passed at compile or run time.
5362
5363 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5364 pcretest to cause it to pass that flag.
5365
5366 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5367 options, and the first character, if set.
5368
5369 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5370
5371
5372 Version 0.91 10-Sep-97
5373 ----------------------
5374
5375 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5376 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5377
5378 2. PCRE was looping on encountering an indefinitely repeated back reference to
5379 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5380 Perl does - treats the match as successful.
5381
5382 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5