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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1425 - (show annotations)
Tue Dec 31 17:44:40 2013 UTC (5 years, 10 months ago) by ph10
File size: 259721 byte(s)
Error occurred while calculating annotation data.
Get rid of confusing RAWUCHAR set of macros.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.35-RC1 xx-xxxx-201x
5 -----------------------------
6
7 1. A new flag is set, when property checks are present in an XCLASS.
8 When this flag is not set, PCRE can perform certain optimizations
9 such as studying these XCLASS-es.
10
11 2. The auto-possessification of character sets were improved: a normal
12 and an extended character set can be compared now. Furthermore
13 the JIT compiler optimizes more character set checks.
14
15 3. Got rid of some compiler warnings for potentially uninitialized variables
16 that show up only when compiled with -O2.
17
18 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
19 of a match later then the end of the match. The pcretest program was not
20 handling the case sensibly - it was outputting from the start to the next
21 binary zero. It now reports this situation in a message, and outputs the
22 text from the end to the start.
23
24 5. Fast forward search is improved in JIT. Instead of the first three
25 characters, any three characters with fixed position can be searched.
26 Search order: first, last, middle.
27
28 6. Improve character range checks in JIT. Characters are read by an inprecise
29 function now, which returns with an unknown value if the character code is
30 above a certain treshold (e.g: 256). The only limitation is that the value
31 must be bigger than the treshold as well. This function is useful, when
32 the characters above the treshold are handled in the same way.
33
34 7. The macros RAWUCHAR and RAWUCHARTEST were identical (and the latter was not
35 testing anything, contrary to its name and comment) and were not really
36 fulfilling any useful function, so I have replaced their use by plain code.
37 Similarly for RAWUCHARINC and RAWUCHARINCTEST.
38
39
40 Version 8.34 15-December-2013
41 -----------------------------
42
43 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
44 executable memory. Patch inspired by Carsten Klein.
45
46 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
47 this macro is never tested and has no effect, because the work to support
48 coverage involves only compiling and linking options and special targets in
49 the Makefile. The comment in config.h implied that defining the macro would
50 enable coverage support, which is totally false. There was also support for
51 setting this macro in the CMake files (my fault, I just copied it from
52 configure). SUPPORT_GCOV has now been removed.
53
54 3. Make a small performance improvement in strlen16() and strlen32() in
55 pcretest.
56
57 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
58 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
59 these cases). There was also one in pcretest.c.
60
61 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
62
63 6. In UTF mode, the code for checking whether a group could match an empty
64 string (which is used for indefinitely repeated groups to allow for
65 breaking an infinite loop) was broken when the group contained a repeated
66 negated single-character class with a character that occupied more than one
67 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
68 UTF-8 mode). The effect was undefined: the group might or might not be
69 deemed as matching an empty string, or the program might have crashed.
70
71 7. The code for checking whether a group could match an empty string was not
72 recognizing that \h, \H, \v, \V, and \R must match a character.
73
74 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
75 an empty string. If it can, pcretest shows this in its information output.
76
77 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
78 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
79 matched by pcre_exec() without using JIT:
80
81 (a) If the rest of the pattern did not match after a maximal run of
82 grapheme clusters, the code for backing up to try with fewer of them
83 did not always back up over a full grapheme when characters that do not
84 have the modifier quality were involved, e.g. Hangul syllables.
85
86 (b) If the match point in a subject started with modifier character, and
87 there was no match, the code could incorrectly back up beyond the match
88 point, and potentially beyond the first character in the subject,
89 leading to a segfault or an incorrect match result.
90
91 10. A conditional group with an assertion condition could lead to PCRE
92 recording an incorrect first data item for a match if no other first data
93 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
94 first data item, and therefore matched "ca" after "c" instead of at the
95 start.
96
97 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
98 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
99
100 12. The source of pcregrep now includes z/OS-specific code so that it can be
101 compiled for z/OS as part of the special z/OS distribution.
102
103 13. Added the -T and -TM options to pcretest.
104
105 14. The code in pcre_compile.c for creating the table of named capturing groups
106 has been refactored. Instead of creating the table dynamically during the
107 actual compiling pass, the information is remembered during the pre-compile
108 pass (on the stack unless there are more than 20 named groups, in which
109 case malloc() is used) and the whole table is created before the actual
110 compile happens. This has simplified the code (it is now nearly 150 lines
111 shorter) and prepared the way for better handling of references to groups
112 with duplicate names.
113
114 15. A back reference to a named subpattern when there is more than one of the
115 same name now checks them in the order in which they appear in the pattern.
116 The first one that is set is used for the reference. Previously only the
117 first one was inspected. This change makes PCRE more compatible with Perl.
118
119 16. Unicode character properties were updated from Unicode 6.3.0.
120
121 17. The compile-time code for auto-possessification has been refactored, based
122 on a patch by Zoltan Herczeg. It now happens after instead of during
123 compilation. The code is cleaner, and more cases are handled. The option
124 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
125 options in pcretest are provided to set it. It can also be set by
126 (*NO_AUTO_POSSESS) at the start of a pattern.
127
128 18. The character VT has been added to the default ("C" locale) set of
129 characters that match \s and are generally treated as white space,
130 following this same change in Perl 5.18. There is now no difference between
131 "Perl space" and "POSIX space". Whether VT is treated as white space in
132 other locales depends on the locale.
133
134 19. The code for checking named groups as conditions, either for being set or
135 for being recursed, has been refactored (this is related to 14 and 15
136 above). Processing unduplicated named groups should now be as fast at
137 numerical groups, and processing duplicated groups should be faster than
138 before.
139
140 20. Two patches to the CMake build system, by Alexander Barkov:
141
142 (1) Replace the "source" command by "." in CMakeLists.txt because
143 "source" is a bash-ism.
144
145 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
146 without these the CMake build does not work on Solaris.
147
148 21. Perl has changed its handling of \8 and \9. If there is no previously
149 encountered capturing group of those numbers, they are treated as the
150 literal characters 8 and 9 instead of a binary zero followed by the
151 literals. PCRE now does the same.
152
153 22. Following Perl, added \o{} to specify codepoints in octal, making it
154 possible to specify values greater than 0777 and also making them
155 unambiguous.
156
157 23. Perl now gives an error for missing closing braces after \x{... instead of
158 treating the string as literal. PCRE now does the same.
159
160 24. RunTest used to grumble if an inappropriate test was selected explicitly,
161 but just skip it when running all tests. This make it awkward to run ranges
162 of tests when one of them was inappropriate. Now it just skips any
163 inappropriate tests, as it always did when running all tests.
164
165 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
166 character types such as \d or \w, too many callouts were inserted, and the
167 data that they returned was rubbish.
168
169 26. In UCP mode, \s was not matching two of the characters that Perl matches,
170 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
171 were matched by \h. The code has now been refactored so that the lists of
172 the horizontal and vertical whitespace characters used for \h and \v (which
173 are defined only in one place) are now also used for \s.
174
175 27. Add JIT support for the 64 bit TileGX architecture.
176 Patch by Jiong Wang (Tilera Corporation).
177
178 28. Possessive quantifiers for classes (both explicit and automatically
179 generated) now use special opcodes instead of wrapping in ONCE brackets.
180
181 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
182 (because it's meaningless), this was not happening when PCRE_CASELESS was
183 set. Not wrong, but inefficient.
184
185 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
186 properties for \w, \d, etc) is present in a test regex. Otherwise if the
187 test contains no characters greater than 255, Perl doesn't realise it
188 should be using Unicode semantics.
189
190 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
191 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
192 does in Unicode mode.
193
194 32. Added the "forbid" facility to pcretest so that putting tests into the
195 wrong test files can sometimes be quickly detected.
196
197 33. There is now a limit (default 250) on the depth of nesting of parentheses.
198 This limit is imposed to control the amount of system stack used at compile
199 time. It can be changed at build time by --with-parens-nest-limit=xxx or
200 the equivalent in CMake.
201
202 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
203 errors. Perl warns for these when in warning mode, but PCRE has no facility
204 for giving warnings.
205
206 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
207 However, this was not working for (?!) because it is optimized to (*FAIL),
208 for which PCRE does not allow quantifiers. The optimization is now disabled
209 when a quantifier follows (?!). I can't see any use for this, but it makes
210 things uniform.
211
212 36. Perl no longer allows group names to start with digits, so I have made this
213 change also in PCRE. It simplifies the code a bit.
214
215 37. In extended mode, Perl ignores spaces before a + that indicates a
216 possessive quantifier. PCRE allowed a space before the quantifier, but not
217 before the possessive +. It now does.
218
219 38. The use of \K (reset reported match start) within a repeated possessive
220 group such as (a\Kb)*+ was not working.
221
222 40. Document that the same character tables must be used at compile time and
223 run time, and that the facility to pass tables to pcre_exec() and
224 pcre_dfa_exec() is for use only with saved/restored patterns.
225
226 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
227 features for Builds with MSVC:
228
229 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
230 builds). The .rc files can be used to set FileDescription and many other
231 attributes.
232
233 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
234 This allows higher-level build scripts which want .pdb files to avoid
235 hard-coding the exact files needed."
236
237 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
238 mean "start of word" and "end of word", respectively, as a transition aid.
239
240 43. A minimizing repeat of a class containing codepoints greater than 255 in
241 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
242 compiled to use the heap for recursion.
243
244 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
245 is configured.
246
247
248 Version 8.33 28-May-2013
249 ------------------------
250
251 1. Added 'U' to some constants that are compared to unsigned integers, to
252 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
253 variables that are added to signed variables, to ensure the result is
254 signed and can be negated.
255
256 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
257 CMake config files.
258
259 3. Revise the creation of config.h.generic so that all boolean macros are
260 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
261 overriding via -D on the command line possible.
262
263 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
264 to unsigned int is reported to make a quite noticeable speed difference in
265 a specific Windows environment. Testing on Linux did also appear to show
266 some benefit (and it is clearly not harmful). Also fixed the definition of
267 Xop which should be unsigned.
268
269 5. Related to (4), changing the definition of the intermediate variable cc
270 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
271 improvements.
272
273 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
274 unnecessary spaces.
275
276 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
277 and later.
278
279 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
280
281 9. Optimizing fast_forward_start_bits in JIT.
282
283 10. Adding support for callouts in JIT, and fixing some issues revealed
284 during this work. Namely:
285
286 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
287
288 (b) Minimum length was not checked before the matching is started.
289
290 11. The value of capture_last that is passed to callouts was incorrect in some
291 cases when there was a capture on one path that was subsequently abandoned
292 after a backtrack. Also, the capture_last value is now reset after a
293 recursion, since all captures are also reset in this case.
294
295 12. The interpreter no longer returns the "too many substrings" error in the
296 case when an overflowing capture is in a branch that is subsequently
297 abandoned after a backtrack.
298
299 13. In the pathological case when an offset vector of size 2 is used, pcretest
300 now prints out the matched string after a yield of 0 or 1.
301
302 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
303 Only supported by the JIT compiler at the moment.
304
305 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
306
307 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
308 offset of the starting point of the matching process, provided the offsets
309 vector is large enough.
310
311 17. The \A escape now records a lookbehind value of 1, though its execution
312 does not actually inspect the previous character. This is to ensure that,
313 in partial multi-segment matching, at least one character from the old
314 segment is retained when a new segment is processed. Otherwise, if there
315 are no lookbehinds in the pattern, \A might match incorrectly at the start
316 of a new segment.
317
318 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
319
320 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
321 gives some modest performance improvement in 8-bit mode.
322
323 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
324 be expressed in certain programming languages using Universal Character
325 Names.
326
327 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
328 which points out that "non characters" are not "characters that may not
329 appear in Unicode strings" but rather "characters that are reserved for
330 internal use and have only local meaning".
331
332 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
333 there was a conditional group that depended on an assertion, if the
334 assertion was false, the callout that immediately followed the alternation
335 in the condition was skipped when pcre_exec() was used for matching.
336
337 23. Allow an explicit callout to be inserted before an assertion that is the
338 condition for a conditional group, for compatibility with automatic
339 callouts, which always insert a callout at this point.
340
341 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
342 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
343
344 25. (*PRUNE) is now supported by the JIT compiler.
345
346 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
347
348 27. Fix the case where there are two or more SKIPs with arguments that may be
349 ignored.
350
351 28. (*SKIP) is now supported by the JIT compiler.
352
353 29. (*THEN) is now supported by the JIT compiler.
354
355 30. Update RunTest with additional test selector options.
356
357 31. The way PCRE handles backtracking verbs has been changed in two ways.
358
359 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
360 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
361 backtracking. In some cases this makes it more Perl-compatible, but Perl's
362 rather obscure rules do not always do the same thing.
363
364 (2) Previously, backtracking verbs were confined within assertions. This is
365 no longer the case for positive assertions, except for (*ACCEPT). Again,
366 this sometimes improves Perl compatibility, and sometimes does not.
367
368 32. A number of tests that were in test 2 because Perl did things differently
369 have been moved to test 1, because either Perl or PCRE has changed, and
370 these tests are now compatible.
371
372 32. Backtracking control verbs are now handled in the same way in JIT and
373 interpreter.
374
375 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
376 contained a forward subroutine reference caused a compile error.
377
378 34. Auto-detect and optimize limited repetitions in JIT.
379
380 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
381 blocking (*UTF) etc.
382
383 36. In the interpreter, maximizing pattern repetitions for characters and
384 character types now use tail recursion, which reduces stack usage.
385
386 37. The value of the max lookbehind was not correctly preserved if a compiled
387 and saved regex was reloaded on a host of different endianness.
388
389 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
390 of the compiled pattern block, expand the flags field from 16 to 32 bits
391 because it was almost full.
392
393 39. Try madvise first before posix_madvise.
394
395 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
396 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
397 original change locked out all matches of empty strings. This has been
398 changed so that one match of an empty string per line is recognized.
399 Subsequent searches on the same line (for colouring or for --only-matching,
400 for example) do not recognize empty strings.
401
402 41. Applied a user patch to fix a number of spelling mistakes in comments.
403
404 42. Data lines longer than 65536 caused pcretest to crash.
405
406 43. Clarified the data type for length and startoffset arguments for pcre_exec
407 and pcre_dfa_exec in the function-specific man pages, where they were
408 explicitly stated to be in bytes, never having been updated. I also added
409 some clarification to the pcreapi man page.
410
411 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
412 a segmentation fault.
413
414
415 Version 8.32 30-November-2012
416 -----------------------------
417
418 1. Improved JIT compiler optimizations for first character search and single
419 character iterators.
420
421 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
422 Patch by Daniel Richard G.
423
424 3. Single character iterator optimizations in the JIT compiler.
425
426 4. Improved JIT compiler optimizations for character ranges.
427
428 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
429 Reported by Giuseppe D'Angelo.
430
431 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
432 of a line, was being set incorrectly in cases where .* appeared inside
433 atomic brackets at the start of a pattern, or where there was a subsequent
434 *PRUNE or *SKIP.
435
436 7. Improved instruction cache flush for POWER/PowerPC.
437 Patch by Daniel Richard G.
438
439 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
440 grep:
441
442 (a) There is now no limit to the number of patterns to be matched.
443
444 (b) An error is given if a pattern is too long.
445
446 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
447 are now supported.
448
449 (d) --exclude-from and --include-from (multiple use) have been added.
450
451 (e) Exclusions and inclusions now apply to all files and directories, not
452 just to those obtained from scanning a directory recursively.
453
454 (f) Multiple uses of -f and --file-list are now supported.
455
456 (g) In a Windows environment, the default for -d has been changed from
457 "read" (the GNU grep default) to "skip", because otherwise the presence
458 of a directory in the file list provokes an error.
459
460 (h) The documentation has been revised and clarified in places.
461
462 9. Improve the matching speed of capturing brackets.
463
464 10. Changed the meaning of \X so that it now matches a Unicode extended
465 grapheme cluster.
466
467 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
468 out POSIX threads when JIT support is configured.
469
470 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
471
472 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
473 structure could go wrong in environments where size_t is not the same size
474 as int.
475
476 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
477 set.
478
479 15. The EBCDIC support had decayed; later updates to the code had included
480 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
481 general tidy up of EBCDIC-related issues, and the documentation was also
482 not quite right. There is now a test that can be run on ASCII systems to
483 check some of the EBCDIC-related things (but is it not a full test).
484
485 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
486 in a small tidy to the code.
487
488 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
489
490 18. If the --only-matching (-o) option in pcregrep is specified multiple
491 times, each one causes appropriate output. For example, -o1 -o2 outputs the
492 substrings matched by the 1st and 2nd capturing parentheses. A separating
493 string can be specified by --om-separator (default empty).
494
495 19. Improving the first n character searches.
496
497 20. Turn case lists for horizontal and vertical white space into macros so that
498 they are defined only once.
499
500 21. This set of changes together give more compatible Unicode case-folding
501 behaviour for characters that have more than one other case when UCP
502 support is available.
503
504 (a) The Unicode property table now has offsets into a new table of sets of
505 three or more characters that are case-equivalent. The MultiStage2.py
506 script that generates these tables (the pcre_ucd.c file) now scans
507 CaseFolding.txt instead of UnicodeData.txt for character case
508 information.
509
510 (b) The code for adding characters or ranges of characters to a character
511 class has been abstracted into a generalized function that also handles
512 case-independence. In UTF-mode with UCP support, this uses the new data
513 to handle characters with more than one other case.
514
515 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
516 whose other case is greater than 256 are now correctly matched
517 caselessly. Previously, the high codepoint matched the low one, but not
518 vice versa.
519
520 (d) The processing of \h, \H, \v, and \ in character classes now makes use
521 of the new class addition function, using character lists defined as
522 macros alongside the case definitions of 20 above.
523
524 (e) Caseless back references now work with characters that have more than
525 one other case.
526
527 (f) General caseless matching of characters with more than one other case
528 is supported.
529
530 22. Unicode character properties were updated from Unicode 6.2.0
531
532 23. Improved CMake support under Windows. Patch by Daniel Richard G.
533
534 24. Add support for 32-bit character strings, and UTF-32
535
536 25. Major JIT compiler update (code refactoring and bugfixing).
537 Experimental Sparc 32 support is added.
538
539 26. Applied a modified version of Daniel Richard G's patch to create
540 pcre.h.generic and config.h.generic by "make" instead of in the
541 PrepareRelease script.
542
543 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
544 pcre_compile.c when checking for a zero character.
545
546 28. Introducing a native interface for JIT. Through this interface, the compiled
547 machine code can be directly executed. The purpose of this interface is to
548 provide fast pattern matching, so several sanity checks are not performed.
549 However, feature tests are still performed. The new interface provides
550 1.4x speedup compared to the old one.
551
552 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
553 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
554 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
555
556 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
557 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
558 the "old" RFC 2279). Instead, it ended up passing a negative length to
559 pcre_exec().
560
561 31. Add support for GCC's visibility feature to hide internal functions.
562
563 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
564 "unknown -C option" after outputting 0 or 1.
565
566 33. There is now support for generating a code coverage report for the test
567 suite in environments where gcc is the compiler and lcov is installed. This
568 is mainly for the benefit of the developers.
569
570 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
571 unaddressable using valgrind annotations, allowing valgrind to detect
572 invalid memory accesses. This is mainly for the benefit of the developers.
573
574 25. (*UTF) can now be used to start a pattern in any of the three libraries.
575
576 26. Give configure error if --enable-cpp but no C++ compiler found.
577
578
579 Version 8.31 06-July-2012
580 -------------------------
581
582 1. Fixing a wrong JIT test case and some compiler warnings.
583
584 2. Removed a bashism from the RunTest script.
585
586 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
587 to unsigned type, result still unsigned" that was given by an MS compiler
588 on encountering the code "-sizeof(xxx)".
589
590 4. Partial matching support is added to the JIT compiler.
591
592 5. Fixed several bugs concerned with partial matching of items that consist
593 of more than one character:
594
595 (a) /^(..)\1/ did not partially match "aba" because checking references was
596 done on an "all or nothing" basis. This also applied to repeated
597 references.
598
599 (b) \R did not give a hard partial match if \r was found at the end of the
600 subject.
601
602 (c) \X did not give a hard partial match after matching one or more
603 characters at the end of the subject.
604
605 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
606 a partial match for the string "\r".
607
608 (e) When newline was set to CRLF, the metacharacter "." did not recognize
609 a partial match for a CR character at the end of the subject string.
610
611 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
612 running pcretest, the text "(JIT)" added to the output whenever JIT is
613 actually used to run the match.
614
615 7. Individual JIT compile options can be set in pcretest by following -s+[+]
616 or /S+[+] with a digit between 1 and 7.
617
618 8. OP_NOT now supports any UTF character not just single-byte ones.
619
620 9. (*MARK) control verb is now supported by the JIT compiler.
621
622 10. The command "./RunTest list" lists the available tests without actually
623 running any of them. (Because I keep forgetting what they all are.)
624
625 11. Add PCRE_INFO_MAXLOOKBEHIND.
626
627 12. Applied a (slightly modified) user-supplied patch that improves performance
628 when the heap is used for recursion (compiled with --disable-stack-for-
629 recursion). Instead of malloc and free for each heap frame each time a
630 logical recursion happens, frames are retained on a chain and re-used where
631 possible. This sometimes gives as much as 30% improvement.
632
633 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
634 call.
635
636 14. As documented, (*COMMIT) is now confined to within a positive assertion.
637
638 15. It is now possible to link pcretest with libedit as an alternative to
639 libreadline.
640
641 16. (*COMMIT) control verb is now supported by the JIT compiler.
642
643 17. The Unicode data tables have been updated to Unicode 6.1.0.
644
645 18. Added --file-list option to pcregrep.
646
647 19. Added binary file support to pcregrep, including the -a, --binary-files,
648 -I, and --text options.
649
650 20. The madvise function is renamed for posix_madvise for QNX compatibility
651 reasons. Fixed by Giuseppe D'Angelo.
652
653 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
654
655 22. Changed the option for creating symbolic links for 16-bit man pages from
656 -s to -sf so that re-installing does not cause issues.
657
658 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
659
660 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
661 matches in certain environments (the workspace was not being correctly
662 retained). Also added to pcre_dfa_exec() a simple plausibility check on
663 some of the workspace data at the beginning of a restart.
664
665 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
666 was not doing so when it should - probably a typo introduced by SVN 528
667 (change 8.10/14).
668
669 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
670 \w+ when the character tables indicated that \x{c4} was a word character.
671 There were several related cases, all because the tests for doing a table
672 lookup were testing for characters less than 127 instead of 255.
673
674 27. If a pattern contains capturing parentheses that are not used in a match,
675 their slots in the ovector are set to -1. For those that are higher than
676 any matched groups, this happens at the end of processing. In the case when
677 there were back references that the ovector was too small to contain
678 (causing temporary malloc'd memory to be used during matching), and the
679 highest capturing number was not used, memory off the end of the ovector
680 was incorrectly being set to -1. (It was using the size of the temporary
681 memory instead of the true size.)
682
683 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
684 ovector size, it uses memory at the end of the block that it has got.
685
686 29. Check for an overlong MARK name and give an error at compile time. The
687 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
688
689 30. JIT compiler update.
690
691 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
692 Rill for the patch.
693
694 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
695
696 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
697
698 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
699 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
700 was enabled.
701
702 35. Improve JIT code generation for greedy plus quantifier.
703
704 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
705 1 to "aa" instead of to an empty string. The bug affected repeated groups
706 that could potentially match an empty string.
707
708 37. Optimizing single character iterators in JIT.
709
710 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
711 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
712 codepoints that are too big for the mode are faulted, and in a UTF mode,
713 disallowed codepoints are also faulted.
714
715 39. If PCRE was compiled with UTF support, in three places in the DFA
716 matcher there was code that should only have been obeyed in UTF mode, but
717 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
718 processing when bytes with values greater than 127 were present. In 16-bit
719 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
720 both cases the values are those that cannot be the first data item in a UTF
721 character. The three items that might have provoked this were recursions,
722 possessively repeated groups, and atomic groups.
723
724 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
725 and pcregrep, because some OS require shared objects to be explicitly
726 passed to ld, causing the link step to fail if they are not.
727
728 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
729 mode, patterns that started with \h* or \R* might be incorrectly matched.
730
731
732 Version 8.30 04-February-2012
733 -----------------------------
734
735 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
736 name is defined in ctype.h.
737
738 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
739 only in quite long subpatterns.
740
741 3. Removed the function pcre_info(), which has been obsolete and deprecated
742 since it was replaced by pcre_fullinfo() in February 2000.
743
744 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
745 match a (*MARK), and the match failed at the start of the subject, a
746 reference to memory before the start of the subject could occur. This bug
747 was introduced by fix 17 of release 8.21.
748
749 5. A reference to an unset group with zero minimum repetition was giving
750 totally wrong answers (in non-JavaScript-compatibility mode). For example,
751 /(another)?(\1?)test/ matched against "hello world test". This bug was
752 introduced in release 8.13.
753
754 6. Add support for 16-bit character strings (a large amount of work involving
755 many changes and refactorings).
756
757 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
758 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
759 from a file.
760
761 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
762 rounding is not applied in this particular case).
763
764 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
765 if they appear, or are escaped, in patterns.
766
767 10. Get rid of a number of -Wunused-but-set-variable warnings.
768
769 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
770 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
771 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
772 also returns the mark "x". This bug applied to capturing parentheses,
773 non-capturing parentheses, and atomic parentheses. It also applied to some
774 assertions.
775
776 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
777 information out of configure.ac instead of relying on pcre.h.generic, which
778 is not stored in the repository.
779
780 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
781 -lreadline.
782
783 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
784
785 15. Applied Graycode's patch to put the top-level frame on the stack rather
786 than the heap when not using the stack for recursion. This gives a
787 performance improvement in many cases when recursion is not deep.
788
789 16. Experimental code added to "pcretest -C" to output the stack frame size.
790
791
792 Version 8.21 12-Dec-2011
793 ------------------------
794
795 1. Updating the JIT compiler.
796
797 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
798 are added as well.
799
800 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
801 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
802 calling _pcre_jit_exec. Some extra comments are added.
803
804 4. (*MARK) settings inside atomic groups that do not contain any capturing
805 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
806 was introduced by change 18 for 8.20.
807
808 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
809 ECMA-262 standard.
810
811 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
812 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
813 This bug was probably introduced by change 9 of 8.13.
814
815 7. While fixing 6 above, I noticed that a number of other items were being
816 incorrectly rejected as "not fixed length". This arose partly because newer
817 opcodes had not been added to the fixed-length checking code. I have (a)
818 corrected the bug and added tests for these items, and (b) arranged for an
819 error to occur if an unknown opcode is encountered while checking for fixed
820 length instead of just assuming "not fixed length". The items that were
821 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
822 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
823 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
824
825 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
826 being incorrectly compiled and would have given unpredicatble results.
827
828 9. A possessively repeated subpattern with minimum repeat count greater than
829 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
830 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
831 the first (A) could occur when it should not.
832
833 10. Add a cast and remove a redundant test from the code.
834
835 11. JIT should use pcre_malloc/pcre_free for allocation.
836
837 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
838 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
839 is anything other than /usr, -L is still shown).
840
841 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
842
843 14. Perl does not support \N without a following name in a [] class; PCRE now
844 also gives an error.
845
846 15. If a forward reference was repeated with an upper limit of around 2000,
847 it caused the error "internal error: overran compiling workspace". The
848 maximum number of forward references (including repeats) was limited by the
849 internal workspace, and dependent on the LINK_SIZE. The code has been
850 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
851 the default depends on LINK_SIZE. There is a new upper limit (for safety)
852 of around 200,000 forward references. While doing this, I also speeded up
853 the filling in of repeated forward references.
854
855 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
856 incorrectly expecting the subject to contain another "a" after the start.
857
858 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
859 in the match, the SKIP should be ignored. This was not happening; instead
860 the SKIP was being treated as NOMATCH. For patterns such as
861 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
862 tested.
863
864 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
865 now much more compatible with Perl, in particular in cases where the result
866 is a non-match for a non-anchored pattern. For example, if
867 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
868 "m", where previously it did not return a name. A side effect of this
869 change is that for partial matches, the last encountered mark name is
870 returned, as for non matches. A number of tests that were previously not
871 Perl-compatible have been moved into the Perl-compatible test files. The
872 refactoring has had the pleasing side effect of removing one argument from
873 the match() function, thus reducing its stack requirements.
874
875 19. If the /S+ option was used in pcretest to study a pattern using JIT,
876 subsequent uses of /S (without +) incorrectly behaved like /S+.
877
878 21. Retrieve executable code size support for the JIT compiler and fixing
879 some warnings.
880
881 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
882 not work when the shorter character appeared right at the end of the
883 subject string.
884
885 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
886 systems.
887
888 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
889 output it when the /M option is used in pcretest.
890
891 25. The CheckMan script was not being included in the distribution. Also, added
892 an explicit "perl" to run Perl scripts from the PrepareRelease script
893 because this is reportedly needed in Windows.
894
895 26. If study data was being save in a file and studying had not found a set of
896 "starts with" bytes for the pattern, the data written to the file (though
897 never used) was taken from uninitialized memory and so caused valgrind to
898 complain.
899
900 27. Updated RunTest.bat as provided by Sheri Pierce.
901
902 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
903
904 29. Computation of memory usage for the table of capturing group names was
905 giving an unnecessarily large value.
906
907
908 Version 8.20 21-Oct-2011
909 ------------------------
910
911 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
912 a POSIX class. After further experiments with Perl, which convinced me that
913 Perl has bugs and confusions, a closing square bracket is no longer allowed
914 in a POSIX name. This bug also affected patterns with classes that started
915 with full stops.
916
917 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
918 captured substring, but while checking the failing first alternative,
919 substring 1 is temporarily captured. If the output vector supplied to
920 pcre_exec() was not big enough for this capture, the yield of the function
921 was still zero ("insufficient space for captured substrings"). This cannot
922 be totally fixed without adding another stack variable, which seems a lot
923 of expense for a edge case. However, I have improved the situation in cases
924 such as /(a)(b)x|abc/ matched against "abc", where the return code
925 indicates that fewer than the maximum number of slots in the ovector have
926 been set.
927
928 3. Related to (2) above: when there are more back references in a pattern than
929 slots in the output vector, pcre_exec() uses temporary memory during
930 matching, and copies in the captures as far as possible afterwards. It was
931 using the entire output vector, but this conflicts with the specification
932 that only 2/3 is used for passing back captured substrings. Now it uses
933 only the first 2/3, for compatibility. This is, of course, another edge
934 case.
935
936 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
937 main code base, and can be used by building with --enable-jit. When this is
938 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
939 runtime --no-jit option is given.
940
941 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
942 ovector, the return from the function was zero, implying that there were
943 other matches that did not fit. The correct "exactly full" value is now
944 returned.
945
946 6. If a subpattern that was called recursively or as a subroutine contained
947 (*PRUNE) or any other control that caused it to give a non-standard return,
948 invalid errors such as "Error -26 (nested recursion at the same subject
949 position)" or even infinite loops could occur.
950
951 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
952 computing the minimum length on reaching *ACCEPT, and so ended up with the
953 wrong value of 1 rather than 0. Further investigation indicates that
954 computing a minimum subject length in the presence of *ACCEPT is difficult
955 (think back references, subroutine calls), and so I have changed the code
956 so that no minimum is registered for a pattern that contains *ACCEPT.
957
958 8. If (*THEN) was present in the first (true) branch of a conditional group,
959 it was not handled as intended. [But see 16 below.]
960
961 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
962 Sheri Pierce.
963
964 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
965 the first byte in a match must be "a".
966
967 11. Change 17 for 8.13 increased the recursion depth for patterns like
968 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
969 pattern contains any instances of (*THEN). If it does not, the old
970 optimizations are restored. It would be nice to do this on a per-group
971 basis, but at the moment that is not feasible.
972
973 12. In some environments, the output of pcretest -C is CRLF terminated. This
974 broke RunTest's code that checks for the link size. A single white space
975 character after the value is now allowed for.
976
977 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
978 For "fr", it uses the Windows-specific input and output files.
979
980 14. If (*THEN) appeared in a group that was called recursively or as a
981 subroutine, it did not work as intended. [But see next item.]
982
983 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
984 pattern fragments (but not containing any | characters). If A and B are
985 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
986 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
987 D. In other words, Perl considers parentheses that do not contain any |
988 characters to be part of a surrounding alternative, whereas PCRE was
989 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
990 differently. PCRE now behaves in the same way as Perl, except in the case
991 of subroutine/recursion calls such as (?1) which have in any case always
992 been different (but PCRE had them first :-).
993
994 16. Related to 15 above: Perl does not treat the | in a conditional group as
995 creating alternatives. Such a group is treated in the same way as an
996 ordinary group without any | characters when processing (*THEN). PCRE has
997 been changed to match Perl's behaviour.
998
999 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1000 RunGrepTest script failed.
1001
1002 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1003 inevitable for groups that contain captures, but it can lead to a lot of
1004 stack use in large patterns. The old behaviour has been restored for atomic
1005 groups that do not contain any capturing parentheses.
1006
1007 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1008 suppress the check for a minimum subject length at run time. (If it was
1009 given to pcre_exec() or pcre_dfa_exec() it did work.)
1010
1011 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1012 fail to work when decoding hex characters in data strings in EBCDIC
1013 environments.
1014
1015 21. It appears that in at least one Mac OS environment, the isxdigit() function
1016 is implemented as a macro that evaluates to its argument more than once,
1017 contravening the C 90 Standard (I haven't checked a later standard). There
1018 was an instance in pcretest which caused it to go wrong when processing
1019 \x{...} escapes in subject strings. The has been rewritten to avoid using
1020 things like p++ in the argument of isxdigit().
1021
1022
1023 Version 8.13 16-Aug-2011
1024 ------------------------
1025
1026 1. The Unicode data tables have been updated to Unicode 6.0.0.
1027
1028 2. Two minor typos in pcre_internal.h have been fixed.
1029
1030 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1031 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1032 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1033
1034 4. There were a number of related bugs in the code for matching backrefences
1035 caselessly in UTF-8 mode when codes for the characters concerned were
1036 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1037 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1038 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1039 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1040 2-byte code at the end of the subject (it thought there wasn't enough data
1041 left).
1042
1043 5. Comprehensive information about what went wrong is now returned by
1044 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1045 as the output vector has at least 2 elements. The offset of the start of
1046 the failing character and a reason code are placed in the vector.
1047
1048 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1049 now returned is for the first byte of the failing character, instead of the
1050 last byte inspected. This is an incompatible change, but I hope it is small
1051 enough not to be a problem. It makes the returned offset consistent with
1052 pcre_exec() and pcre_dfa_exec().
1053
1054 7. pcretest now gives a text phrase as well as the error number when
1055 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1056 failure, the offset and reason code are output.
1057
1058 8. When \R was used with a maximizing quantifier it failed to skip backwards
1059 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1060 back over a single character (\n). This seems wrong (because it treated the
1061 two characters as a single entity when going forwards), conflicts with the
1062 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1063 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1064 has been changed.
1065
1066 9. Some internal refactoring has changed the processing so that the handling
1067 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1068 time (the PCRE_DOTALL option was changed this way some time ago: version
1069 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1070 which was always a bit of a fudge. It also means that there is one less
1071 argument for the match() function, which reduces its stack requirements
1072 slightly. This change also fixes an incompatibility with Perl: the pattern
1073 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1074
1075 10. More internal refactoring has drastically reduced the number of recursive
1076 calls to match() for possessively repeated groups such as (abc)++ when
1077 using pcre_exec().
1078
1079 11. While implementing 10, a number of bugs in the handling of groups were
1080 discovered and fixed:
1081
1082 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1083 (a|)*(?1) gave a compile-time internal error.
1084 ((a|)+)+ did not notice that the outer group could match an empty string.
1085 (^a|^)+ was not marked as anchored.
1086 (.*a|.*)+ was not marked as matching at start or after a newline.
1087
1088 12. Yet more internal refactoring has removed another argument from the match()
1089 function. Special calls to this function are now indicated by setting a
1090 value in a variable in the "match data" data block.
1091
1092 13. Be more explicit in pcre_study() instead of relying on "default" for
1093 opcodes that mean there is no starting character; this means that when new
1094 ones are added and accidentally left out of pcre_study(), testing should
1095 pick them up.
1096
1097 14. The -s option of pcretest has been documented for ages as being an old
1098 synonym of -m (show memory usage). I have changed it to mean "force study
1099 for every regex", that is, assume /S for every regex. This is similar to -i
1100 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1101 using it. It makes it easier to run collections of tests with and without
1102 study enabled, and thereby test pcre_study() more easily. All the standard
1103 tests are now run with and without -s (but some patterns can be marked as
1104 "never study" - see 20 below).
1105
1106 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1107 restoration of the capturing data to the outer values was not happening
1108 correctly.
1109
1110 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1111 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1112 pattern had matched an empty string, and so incorrectly returned a no
1113 match.
1114
1115 17. There was optimizing code for the last branch of non-capturing parentheses,
1116 and also for the obeyed branch of a conditional subexpression, which used
1117 tail recursion to cut down on stack usage. Unfortunately, now that there is
1118 the possibility of (*THEN) occurring in these branches, tail recursion is
1119 no longer possible because the return has to be checked for (*THEN). These
1120 two optimizations have therefore been removed. [But see 8.20/11 above.]
1121
1122 18. If a pattern containing \R was studied, it was assumed that \R always
1123 matched two bytes, thus causing the minimum subject length to be
1124 incorrectly computed because \R can also match just one byte.
1125
1126 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1127 was incorrectly computed.
1128
1129 20. If /S is present twice on a test pattern in pcretest input, it now
1130 *disables* studying, thereby overriding the use of -s on the command line
1131 (see 14 above). This is necessary for one or two tests to keep the output
1132 identical in both cases.
1133
1134 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1135 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1136
1137 22. When an atomic group that contained a capturing parenthesis was
1138 successfully matched, but the branch in which it appeared failed, the
1139 capturing was not being forgotten if a higher numbered group was later
1140 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1141 group 1 to "a", when in fact it should be unset. This applied to multi-
1142 branched capturing and non-capturing groups, repeated or not, and also to
1143 positive assertions (capturing in negative assertions does not happen
1144 in PCRE) and also to nested atomic groups.
1145
1146 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1147 subject after a captured substring, to make it easier to tell which of a
1148 number of identical substrings has been captured.
1149
1150 24. The way atomic groups are processed by pcre_exec() has been changed so that
1151 if they are repeated, backtracking one repetition now resets captured
1152 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1153 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1154 "aaa". Previously, it would have been "a". As part of this code
1155 refactoring, the way recursive calls are handled has also been changed.
1156
1157 25. If an assertion condition captured any substrings, they were not passed
1158 back unless some other capturing happened later. For example, if
1159 (?(?=(a))a) was matched against "a", no capturing was returned.
1160
1161 26. When studying a pattern that contained subroutine calls or assertions,
1162 the code for finding the minimum length of a possible match was handling
1163 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1164 group 1 called group 2 while simultaneously a separate group 2 called group
1165 1). A stack overflow occurred in this case. I have fixed this by limiting
1166 the recursion depth to 10.
1167
1168 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1169 Fortmann. This supports explicit test numbers on the command line, and has
1170 argument validation and error reporting.
1171
1172 28. An instance of \X with an unlimited repeat could fail if at any point the
1173 first character it looked at was a mark character.
1174
1175 29. Some minor code refactoring concerning Unicode properties and scripts
1176 should reduce the stack requirement of match() slightly.
1177
1178 30. Added the '=' option to pcretest to check the setting of unused capturing
1179 slots at the end of the pattern, which are documented as being -1, but are
1180 not included in the return count.
1181
1182 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1183 compiled something random. Now it gives a compile-time error (as does
1184 Perl).
1185
1186 32. A *MARK encountered during the processing of a positive assertion is now
1187 recorded and passed back (compatible with Perl).
1188
1189 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1190 had alternative anchored branches, the search for a second match in a line
1191 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1192 matched the line "0102" twice. The same bug affected patterns that started
1193 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1194 twice.
1195
1196 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1197 does, and because of capturing effects, quantifying parenthesized
1198 assertions may at times be useful. Quantifiers are now allowed for
1199 parenthesized assertions.
1200
1201 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1202
1203 36. \g was being checked for fancy things in a character class, when it should
1204 just be a literal "g".
1205
1206 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1207 appearance of a nested POSIX class supersedes an apparent external class.
1208 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1209 unescaped square brackets may also appear as part of class names. For
1210 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1211 more like Perl. (But see 8.20/1 above.)
1212
1213 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1214 was because it thought it was \N{name}, which is not supported).
1215
1216 39. Add minix to OS list not supporting the -S option in pcretest.
1217
1218 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1219 cannot analyze patterns in sufficient detail to catch mutual recursions
1220 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1221 subgroup is called recursively as a subpattern for a second time at the
1222 same position in the subject string. In previous releases this might have
1223 been caught by the recursion limit, or it might have run out of stack.
1224
1225 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1226 happen only once. PCRE was, however incorrectly giving a compile time error
1227 "recursive call could loop indefinitely" because it cannot analyze the
1228 pattern in sufficient detail. The compile time test no longer happens when
1229 PCRE is compiling a conditional subpattern, but actual runaway loops are
1230 now caught at runtime (see 40 above).
1231
1232 42. It seems that Perl allows any characters other than a closing parenthesis
1233 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1234 has been changed to be the same.
1235
1236 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1237 as not to get warnings when autogen.sh is called. Also changed
1238 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1239
1240 44. To help people who use pcregrep to scan files containing exceedingly long
1241 lines, the following changes have been made:
1242
1243 (a) The default value of the buffer size parameter has been increased from
1244 8K to 20K. (The actual buffer used is three times this size.)
1245
1246 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1247 PCRE is built.
1248
1249 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1250 to be set at run time.
1251
1252 (d) Numerical values in pcregrep options can be followed by K or M, for
1253 example --buffer-size=50K.
1254
1255 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1256 given and the return code is set to 2.
1257
1258 45. Add a pointer to the latest mark to the callout data block.
1259
1260 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1261 partial match of an empty string instead of no match. This was specific to
1262 the use of ".".
1263
1264 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1265 complete match instead of a partial match. This bug was dependent on both
1266 the PCRE_UTF8 and PCRE_DOTALL options being set.
1267
1268 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1269 starting byte set, because \b was not being ignored.
1270
1271
1272 Version 8.12 15-Jan-2011
1273 ------------------------
1274
1275 1. Fixed some typos in the markup of the man pages, and wrote a script that
1276 checks for such things as part of the documentation building process.
1277
1278 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1279 --match-limit and --recursion-limit options (added for 8.11). In
1280 particular, this made one of the standard tests fail. (The integer value
1281 went into the wrong half of a long int.)
1282
1283 3. If the --colour option was given to pcregrep with -v (invert match), it
1284 did strange things, either producing crazy output, or crashing. It should,
1285 of course, ignore a request for colour when reporting lines that do not
1286 match.
1287
1288 4. Another pcregrep bug caused similar problems if --colour was specified with
1289 -M (multiline) and the pattern match finished with a line ending.
1290
1291 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1292 matched in multiline mode, the following line was shown as part of the
1293 match. This seems wrong, so I have changed it.
1294
1295 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1296 the check for further matches in the same line (so they could be coloured)
1297 to overrun the end of the current line. If another match was found, it was
1298 incorrectly shown (and then shown again when found in the next line).
1299
1300 7. If pcregrep was compiled under Windows, there was a reference to the
1301 function pcregrep_exit() before it was defined. I am assuming this was
1302 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1303 reported by a user. I've moved the definition above the reference.
1304
1305
1306 Version 8.11 10-Dec-2010
1307 ------------------------
1308
1309 1. (*THEN) was not working properly if there were untried alternatives prior
1310 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1311 backtracked to try for "b" instead of moving to the next alternative branch
1312 at the same level (in this case, to look for "c"). The Perl documentation
1313 is clear that when (*THEN) is backtracked onto, it goes to the "next
1314 alternative in the innermost enclosing group".
1315
1316 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1317 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1318 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1319 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1320 (*THEN).
1321
1322 3. If \s appeared in a character class, it removed the VT character from
1323 the class, even if it had been included by some previous item, for example
1324 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1325 of \s, but is part of the POSIX "space" class.)
1326
1327 4. A partial match never returns an empty string (because you can always
1328 match an empty string at the end of the subject); however the checking for
1329 an empty string was starting at the "start of match" point. This has been
1330 changed to the "earliest inspected character" point, because the returned
1331 data for a partial match starts at this character. This means that, for
1332 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1333 (previously it gave "no match").
1334
1335 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1336 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1337 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1338 has an implication that the given string is incomplete (because a partial
1339 match is preferred over a full match). For this reason, these items now
1340 give a partial match in this situation. [Aside: previously, the one case
1341 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1342 match rather than a full match, which was wrong by the old rules, but is
1343 now correct.]
1344
1345 6. There was a bug in the handling of #-introduced comments, recognized when
1346 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1347 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1348 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1349 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1350 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1351 places in pcre_compile().
1352
1353 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1354 comments when looking ahead for named forward references to subpatterns,
1355 the only newline sequence it recognized was NL. It now handles newlines
1356 according to the set newline convention.
1357
1358 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1359 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1360 cater for a lack of strerror(). These oversights have been fixed.
1361
1362 9. Added --match-limit and --recursion-limit to pcregrep.
1363
1364 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1365
1366 11. When the -o option was used, pcregrep was setting a return code of 1, even
1367 when matches were found, and --line-buffered was not being honoured.
1368
1369 12. Added an optional parentheses number to the -o and --only-matching options
1370 of pcregrep.
1371
1372 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1373 can match an empty string. The code to do it in pcretest and pcredemo
1374 needed fixing:
1375
1376 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1377 only one byte after an empty string match just before CRLF (this case
1378 just got forgotten; "any" and "anycrlf" were OK).
1379
1380 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1381 mode when an empty string match preceded an ASCII character followed by
1382 a non-ASCII character. (The code for advancing by one character rather
1383 than one byte was nonsense.)
1384
1385 (c) The pcredemo.c sample program did not have any code at all to handle
1386 the cases when CRLF is a valid newline sequence.
1387
1388 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1389 as a starting offset was within the subject string. There is now a new
1390 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1391 negative or greater than the length of the string. In order to test this,
1392 pcretest is extended to allow the setting of negative starting offsets.
1393
1394 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1395 starting offset points to the beginning of a UTF-8 character was
1396 unnecessarily clumsy. I tidied it up.
1397
1398 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1399 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1400
1401 17. Nobody had reported that the --include_dir option, which was added in
1402 release 7.7 should have been called --include-dir (hyphen, not underscore)
1403 for compatibility with GNU grep. I have changed it to --include-dir, but
1404 left --include_dir as an undocumented synonym, and the same for
1405 --exclude-dir, though that is not available in GNU grep, at least as of
1406 release 2.5.4.
1407
1408 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1409 characters from a string of bytes) have been redefined so as not to use
1410 loops, in order to improve performance in some environments. At the same
1411 time, I abstracted some of the common code into auxiliary macros to save
1412 repetition (this should not affect the compiled code).
1413
1414 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1415 compile-time error is now given if \c is not followed by an ASCII
1416 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1417 different, and any byte value is allowed.)
1418
1419 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1420 START_OPTIMIZE option, which is now allowed at compile time - but just
1421 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1422 to pcregrep and other applications that have no direct access to PCRE
1423 options. The new /Y option in pcretest sets this option when calling
1424 pcre_compile().
1425
1426 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1427 back references. Groups containing recursive back references were forced to
1428 be atomic by that change, but in the case of named groups, the amount of
1429 memory required was incorrectly computed, leading to "Failed: internal
1430 error: code overflow". This has been fixed.
1431
1432 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1433 pcretest.c, to avoid build problems in some Borland environments.
1434
1435
1436 Version 8.10 25-Jun-2010
1437 ------------------------
1438
1439 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1440 THEN.
1441
1442 2. (*ACCEPT) was not working when inside an atomic group.
1443
1444 3. Inside a character class, \B is treated as a literal by default, but
1445 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1446 causes the error). The code is unchanged, but I tidied the documentation.
1447
1448 4. Inside a character class, PCRE always treated \R and \X as literals,
1449 whereas Perl faults them if its -w option is set. I have changed PCRE so
1450 that it faults them when PCRE_EXTRA is set.
1451
1452 5. Added support for \N, which always matches any character other than
1453 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1454
1455 6. When compiling pcregrep with newer versions of gcc which may have
1456 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1457 declared with attribute warn_unused_result" were given. Just casting the
1458 result to (void) does not stop the warnings; a more elaborate fudge is
1459 needed. I've used a macro to implement this.
1460
1461 7. Minor change to pcretest.c to avoid a compiler warning.
1462
1463 8. Added four artifical Unicode properties to help with an option to make
1464 \s etc use properties (see next item). The new properties are: Xan
1465 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1466
1467 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1468 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1469 this option. Modified pcretest to add /W to test this facility. Added
1470 REG_UCP to make it available via the POSIX interface.
1471
1472 10. Added --line-buffered to pcregrep.
1473
1474 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1475 studied, and the match started with a letter with a code point greater than
1476 127 whose first byte was different to the first byte of the other case of
1477 the letter, the other case of this starting letter was not recognized
1478 (#976).
1479
1480 12. If a pattern that was studied started with a repeated Unicode property
1481 test, for example, \p{Nd}+, there was the theoretical possibility of
1482 setting up an incorrect bitmap of starting bytes, but fortunately it could
1483 not have actually happened in practice until change 8 above was made (it
1484 added property types that matched character-matching opcodes).
1485
1486 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1487 possible starting bytes for non-anchored patterns.
1488
1489 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1490 \R, and also a number of cases that involve Unicode properties, both
1491 explicit and implicit when PCRE_UCP is set.
1492
1493 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1494 input, it could crash or give wrong results if characters with values
1495 greater than 0xc0 were present in the subject string. (Detail: it assumed
1496 UTF-8 input when processing these items.)
1497
1498 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1499 size_t is 64-bit (#991).
1500
1501 17. Added a check for running out of memory when PCRE is compiled with
1502 --disable-stack-for-recursion (#990).
1503
1504 18. If the last data line in a file for pcretest does not have a newline on
1505 the end, a newline was missing in the output.
1506
1507 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1508 less than 128) in its various bitmaps. However, there is a facility for
1509 generating tables according to the current locale when PCRE is compiled. It
1510 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1511 characters, are recognized by isspace() and therefore were getting set in
1512 these tables, and indeed these tables seem to approximate to ISO 8859. This
1513 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1514 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1515 which of course cannot start UTF-8 characters. I have changed the code so
1516 that only real ASCII characters (less than 128) and the correct starting
1517 bytes for UTF-8 encodings are set for characters greater than 127 when in
1518 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1519 altogether.)
1520
1521 20. Added the /T option to pcretest so as to be able to run tests with non-
1522 standard character tables, thus making it possible to include the tests
1523 used for 19 above in the standard set of tests.
1524
1525 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1526 reference to a subpattern the other side of a comment that contains an
1527 opening parenthesis caused either an internal compiling error, or a
1528 reference to the wrong subpattern.
1529
1530
1531 Version 8.02 19-Mar-2010
1532 ------------------------
1533
1534 1. The Unicode data tables have been updated to Unicode 5.2.0.
1535
1536 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1537 configured.
1538
1539 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1540 original author of that file, following a query about its status.
1541
1542 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1543 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1544
1545 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1546 quantifier applied to a forward-referencing subroutine call, could compile
1547 incorrect code or give the error "internal error: previously-checked
1548 referenced subpattern not found".
1549
1550 6. Both MS Visual Studio and Symbian OS have problems with initializing
1551 variables to point to external functions. For these systems, therefore,
1552 pcre_malloc etc. are now initialized to local functions that call the
1553 relevant global functions.
1554
1555 7. There were two entries missing in the vectors called coptable and poptable
1556 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1557 I've fixed the data, and added a kludgy way of testing at compile time that
1558 the lengths are correct (equal to the number of opcodes).
1559
1560 8. Following on from 7, I added a similar kludge to check the length of the
1561 eint vector in pcreposix.c.
1562
1563 9. Error texts for pcre_compile() are held as one long string to avoid too
1564 much relocation at load time. To find a text, the string is searched,
1565 counting zeros. There was no check for running off the end of the string,
1566 which could happen if a new error number was added without updating the
1567 string.
1568
1569 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1570
1571 11. \K was not working if it appeared in an atomic group or in a group that
1572 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1573 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1574 the assertion is positive, but not if it is negative.
1575
1576 12. Change 11 fortuitously reduced the size of the stack frame used in the
1577 "match()" function of pcre_exec.c by one pointer. Forthcoming
1578 implementation of support for (*MARK) will need an extra pointer on the
1579 stack; I have reserved it now, so that the stack frame size does not
1580 decrease.
1581
1582 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1583 item in branch that calls a recursion is a subroutine call - as in the
1584 second branch in the above example - was incorrectly given the compile-
1585 time error "recursive call could loop indefinitely" because pcre_compile()
1586 was not correctly checking the subroutine for matching a non-empty string.
1587
1588 14. The checks for overrunning compiling workspace could trigger after an
1589 overrun had occurred. This is a "should never occur" error, but it can be
1590 triggered by pathological patterns such as hundreds of nested parentheses.
1591 The checks now trigger 100 bytes before the end of the workspace.
1592
1593 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1594
1595
1596 Version 8.01 19-Jan-2010
1597 ------------------------
1598
1599 1. If a pattern contained a conditional subpattern with only one branch (in
1600 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1601 computed the wrong minimum data length (which is of course zero for such
1602 subpatterns). This could cause incorrect "no match" results.
1603
1604 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1605 the pattern is reset in the first branch, pcre_compile() failed with
1606 "internal error: code overflow at offset...". This happened only when
1607 the reset was to the original external option setting. (An optimization
1608 abstracts leading options settings into an external setting, which was the
1609 cause of this.)
1610
1611 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1612 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1613 assertion pattern did not match (meaning that the assertion was true), it
1614 was incorrectly treated as false if the SKIP had been reached during the
1615 matching. This also applied to assertions used as conditions.
1616
1617 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1618 assertion subpattern, including such a pattern used as a condition,
1619 unpredictable results occurred, instead of the error return
1620 PCRE_ERROR_DFA_UITEM.
1621
1622 5. The C++ GlobalReplace function was not working like Perl for the special
1623 situation when an empty string is matched. It now does the fancy magic
1624 stuff that is necessary.
1625
1626 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1627 removed. (These were left over from very, very early versions of PCRE.)
1628
1629 7. Some cosmetic changes to the code to make life easier when compiling it
1630 as part of something else:
1631
1632 (a) Change DEBUG to PCRE_DEBUG.
1633
1634 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1635 called "current" as "current_branch", to prevent a collision with the
1636 Linux macro when compiled as a kernel module.
1637
1638 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1639 prevent a collision with the Linux macro when compiled as a kernel
1640 module.
1641
1642 8. In pcre_compile() there are some checks for integer overflows that used to
1643 cast potentially large values to (double). This has been changed to that
1644 when building, a check for int64_t is made, and if it is found, it is used
1645 instead, thus avoiding the use of floating point arithmetic. (There is no
1646 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1647 double.
1648
1649 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1650 2005 (difference between two addresses compared to an unsigned value).
1651
1652 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1653 custom one, because of the following reported problem in Windows:
1654
1655 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1656 under Win32.
1657 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1658 therefore missing the function definition.
1659 - The compiler thus generates a "C" signature for the test function.
1660 - The linker fails to find the "C" function.
1661 - PCRE fails to configure if asked to do so against libbz2.
1662
1663 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1664 messages were output:
1665
1666 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1667 rerunning libtoolize, to keep the correct libtool macros in-tree.
1668 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1669
1670 I have done both of these things.
1671
1672 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1673 most of the time, it *can* run out if it is given a pattern that contains a
1674 runaway infinite recursion. I updated the discussion in the pcrestack man
1675 page.
1676
1677 13. Now that we have gone to the x.xx style of version numbers, the minor
1678 version may start with zero. Using 08 or 09 is a bad idea because users
1679 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1680 interpreted as invalid octal numbers. I've updated the previous comment in
1681 configure.ac, and also added a check that gives an error if 08 or 09 are
1682 used.
1683
1684 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1685 causing partial matching to fail when the end of the subject matched \W
1686 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1687
1688 15. There were some discrepancies between the declarations in pcre_internal.h
1689 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1690 their definitions. The declarations used "const uschar *" and the
1691 definitions used USPTR. Even though USPTR is normally defined as "const
1692 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1693 reported that: "This difference in casting confuses some C++ compilers, for
1694 example, SunCC recognizes above declarations as different functions and
1695 generates broken code for hbpcre." I have changed the declarations to use
1696 USPTR.
1697
1698 16. GNU libtool is named differently on some systems. The autogen.sh script now
1699 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1700 (FreeBSD).
1701
1702 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1703 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1704 comment: "Figure out how to create a longlong from a string: strtoll and
1705 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1706 instance, but it only takes 2 args instead of 3!"
1707
1708 18. A subtle bug concerned with back references has been fixed by a change of
1709 specification, with a corresponding code fix. A pattern such as
1710 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1711 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1712 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1713 same bug. Such groups have to be quantified to be useful, or contained
1714 inside another quantified group. (If there's no repetition, the reference
1715 can never match.) The problem arises because, having left the group and
1716 moved on to the rest of the pattern, a later failure that backtracks into
1717 the group uses the captured value from the final iteration of the group
1718 rather than the correct earlier one. I have fixed this in PCRE by forcing
1719 any group that contains a reference to itself to be an atomic group; that
1720 is, there cannot be any backtracking into it once it has completed. This is
1721 similar to recursive and subroutine calls.
1722
1723
1724 Version 8.00 19-Oct-09
1725 ----------------------
1726
1727 1. The table for translating pcre_compile() error codes into POSIX error codes
1728 was out-of-date, and there was no check on the pcre_compile() error code
1729 being within the table. This could lead to an OK return being given in
1730 error.
1731
1732 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1733 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1734 in a Windows environment.
1735
1736 3. The pcregrep --count option prints the count for each file even when it is
1737 zero, as does GNU grep. However, pcregrep was also printing all files when
1738 --files-with-matches was added. Now, when both options are given, it prints
1739 counts only for those files that have at least one match. (GNU grep just
1740 prints the file name in this circumstance, but including the count seems
1741 more useful - otherwise, why use --count?) Also ensured that the
1742 combination -clh just lists non-zero counts, with no names.
1743
1744 4. The long form of the pcregrep -F option was incorrectly implemented as
1745 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1746 but it seems right to fix it, and I didn't think it was worth preserving
1747 the old behaviour.
1748
1749 5. The command line items --regex=pattern and --regexp=pattern were not
1750 recognized by pcregrep, which required --regex pattern or --regexp pattern
1751 (with a space rather than an '='). The man page documented the '=' forms,
1752 which are compatible with GNU grep; these now work.
1753
1754 6. No libpcreposix.pc file was created for pkg-config; there was just
1755 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1756
1757 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1758 when UCP support is not needed, by modifying the Python script that
1759 generates it from Unicode data files. This should not matter if the module
1760 is correctly used as a library, but I received one complaint about 50K of
1761 unwanted data. My guess is that the person linked everything into his
1762 program rather than using a library. Anyway, it does no harm.
1763
1764 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1765 was a minimum greater than 1 for a wide character in a possessive
1766 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1767 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1768 character. Chaos in the form of incorrect output or a compiling loop could
1769 result.
1770
1771 9. The restrictions on what a pattern can contain when partial matching is
1772 requested for pcre_exec() have been removed. All patterns can now be
1773 partially matched by this function. In addition, if there are at least two
1774 slots in the offset vector, the offset of the earliest inspected character
1775 for the match and the offset of the end of the subject are set in them when
1776 PCRE_ERROR_PARTIAL is returned.
1777
1778 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1779 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1780 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1781 and may be more useful for multi-segment matching.
1782
1783 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1784 used to be given if ever the end of the subject was reached; now it is
1785 given only if matching could not proceed because another character was
1786 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1787 string "Z", which now yields "no match" instead of "partial match". In the
1788 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1789 final character ended with (*FAIL).
1790
1791 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1792 if the pattern had a "must contain" character that was already found in the
1793 earlier partial match, unless partial matching was again requested. For
1794 example, with the pattern /dog.(body)?/, the "must contain" character is
1795 "g". If the first part-match was for the string "dog", restarting with
1796 "sbody" failed. This bug has been fixed.
1797
1798 13. The string returned by pcre_dfa_exec() after a partial match has been
1799 changed so that it starts at the first inspected character rather than the
1800 first character of the match. This makes a difference only if the pattern
1801 starts with a lookbehind assertion or \b or \B (\K is not supported by
1802 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1803 matching functions compatible, and I think it's the right thing to do.
1804
1805 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1806 so that the demonstration program is easily available in environments where
1807 PCRE has not been installed from source.
1808
1809 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1810 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1811 library.
1812
1813 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1814 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1815 is not the first non-POSIX option to be added. Clearly some people find
1816 these options useful.
1817
1818 17. If a caller to the POSIX matching function regexec() passes a non-zero
1819 value for nmatch with a NULL value for pmatch, the value of
1820 nmatch is forced to zero.
1821
1822 18. RunGrepTest did not have a test for the availability of the -u option of
1823 the diff command, as RunTest does. It now checks in the same way as
1824 RunTest, and also checks for the -b option.
1825
1826 19. If an odd number of negated classes containing just a single character
1827 interposed, within parentheses, between a forward reference to a named
1828 subpattern and the definition of the subpattern, compilation crashed with
1829 an internal error, complaining that it could not find the referenced
1830 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1831 [The bug was that it was starting one character too far in when skipping
1832 over the character class, thus treating the ] as data rather than
1833 terminating the class. This meant it could skip too much.]
1834
1835 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1836 /g option in pcretest when the pattern contains \K, which makes it possible
1837 to have an empty string match not at the start, even when the pattern is
1838 anchored. Updated pcretest and pcredemo to use this option.
1839
1840 21. If the maximum number of capturing subpatterns in a recursion was greater
1841 than the maximum at the outer level, the higher number was returned, but
1842 with unset values at the outer level. The correct (outer level) value is
1843 now given.
1844
1845 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1846 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1847 make it do so. The string so far is captured, making this feature
1848 compatible with Perl.
1849
1850 23. The tests have been re-organized, adding tests 11 and 12, to make it
1851 possible to check the Perl 5.10 features against Perl 5.10.
1852
1853 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1854 pattern matches a fixed length string. PCRE did not allow this; now it
1855 does. Neither allows recursion.
1856
1857 25. I finally figured out how to implement a request to provide the minimum
1858 length of subject string that was needed in order to match a given pattern.
1859 (It was back references and recursion that I had previously got hung up
1860 on.) This code has now been added to pcre_study(); it finds a lower bound
1861 to the length of subject needed. It is not necessarily the greatest lower
1862 bound, but using it to avoid searching strings that are too short does give
1863 some useful speed-ups. The value is available to calling programs via
1864 pcre_fullinfo().
1865
1866 26. While implementing 25, I discovered to my embarrassment that pcretest had
1867 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1868 study optimizations had never been tested with that matching function.
1869 Oops. What is worse, even when it was passed study data, there was a bug in
1870 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1871 were also very few tests of studied patterns with pcre_dfa_exec().
1872
1873 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1874 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1875 on the other side of the coin, they are no longer allowed to have different
1876 names, because these cannot be distinguished in PCRE, and this has caused
1877 confusion. (This is a difference from Perl.)
1878
1879 28. When duplicate subpattern names are present (necessarily with different
1880 numbers, as required by 27 above), and a test is made by name in a
1881 conditional pattern, either for a subpattern having been matched, or for
1882 recursion in such a pattern, all the associated numbered subpatterns are
1883 tested, and the overall condition is true if the condition is true for any
1884 one of them. This is the way Perl works, and is also more like the way
1885 testing by number works.
1886
1887
1888 Version 7.9 11-Apr-09
1889 ---------------------
1890
1891 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1892 (pcretest), all targets were linked against these libraries. This included
1893 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1894 libraries. This caused unwanted dependencies to be created. This problem
1895 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1896 pcretest is linked with readline.
1897
1898 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1899 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1900 moved outside it again, because FALSE and TRUE are already defined in AIX,
1901 but BOOL is not.
1902
1903 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1904 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1905
1906 4. The pcregrep documentation said spaces were inserted as well as colons (or
1907 hyphens) following file names and line numbers when outputting matching
1908 lines. This is not true; no spaces are inserted. I have also clarified the
1909 wording for the --colour (or --color) option.
1910
1911 5. In pcregrep, when --colour was used with -o, the list of matching strings
1912 was not coloured; this is different to GNU grep, so I have changed it to be
1913 the same.
1914
1915 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1916 each matching line was coloured. Now it goes on to look for further matches
1917 of any of the test patterns, which is the same behaviour as GNU grep.
1918
1919 7. A pattern that could match an empty string could cause pcregrep to loop; it
1920 doesn't make sense to accept an empty string match in pcregrep, so I have
1921 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1922 seems to be how GNU grep behaves. [But see later change 40 for release
1923 8.33.]
1924
1925 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1926 start or after a newline", because the conditional assertion was not being
1927 correctly handled. The rule now is that both the assertion and what follows
1928 in the first alternative must satisfy the test.
1929
1930 9. If auto-callout was enabled in a pattern with a conditional group whose
1931 condition was an assertion, PCRE could crash during matching, both with
1932 pcre_exec() and pcre_dfa_exec().
1933
1934 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1935 used for matching.
1936
1937 11. Unicode property support in character classes was not working for
1938 characters (bytes) greater than 127 when not in UTF-8 mode.
1939
1940 12. Added the -M command line option to pcretest.
1941
1942 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1943
1944 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1945
1946 16. Added comments and documentation about mis-use of no_arg in the C++
1947 wrapper.
1948
1949 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1950 from Martin Jerabek that uses macro names for all relevant character and
1951 string constants.
1952
1953 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1954 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1955 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1956 these, but not everybody uses configure.
1957
1958 19. A conditional group that had only one branch was not being correctly
1959 recognized as an item that could match an empty string. This meant that an
1960 enclosing group might also not be so recognized, causing infinite looping
1961 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1962 with the subject "ab", where knowledge that the repeated group can match
1963 nothing is needed in order to break the loop.
1964
1965 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1966 exec(), but without supplying a callout function, matching went wrong.
1967
1968 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1969 leak if the size of the offset vector was greater than 30. When the vector
1970 is smaller, the saved offsets during recursion go onto a local stack
1971 vector, but for larger vectors malloc() is used. It was failing to free
1972 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1973 error, in fact).
1974
1975 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1976 heapframe that is used only when UTF-8 support is enabled. This caused no
1977 problem, but was untidy.
1978
1979 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1980 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1981 included within another project.
1982
1983 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1984 slightly modified by me:
1985
1986 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1987 not building pcregrep.
1988
1989 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1990 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1991
1992 25. Forward references, both numeric and by name, in patterns that made use of
1993 duplicate group numbers, could behave incorrectly or give incorrect errors,
1994 because when scanning forward to find the reference group, PCRE was not
1995 taking into account the duplicate group numbers. A pattern such as
1996 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1997
1998 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1999 the feature of a custom pointer more persuasive (as requested by a user).
2000
2001 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2002 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2003 pre-defined.
2004
2005 28. Added support for (*UTF8) at the start of a pattern.
2006
2007 29. Arrange for flags added by the "release type" setting in CMake to be shown
2008 in the configuration summary.
2009
2010
2011 Version 7.8 05-Sep-08
2012 ---------------------
2013
2014 1. Replaced UCP searching code with optimized version as implemented for Ad
2015 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2016 stage table and inline lookup instead of a function, giving speed ups of 2
2017 to 5 times on some simple patterns that I tested. Permission was given to
2018 distribute the MultiStage2.py script that generates the tables (it's not in
2019 the tarball, but is in the Subversion repository).
2020
2021 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2022 scripts.
2023
2024 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2025 a group with a zero qualifier. The result of the study could be incorrect,
2026 or the function might crash, depending on the pattern.
2027
2028 4. Caseless matching was not working for non-ASCII characters in back
2029 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2030 It now works when Unicode Property Support is available.
2031
2032 5. In pcretest, an escape such as \x{de} in the data was always generating
2033 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2034 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2035 truncation.
2036
2037 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2038
2039 7. Added two (int) casts to pcregrep when printing the difference of two
2040 pointers, in case they are 64-bit values.
2041
2042 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2043 test 2 if it fails.
2044
2045 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2046 and a #define of that name to empty if it is not externally set. This is to
2047 allow users of MSVC to set it if necessary.
2048
2049 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2050 the convenience functions in the pcre_get.c source file.
2051
2052 11. An option change at the start of a pattern that had top-level alternatives
2053 could cause overwriting and/or a crash. This command provoked a crash in
2054 some environments:
2055
2056 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2057
2058 This potential security problem was recorded as CVE-2008-2371.
2059
2060 12. For a pattern where the match had to start at the beginning or immediately
2061 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2062 pcre_dfa_exec() could read past the end of the passed subject if there was
2063 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2064 pcretest so that it places the subject at the end of its malloc-ed buffer.
2065
2066 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2067 exec() might read past the end of the data buffer in UTF-8 mode.
2068
2069 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2070 the data contained the byte 0x85 as part of a UTF-8 character within its
2071 first line. This applied both to normal and DFA matching.
2072
2073 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2074 /^[^d]*?$/8 failed to match "abc".
2075
2076 16. Added a missing copyright notice to pcrecpp_internal.h.
2077
2078 17. Make it more clear in the documentation that values returned from
2079 pcre_exec() in ovector are byte offsets, not character counts.
2080
2081 18. Tidied a few places to stop certain compilers from issuing warnings.
2082
2083 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2084 supplied by Stefan Weber. I made a further small update for 7.8 because
2085 there is a change of source arrangements: the pcre_searchfuncs.c module is
2086 replaced by pcre_ucd.c.
2087
2088
2089 Version 7.7 07-May-08
2090 ---------------------
2091
2092 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2093 a string to a long long, pretend we don't even have a long long." This is
2094 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2095
2096 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2097 pre-7.6 versions, which defined a global no_arg variable instead of putting
2098 it in the RE class. (See also #8 below.)
2099
2100 3. Remove a line of dead code, identified by coverity and reported by Nuno
2101 Lopes.
2102
2103 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2104
2105 (1) The include/exclude patterns were being applied to the whole pathnames
2106 of files, instead of just to the final components.
2107
2108 (2) If there was more than one level of directory, the subdirectories were
2109 skipped unless they satisfied the include/exclude conditions. This is
2110 inconsistent with GNU grep (and could even be seen as contrary to the
2111 pcregrep specification - which I improved to make it absolutely clear).
2112 The action now is always to scan all levels of directory, and just
2113 apply the include/exclude patterns to regular files.
2114
2115 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2116 --exclude_dir in the tests to avoid scanning .svn directories.
2117
2118 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2119 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2120 doesn't support NULs in patterns.
2121
2122 7. Added some missing "const"s to declarations of static tables in
2123 pcre_compile.c and pcre_dfa_exec.c.
2124
2125 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2126 caused by fix #2 above. (Subsequently also a second patch to fix the
2127 first patch. And a third patch - this was a messy problem.)
2128
2129 9. Applied Craig's patch to remove the use of push_back().
2130
2131 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2132 matching function regexec().
2133
2134 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2135 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2136 references. PCRE supports relative numbers with this syntax (I don't think
2137 Oniguruma does).
2138
2139 12. Previously, a group with a zero repeat such as (...){0} was completely
2140 omitted from the compiled regex. However, this means that if the group
2141 was called as a subroutine from elsewhere in the pattern, things went wrong
2142 (an internal error was given). Such groups are now left in the compiled
2143 pattern, with a new opcode that causes them to be skipped at execution
2144 time.
2145
2146 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2147 to the way PCRE behaves:
2148
2149 (a) A lone ] character is dis-allowed (Perl treats it as data).
2150
2151 (b) A back reference to an unmatched subpattern matches an empty string
2152 (Perl fails the current match path).
2153
2154 (c) A data ] in a character class must be notated as \] because if the
2155 first data character in a class is ], it defines an empty class. (In
2156 Perl it is not possible to have an empty class.) The empty class []
2157 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2158 The negative empty class [^] matches any one character, independently
2159 of the DOTALL setting.
2160
2161 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2162 non-existent subpattern following a character class starting with ']' and
2163 containing () gave an internal compiling error instead of "reference to
2164 non-existent subpattern". Fortunately, when the pattern did exist, the
2165 compiled code was correct. (When scanning forwards to check for the
2166 existence of the subpattern, it was treating the data ']' as terminating
2167 the class, so got the count wrong. When actually compiling, the reference
2168 was subsequently set up correctly.)
2169
2170 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2171 it was being rejected as not supported by pcre_dfa_exec(), even though
2172 other assertions are supported. I have made pcre_dfa_exec() support
2173 (*FAIL).
2174
2175 16. The implementation of 13c above involved the invention of a new opcode,
2176 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2177 cannot be changed at match time, I realized I could make a small
2178 improvement to matching performance by compiling OP_ALLANY instead of
2179 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2180 on the OP_ANY path.
2181
2182 17. Compiling pcretest on Windows with readline support failed without the
2183 following two fixes: (1) Make the unistd.h include conditional on
2184 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2185
2186 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2187 ncurses library to be included for pcretest when ReadLine support is
2188 requested, but also to allow for it to be overridden. This patch came from
2189 Daniel Bergström.
2190
2191 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2192 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2193 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2194 spotting this.
2195
2196
2197 Version 7.6 28-Jan-08
2198 ---------------------
2199
2200 1. A character class containing a very large number of characters with
2201 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2202 overflow.
2203
2204 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2205 HAVE_LONG_LONG is not defined.
2206
2207 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2208 bring them up to date and include new features. This patch includes:
2209
2210 - Fixed PH's badly added libz and libbz2 support.
2211 - Fixed a problem with static linking.
2212 - Added pcredemo. [But later removed - see 7 below.]
2213 - Fixed dftables problem and added an option.
2214 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2215 HAVE_LONG_LONG.
2216 - Added readline support for pcretest.
2217 - Added an listing of the option settings after cmake has run.
2218
2219 4. A user submitted a patch to Makefile that makes it easy to create
2220 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2221 Makefile.am that cause it to include this special target, without
2222 affecting anything else. Note that the same mingw target plus all
2223 the other distribution libraries and programs are now supported
2224 when configuring with CMake (see 6 below) instead of with
2225 Configure/Make.
2226
2227 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2228 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2229 exported in the Windows port". It has not yet been confirmed that the patch
2230 solves the problem, but it does no harm.
2231
2232 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2233 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2234 with CMake, and also correct the comment about stack recursion.
2235
2236 7. Remove the automatic building of pcredemo from the ./configure system and
2237 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2238 of a program that users should build themselves after PCRE is installed, so
2239 building it automatically is not really right. What is more, it gave
2240 trouble in some build environments.
2241
2242 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2243
2244
2245 Version 7.5 10-Jan-08
2246 ---------------------
2247
2248 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2249 values in parens when parsing an RE using the C++ wrapper."
2250
2251 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2252 Characters greater than 255 were excluded from the class instead of being
2253 included.
2254
2255 3. The same bug as (2) above applied to negated POSIX classes such as
2256 [:^space:].
2257
2258 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2259 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2260 I have changed it.
2261
2262 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2263 first named subpattern) and a construct such as (?&a) would reference the
2264 first named subpattern whose name started with "a" (in other words, the
2265 length check was missing). Both these problems are fixed. "Subpattern name
2266 expected" is now given for (?&) (a zero-length name), and this patch also
2267 makes it give the same error for \k'' (previously it complained that that
2268 was a reference to a non-existent subpattern).
2269
2270 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2271 this is right because (?- can be followed by option settings as well as by
2272 digits. I have, however, made the messages clearer.
2273
2274 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2275 than the number used in the conditional) now cause a compile-time error.
2276 This is actually not compatible with Perl, which accepts such patterns, but
2277 treats the conditional as always being FALSE (as PCRE used to), but it
2278 seems to me that giving a diagnostic is better.
2279
2280 8. Change "alphameric" to the more common word "alphanumeric" in comments
2281 and messages.
2282
2283 9. Fix two occurrences of "backslash" in comments that should have been
2284 "backspace".
2285
2286 10. Remove two redundant lines of code that can never be obeyed (their function
2287 was moved elsewhere).
2288
2289 11. The program that makes PCRE's Unicode character property table had a bug
2290 which caused it to generate incorrect table entries for sequences of
2291 characters that have the same character type, but are in different scripts.
2292 It amalgamated them into a single range, with the script of the first of
2293 them. In other words, some characters were in the wrong script. There were
2294 thirteen such cases, affecting characters in the following ranges:
2295
2296 U+002b0 - U+002c1
2297 U+0060c - U+0060d
2298 U+0061e - U+00612
2299 U+0064b - U+0065e
2300 U+0074d - U+0076d
2301 U+01800 - U+01805
2302 U+01d00 - U+01d77
2303 U+01d9b - U+01dbf
2304 U+0200b - U+0200f
2305 U+030fc - U+030fe
2306 U+03260 - U+0327f
2307 U+0fb46 - U+0fbb1
2308 U+10450 - U+1049d
2309
2310 12. The -o option (show only the matching part of a line) for pcregrep was not
2311 compatible with GNU grep in that, if there was more than one match in a
2312 line, it showed only the first of them. It now behaves in the same way as
2313 GNU grep.
2314
2315 13. If the -o and -v options were combined for pcregrep, it printed a blank
2316 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2317 does the same. The return code can be used to tell if there were any
2318 non-matching lines.
2319
2320 14. Added --file-offsets and --line-offsets to pcregrep.
2321
2322 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2323 infinitely looping recursion. The bug was that positive lookaheads were not
2324 being skipped when checking for a possible empty match (negative lookaheads
2325 and both kinds of lookbehind were skipped).
2326
2327 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2328 inclusion of <windows.h> to before rather than after the definition of
2329 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2330
2331 17. Specifying a possessive quantifier with a specific limit for a Unicode
2332 character property caused pcre_compile() to compile bad code, which led at
2333 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2334 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2335 caused the error; without that there was no problem.
2336
2337 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2338
2339 19. Added --enable-pcretest-libreadline.
2340
2341 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2342 RE::GlobalReplace(). As a result, the number of replacements returned was
2343 double what it should be. I removed one of the increments, but Craig sent a
2344 later patch that removed the other one (the right fix) and added unit tests
2345 that check the return values (which was not done before).
2346
2347 21. Several CMake things:
2348
2349 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2350 the names libpcre and libpcreposix, not just pcre and pcreposix.
2351
2352 (2) The above change means that pcretest and pcregrep are now correctly
2353 linked with the newly-built libraries, not previously installed ones.
2354
2355 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2356
2357 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2358 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2359 UTF-8 newline character). The key issue is that the pattern starts .*;
2360 this means that the match must be either at the beginning, or after a
2361 newline. The bug was in the code for advancing after a failed match and
2362 checking that the new position followed a newline. It was not taking
2363 account of UTF-8 characters correctly.
2364
2365 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2366 character classes. PCRE was not treating the sequence [:...:] as a
2367 character class unless the ... were all letters. Perl, however, seems to
2368 allow any characters between [: and :], though of course it rejects as
2369 unknown any "names" that contain non-letters, because all the known class
2370 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2371 for example, whereas PCRE did not - it did not recognize a POSIX character
2372 class. This seemed a bit dangerous, so the code has been changed to be
2373 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2374 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2375 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2376 Perl does, and where it didn't before.
2377
2378 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2379 Windows environments %n is disabled by default.
2380
2381
2382 Version 7.4 21-Sep-07
2383 ---------------------
2384
2385 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2386 means that a class such as [\s] counted as "explicit reference to CR or
2387 LF". That isn't really right - the whole point of the change was to try to
2388 help when there was an actual mention of one of the two characters. So now
2389 the change happens only if \r or \n (or a literal CR or LF) character is
2390 encountered.
2391
2392 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2393 of both had grown to the point where there were only 3 bits left.
2394 Fortunately, there was spare space in the data structure, and so I have
2395 moved the internal flags into a new 16-bit field to free up more option
2396 bits.
2397
2398 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2399 but did not set the internal JCHANGED flag - either of these is enough to
2400 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2401 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2402 start sets both bits.
2403
2404 4. Added options (at build time, compile time, exec time) to change \R from
2405 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2406
2407 5. doc/pcresyntax.html was missing from the distribution.
2408
2409 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2410 compatibility, even though it is no longer used.
2411
2412 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2413 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2414 windows.h file is present (where different names are used). [This was
2415 reversed later after testing - see 16 below.]
2416
2417 8. Changed all #include <config.h> to #include "config.h". There were also
2418 some further <pcre.h> cases that I changed to "pcre.h".
2419
2420 9. When pcregrep was used with the --colour option, it missed the line ending
2421 sequence off the lines that it output.
2422
2423 10. It was pointed out to me that arrays of string pointers cause lots of
2424 relocations when a shared library is dynamically loaded. A technique of
2425 using a single long string with a table of offsets can drastically reduce
2426 these. I have refactored PCRE in four places to do this. The result is
2427 dramatic:
2428
2429 Originally: 290
2430 After changing UCP table: 187
2431 After changing error message table: 43
2432 After changing table of "verbs" 36
2433 After changing table of Posix names 22
2434
2435 Thanks to the folks working on Gregex for glib for this insight.
2436
2437 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2438 unicode-properties was also set.
2439
2440 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2441
2442 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2443 checked only for CRLF.
2444
2445 14. Added casts to pcretest.c to avoid compiler warnings.
2446
2447 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2448
2449 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2450 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2451 entirely. This removes changes made in 7 above.
2452
2453 17. The CMake files have been updated, and there is now more information about
2454 building with CMake in the NON-UNIX-USE document.
2455
2456
2457 Version 7.3 28-Aug-07
2458 ---------------------
2459
2460 1. In the rejigging of the build system that eventually resulted in 7.1, the
2461 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2462 brackets there is not right, since it causes compilers to look for an
2463 installed pcre.h, not the version that is in the source that is being
2464 compiled (which of course may be different). I have changed it back to:
2465
2466 #include "pcre.h"
2467
2468 I have a vague recollection that the change was concerned with compiling in
2469 different directories, but in the new build system, that is taken care of
2470 by the VPATH setting the Makefile.
2471
2472 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2473 when the subject happened to end in the byte 0x85 (e.g. if the last
2474 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2475 characters but of course it shouldn't be taken as a newline when it is part
2476 of another character. The bug was that, for an unlimited repeat of . in
2477 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2478 characters when looking for a newline.
2479
2480 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2481
2482 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2483 in debug output.
2484
2485 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2486 long printing in the pcrecpp unittest when running under MinGW.
2487
2488 6. ESC_K was left out of the EBCDIC table.
2489
2490 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2491 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2492 limit also applies to "virtual nesting" when a pattern is recursive, and in
2493 this case 1000 isn't so big. I have been able to remove this limit at the
2494 expense of backing off one optimization in certain circumstances. Normally,
2495 when pcre_exec() would call its internal match() function recursively and
2496 immediately return the result unconditionally, it uses a "tail recursion"
2497 feature to save stack. However, when a subpattern that can match an empty
2498 string has an unlimited repetition quantifier, it no longer makes this
2499 optimization. That gives it a stack frame in which to save the data for
2500 checking that an empty string has been matched. Previously this was taken
2501 from the 1000-entry workspace that had been reserved. So now there is no
2502 explicit limit, but more stack is used.
2503
2504 8. Applied Daniel's patches to solve problems with the import/export magic
2505 syntax that is required for Windows, and which was going wrong for the
2506 pcreposix and pcrecpp parts of the library. These were overlooked when this
2507 problem was solved for the main library.
2508
2509 9. There were some crude static tests to avoid integer overflow when computing
2510 the size of patterns that contain repeated groups with explicit upper
2511 limits. As the maximum quantifier is 65535, the maximum group length was
2512 set at 30,000 so that the product of these two numbers did not overflow a
2513 32-bit integer. However, it turns out that people want to use groups that
2514 are longer than 30,000 bytes (though not repeat them that many times).
2515 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2516 made it possible to implement the integer overflow checks in a much more
2517 dynamic way, which I have now done. The artificial limitation on group
2518 length has been removed - we now have only the limit on the total length of
2519 the compiled pattern, which depends on the LINK_SIZE setting.
2520
2521 10. Fixed a bug in the documentation for get/copy named substring when
2522 duplicate names are permitted. If none of the named substrings are set, the
2523 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2524 empty string.
2525
2526 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2527 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2528 because the ] is interpreted as the first data character and the
2529 terminating ] is not found. PCRE has been made compatible with Perl in this
2530 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2531 cause memory overwriting.
2532
2533 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2534 string has been matched (to stop an infinite loop). It was not recognizing
2535 a conditional subpattern that could match an empty string if that
2536 subpattern was within another subpattern. For example, it looped when
2537 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2538 condition was not nested. This bug has been fixed.
2539
2540 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2541 past the start of the subject in the presence of bytes with the top bit
2542 set, for example "\x8aBCD".
2543
2544 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2545 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2546
2547 14. Optimized (?!) to (*FAIL).
2548
2549 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2550 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2551 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2552 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2553 does: it's just the validity check that is more restrictive.
2554
2555 16. Inserted checks for integer overflows during escape sequence (backslash)
2556 processing, and also fixed erroneous offset values for syntax errors during
2557 backslash processing.
2558
2559 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2560 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2561
2562 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2563 caused an overrun.
2564
2565 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2566 something other than just ASCII characters) inside a group that had an
2567 unlimited repeat caused a loop at compile time (while checking to see
2568 whether the group could match an empty string).
2569
2570 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2571 [\P{Any}] did so. (Error in the code for printing property names.)
2572
2573 21. An orphan \E inside a character class could cause a crash.
2574
2575 22. A repeated capturing bracket such as (A)? could cause a wild memory
2576 reference during compilation.
2577
2578 23. There are several functions in pcre_compile() that scan along a compiled
2579 expression for various reasons (e.g. to see if it's fixed length for look
2580 behind). There were bugs in these functions when a repeated \p or \P was
2581 present in the pattern. These operators have additional parameters compared
2582 with \d, etc, and these were not being taken into account when moving along
2583 the compiled data. Specifically:
2584
2585 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2586 length.
2587
2588 (b) An item such as \pL+ within a repeated group could cause crashes or
2589 loops.
2590
2591 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2592 "reference to non-existent subpattern" error.
2593
2594 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2595
2596 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2597 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2598
2599 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2600
2601 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2602 character were causing crashes (broken optimization).
2603
2604 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2605 \p or \P) caused a compile-time loop.
2606
2607 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2608 break. For example, the unstudied pattern [\r\n]A does not match the string
2609 "\r\nA" because change 7.0/46 below moves the current point on by two
2610 characters after failing to match at the start. However, the pattern \nA
2611 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2612 the same is true. There doesn't seem any very clean way out of this, but
2613 what I have chosen to do makes the common cases work: PCRE now takes note
2614 of whether there can be an explicit match for \r or \n anywhere in the
2615 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2616 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2617 pattern has explicit CR or LF references.
2618
2619 29. Added (*CR) etc for changing newline setting at start of pattern.
2620
2621
2622 Version 7.2 19-Jun-07
2623 ---------------------
2624
2625 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2626 which is apparently normally available under Windows.
2627
2628 2. Re-jig the pcregrep tests with different newline settings in an attempt
2629 to make them independent of the local environment's newline setting.
2630
2631 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2632
2633 4. Some of the "internals" tests were previously cut out when the link size
2634 was not 2, because the output contained actual offsets. The recent new
2635 "Z" feature of pcretest means that these can be cut out, making the tests
2636 usable with all link sizes.
2637
2638 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2639 stack recursion. This gives a massive performance boost under BSD, but just
2640 a small improvement under Linux. However, it saves one field in the frame
2641 in all cases.
2642
2643 6. Added more features from the forthcoming Perl 5.10:
2644
2645 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2646 recursion call. It refers to the nth most recently opened parentheses.
2647
2648 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2649 to be opened parentheses.
2650
2651 (c) Conditions that refer to capturing parentheses can be specified
2652 relatively, for example, (?(-2)... or (?(+3)...
2653
2654 (d) \K resets the start of the current match so that everything before
2655 is not part of it.
2656
2657 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2658
2659 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2660 reference syntax.
2661
2662 (g) (?| introduces a group in which the numbering of parentheses in each
2663 alternative starts with the same number.
2664
2665 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2666
2667 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2668 PCRE_INFO_JCHANGED.
2669
2670 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2671 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2672 for detecting groups that can match an empty string.
2673
2674 9. A pattern with a very large number of alternatives (more than several
2675 hundred) was running out of internal workspace during the pre-compile
2676 phase, where pcre_compile() figures out how much memory will be needed. A
2677 bit of new cunning has reduced the workspace needed for groups with
2678 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2679 workspace instead of running out of the 4096 that are available.
2680
2681 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2682
2683 11. Applied patch from Google to remove an optimization that didn't quite work.
2684 The report of the bug said:
2685
2686 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2687 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2688 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2689
2690 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2691 it matched the wrong number of bytes.
2692
2693
2694 Version 7.1 24-Apr-07
2695 ---------------------
2696
2697 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2698 that is more "standard", making use of automake and other Autotools. There
2699 is some re-arrangement of the files and adjustment of comments consequent
2700 on this.
2701
2702 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2703 for recursive directory scanning broke on some systems because the files
2704 are not scanned in any specific order and on different systems the order
2705 was different. A call to "sort" has been inserted into RunGrepTest for the
2706 approprate test as a short-term fix. In the longer term there may be an
2707 alternative.
2708
2709 3. I had an email from Eric Raymond about problems translating some of PCRE's
2710 man pages to HTML (despite the fact that I distribute HTML pages, some
2711 people do their own conversions for various reasons). The problems
2712 concerned the use of low-level troff macros .br and .in. I have therefore
2713 removed all such uses from the man pages (some were redundant, some could
2714 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2715 HTML has been updated to handle .nf/.fi and to complain if it encounters
2716 .br or .in.
2717
2718 4. Updated comments in configure.ac that get placed in config.h.in and also
2719 arranged for config.h to be included in the distribution, with the name
2720 config.h.generic, for the benefit of those who have to compile without
2721 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2722
2723 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2724 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2725 makevp.bat for the current PCRE, using the additional files
2726 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2727
2728 6. A Windows user reported a minor discrepancy with test 2, which turned out
2729 to be caused by a trailing space on an input line that had got lost in his
2730 copy. The trailing space was an accident, so I've just removed it.
2731
2732 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2733 that is needed.
2734
2735 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2736 as "const" (a) because they are and (b) because it helps the PHP
2737 maintainers who have recently made a script to detect big data structures
2738 in the php code that should be moved to the .rodata section. I remembered
2739 to update Builducptable as well, so it won't revert if ucptable.h is ever
2740 re-created.
2741
2742 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2743 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2744 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2745 support is not required. This saves 1.5-2K of code, which is important in
2746 some applications.
2747
2748 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2749 so as not to refer to the tables, even though these functions will never be
2750 called when UTF-8 support is disabled. Otherwise there are problems with a
2751 shared library.
2752
2753 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2754
2755 (a) It was defining its arguments as char * instead of void *.
2756
2757 (b) It was assuming that all moves were upwards in memory; this was true
2758 a long time ago when I wrote it, but is no longer the case.
2759
2760 The emulated memove() is provided for those environments that have neither
2761 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2762 is clearly not the case, as these two bugs were recently reported.
2763
2764 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2765 and Detrail to create the HTML documentation, the .txt form of the man
2766 pages, and it removes trailing spaces from listed files. It also creates
2767 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2768 case, it wraps all the #defines with #ifndefs. This script should be run
2769 before "make dist".
2770
2771 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2772 with Unicode property support.
2773
2774 (a) For a maximizing quantifier, if the two different cases of the
2775 character were of different lengths in their UTF-8 codings (there are
2776 some cases like this - I found 11), and the matching function had to
2777 back up over a mixture of the two cases, it incorrectly assumed they
2778 were both the same length.
2779
2780 (b) When PCRE was configured to use the heap rather than the stack for
2781 recursion during matching, it was not correctly preserving the data for
2782 the other case of a UTF-8 character when checking ahead for a match
2783 while processing a minimizing repeat. If the check also involved
2784 matching a wide character, but failed, corruption could cause an
2785 erroneous result when trying to check for a repeat of the original
2786 character.
2787
2788 13. Some tidying changes to the testing mechanism:
2789
2790 (a) The RunTest script now detects the internal link size and whether there
2791 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2792 values substituted by "configure". (The RunGrepTest script already did
2793 this for UTF-8.) The configure.ac script no longer substitutes the
2794 relevant variables.
2795
2796 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2797 with length and offset values. This means that the output is different
2798 for different internal link sizes. Test 2 is skipped for link sizes
2799 other than 2 because of this, bypassing the problem. Unfortunately,
2800 there was also a test in test 3 (the locale tests) that used /B and
2801 failed for link sizes other than 2. Rather than cut the whole test out,
2802 I have added a new /Z option to pcretest that replaces the length and
2803 offset values with spaces. This is now used to make test 3 independent
2804 of link size. (Test 2 will be tidied up later.)
2805
2806 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2807 segmentation fault instead of returning the appropriate error message.
2808
2809 15. In multiline mode when the newline sequence was set to "any", the pattern
2810 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2811 This doesn't seem right; it now treats the CRLF combination as the line
2812 ending, and so does not match in that case. It's only a pattern such as ^$
2813 that would hit this one: something like ^ABC$ would have failed after \r
2814 and then tried again after \r\n.
2815
2816 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2817 in an attempt to make files that differ only in their line terminators
2818 compare equal. This works on Linux.
2819
2820 17. Under certain error circumstances pcregrep might try to free random memory
2821 as it exited. This is now fixed, thanks to valgrind.
2822
2823 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2824 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2825 was because its rules for how to advance for /g after matching an empty
2826 string at the end of a line did not allow for this case. They now check for
2827 it specially.
2828
2829 20. pcretest is supposed to handle patterns and data of any length, by
2830 extending its buffers when necessary. It was getting this wrong when the
2831 buffer for a data line had to be extended.
2832
2833 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2834 CRLF as a newline sequence.
2835
2836 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2837 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2838 I have nevertheless tidied it up.
2839
2840 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2841
2842 24. Added a man page for pcre-config.
2843
2844
2845 Version 7.0 19-Dec-06
2846 ---------------------
2847
2848 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2849 moving to gcc 4.1.1.
2850
2851 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2852 sys/time.h, which is documented as needed for this function. It doesn't
2853 seem to matter on Linux, but it showed up on some releases of OS X.
2854
2855 3. It seems that there are systems where bytes whose values are greater than
2856 127 match isprint() in the "C" locale. The "C" locale should be the
2857 default when a C program starts up. In most systems, only ASCII printing
2858 characters match isprint(). This difference caused the output from pcretest
2859 to vary, making some of the tests fail. I have changed pcretest so that:
2860
2861 (a) When it is outputting text in the compiled version of a pattern, bytes
2862 other than 32-126 are always shown as hex escapes.
2863
2864 (b) When it is outputting text that is a matched part of a subject string,
2865 it does the same, unless a different locale has been set for the match
2866 (using the /L modifier). In this case, it uses isprint() to decide.
2867
2868 4. Fixed a major bug that caused incorrect computation of the amount of memory
2869 required for a compiled pattern when options that changed within the
2870 pattern affected the logic of the preliminary scan that determines the
2871 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2872 that the computed length was too small. The symptoms of this bug were
2873 either the PCRE error "internal error: code overflow" from pcre_compile(),
2874 or a glibc crash with a message such as "pcretest: free(): invalid next
2875 size (fast)". Examples of patterns that provoked this bug (shown in
2876 pcretest format) are:
2877
2878 /(?-x: )/x
2879 /(?x)(?-x: \s*#\s*)/
2880 /((?i)[\x{c0}])/8
2881 /(?i:[\x{c0}])/8
2882
2883 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2884 is now done differently.
2885
2886 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2887 wrapper classes; (b) implement a new function in the C++ scanner that is
2888 more efficient than the old way of doing things because it avoids levels of
2889 recursion in the regex matching; (c) add a paragraph to the documentation
2890 for the FullMatch() function.
2891
2892 6. The escape sequence \n was being treated as whatever was defined as
2893 "newline". Not only was this contrary to the documentation, which states
2894 that \n is character 10 (hex 0A), but it also went horribly wrong when
2895 "newline" was defined as CRLF. This has been fixed.
2896
2897 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2898 was being set to -1 for the "end of line" case (supposedly a value that no
2899 character can have). Though this value is never used (the check for end of
2900 line is "zero bytes in current character"), it caused compiler complaints.
2901 I've changed it to 0xffffffff.
2902
2903 8. In pcre_version.c, the version string was being built by a sequence of
2904 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2905 string (as it is for production releases) called a macro with an empty
2906 argument. The C standard says the result of this is undefined. The gcc
2907 compiler treats it as an empty string (which was what was wanted) but it is
2908 reported that Visual C gives an error. The source has been hacked around to
2909 avoid this problem.
2910
2911 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2912 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2913 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2914 of them did).
2915
2916 10. Originally, pcretest opened its input and output without "b"; then I was
2917 told that "b" was needed in some environments, so it was added for release
2918 5.0 to both the input and output. (It makes no difference on Unix-like
2919 systems.) Later I was told that it is wrong for the input on Windows. I've
2920 now abstracted the modes into two macros, to make it easier to fiddle with
2921 them, and removed "b" from the input mode under Windows.
2922
2923 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2924
2925 12. Added -help and --help to pcretest as an official way of being reminded
2926 of the options.
2927
2928 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2929 and pcrecpp.cc because they annoy compilers at high warning levels.
2930
2931 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2932
2933 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2934 scripts are not C programs :-) and which was not noticed because it works
2935 on Linux.
2936
2937 16. pcretest is supposed to handle any length of pattern and data line (as one
2938 line or as a continued sequence of lines) by extending its input buffer if
2939 necessary. This feature was broken for very long pattern lines, leading to
2940 a string of junk being passed to pcre_compile() if the pattern was longer
2941 than about 50K.
2942
2943 17. I have done a major re-factoring of the way pcre_compile() computes the
2944 amount of memory needed for a compiled pattern. Previously, there was code
2945 that made a preliminary scan of the pattern in order to do this. That was
2946 OK when PCRE was new, but as the facilities have expanded, it has become
2947 harder and harder to keep it in step with the real compile phase, and there
2948 have been a number of bugs (see for example, 4 above). I have now found a
2949 cunning way of running the real compile function in a "fake" mode that
2950 enables it to compute how much memory it would need, while actually only
2951 ever using a few hundred bytes of working memory and without too many
2952 tests of the mode. This should make future maintenance and development
2953 easier. A side effect of this work is that the limit of 200 on the nesting
2954 depth of parentheses has been removed (though this was never a serious
2955 limitation, I suspect). However, there is a downside: pcre_compile() now
2956 runs more slowly than before (30% or more, depending on the pattern). I
2957 hope this isn't a big issue. There is no effect on runtime performance.
2958
2959 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2960 newline (only possible for the last line of a file) and it was a
2961 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2962
2963 19. Added additional timing features to pcretest. (1) The -tm option now times
2964 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2965 separate command line item, by a number that specifies the number of
2966 repeats to use when timing. The default is 50000; this gives better
2967 precision, but takes uncomfortably long for very large patterns.
2968
2969 20. Extended pcre_study() to be more clever in cases where a branch of a
2970 subpattern has no definite first character. For example, (a*|b*)[cd] would
2971 previously give no result from pcre_study(). Now it recognizes that the
2972 first character must be a, b, c, or d.
2973
2974 21. There was an incorrect error "recursive call could loop indefinitely" if
2975 a subpattern (or the entire pattern) that was being tested for matching an
2976 empty string contained only one non-empty item after a nested subpattern.
2977 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2978 incorrectly, because the \d was being skipped in the check.
2979
2980 22. The pcretest program now has a new pattern option /B and a command line
2981 option -b, which is equivalent to adding /B to every pattern. This causes
2982 it to show the compiled bytecode, without the additional information that
2983 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2984 is the same as /B/I).
2985
2986 23. A new optimization is now able automatically to treat some sequences such
2987 as a*b as a*+b. More specifically, if something simple (such as a character
2988 or a simple class like \d) has an unlimited quantifier, and is followed by
2989 something that cannot possibly match the quantified thing, the quantifier
2990 is automatically "possessified".
2991
2992 24. A recursive reference to a subpattern whose number was greater than 39
2993 went wrong under certain circumstances in UTF-8 mode. This bug could also
2994 have affected the operation of pcre_study().
2995
2996 25. Realized that a little bit of performance could be had by replacing
2997 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2998
2999 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3000
3001 27. Possessive quantifiers such as a++ were previously implemented by turning
3002 them into atomic groups such as ($>a+). Now they have their own opcodes,
3003 which improves performance. This includes the automatically created ones
3004 from 23 above.
3005
3006 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3007 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3008 the first matched character to be a colon. This applied both to named and
3009 numbered groups.
3010
3011 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3012
3013 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3014 building PCRE on Windows easier, so I have included it in the distribution.
3015
3016 31. There is now a check in pcretest against a ridiculously large number being
3017 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3018 loop, the loop is abandoned.
3019
3020 32. Forward references to subpatterns in conditions such as (?(2)...) where
3021 subpattern 2 is defined later cause pcre_compile() to search forwards in
3022 the pattern for the relevant set of parentheses. This search went wrong
3023 when there were unescaped parentheses in a character class, parentheses
3024 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3025
3026 33. "Subroutine" calls and backreferences were previously restricted to
3027 referencing subpatterns earlier in the regex. This restriction has now
3028 been removed.
3029
3030 34. Added a number of extra features that are going to be in Perl 5.10. On the
3031 whole, these are just syntactic alternatives for features that PCRE had
3032 previously implemented using the Python syntax or my own invention. The
3033 other formats are all retained for compatibility.
3034
3035 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3036 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3037 also .NET compatible.
3038
3039 (b) A recursion or subroutine call to a named group can now be defined as
3040 (?&name) as well as (?P>name).
3041
3042 (c) A backreference to a named group can now be defined as \k<name> or
3043 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3044 5.10, are also .NET compatible.
3045
3046 (d) A conditional reference to a named group can now use the syntax
3047 (?(<name>) or (?('name') as well as (?(name).
3048
3049 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3050 groups (named and numbered) that are never evaluated inline, but can be
3051 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3052 is always false. There may be only one alternative in such a group.
3053
3054 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3055 as the simple (?(R). The condition is true only if the most recent
3056 recursion is that of the given number or name. It does not search out
3057 through the entire recursion stack.
3058
3059 (g) The escape \gN or \g{N} has been added, where N is a positive or
3060 negative number, specifying an absolute or relative reference.
3061
3062 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3063 some "unreachable code" warnings.
3064
3065 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3066 things, this adds five new scripts.
3067
3068 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3069 There were also incompatibilities regarding the handling of \Q..\E inside
3070 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3071 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3072
3073 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3074 matches an empty string, and forcibly breaks the loop. There were bugs in
3075 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3076 against aaaa the result was just "a" rather than "aaaa", for example. Two
3077 separate and independent bugs (that affected different cases) have been
3078 fixed.
3079
3080 39. Refactored the code to abolish the use of different opcodes for small
3081 capturing bracket numbers. This is a tidy that I avoided doing when I
3082 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3083 The new approach is not only tidier, it makes it possible to reduce the
3084 memory needed to fix the previous bug (38).
3085
3086 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3087 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3088 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3089 mode.
3090
3091 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3092 report.
3093
3094 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3095 copy construction and assignment in the C++ wrapper.
3096
3097 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3098 couple of bugs that could have given wrong results in the "--newline=crlf"
3099 case.
3100
3101 44. Added a number of casts and did some reorganization of signed/unsigned int
3102 variables following suggestions from Dair Grant. Also renamed the variable
3103 "this" as "item" because it is a C++ keyword.
3104
3105 45. Arranged for dftables to add
3106
3107 #include "pcre_internal.h"
3108
3109 to pcre_chartables.c because without it, gcc 4.x may remove the array
3110 definition from the final binary if PCRE is built into a static library and
3111 dead code stripping is activated.
3112
3113 46. For an unanchored pattern, if a match attempt fails at the start of a
3114 newline sequence, and the newline setting is CRLF or ANY, and the next two
3115 characters are CRLF, advance by two characters instead of one.
3116
3117
3118 Version 6.7 04-Jul-06
3119 ---------------------
3120
3121 1. In order to handle tests when input lines are enormously long, pcretest has
3122 been re-factored so that it automatically extends its buffers when
3123 necessary. The code is crude, but this _is_ just a test program. The
3124 default size has been increased from 32K to 50K.
3125
3126 2. The code in pcre_study() was using the value of the re argument before
3127 testing it for NULL. (Of course, in any sensible call of the function, it
3128 won't be NULL.)
3129
3130 3. The memmove() emulation function in pcre_internal.h, which is used on
3131 systems that lack both memmove() and bcopy() - that is, hardly ever -
3132 was missing a "static" storage class specifier.
3133
3134 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3135 containing an extended class (one that cannot be represented by a bitmap
3136 because it contains high-valued characters or Unicode property items, e.g.
3137 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3138 pattern, but PCRE should not loop if you do not (it no longer does).
3139 [Detail: two cases were found: (a) a repeated subpattern containing an
3140 extended class; (b) a recursive reference to a subpattern that followed a
3141 previous extended class. It wasn't skipping over the extended class
3142 correctly when UTF-8 mode was not set.]
3143
3144 5. A negated single-character class was not being recognized as fixed-length
3145 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3146 compile error "lookbehind assertion is not fixed length".
3147
3148 6. The RunPerlTest auxiliary script was showing an unexpected difference
3149 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3150 write a Perl script that can interpret lines of an input file either as
3151 byte characters or as UTF-8, which is what "perltest" was being required to
3152 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3153 can't do is switch easily at run time between having the "use utf8;" pragma
3154 or not. In the end, I fudged it by using the RunPerlTest script to insert
3155 "use utf8;" explicitly for the UTF-8 tests.
3156
3157 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3158 the end of the subject string, contrary to the documentation and to what
3159 Perl does. This was true of both matching functions. Now it matches only at
3160 the start of the subject and immediately after *internal* newlines.
3161
3162 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3163 a pointer to an int instead of a pointer to an unsigned long int. This
3164 caused problems on 64-bit systems.
3165
3166 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3167 instance of the 'standard' template library not being so standard".
3168
3169 10. There was no check on the number of named subpatterns nor the maximum
3170 length of a subpattern name. The product of these values is used to compute
3171 the size of the memory block for a compiled pattern. By supplying a very
3172 long subpattern name and a large number of named subpatterns, the size
3173 computation could be caused to overflow. This is now prevented by limiting
3174 the length of names to 32 characters, and the number of named subpatterns
3175 to 10,000.
3176
3177 11. Subpatterns that are repeated with specific counts have to be replicated in
3178 the compiled pattern. The size of memory for this was computed from the
3179 length of the subpattern and the repeat count. The latter is limited to
3180 65535, but there was no limit on the former, meaning that integer overflow
3181 could in principle occur. The compiled length of a repeated subpattern is
3182 now limited to 30,000 bytes in order to prevent this.
3183
3184 12. Added the optional facility to have named substrings with the same name.
3185
3186 13. Added the ability to use a named substring as a condition, using the
3187 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3188 are numbers (not recommended). Forward references are permitted.
3189
3190 14. Added forward references in named backreferences (if you see what I mean).
3191
3192 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3193 pattern could run off the end of the subject. For example, the pattern
3194 "(?s)(.{1,5})"8 did this with the subject "ab".
3195
3196 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3197 PCRE_CASELESS was set when matching characters that were quantified with ?
3198 or *.
3199
3200 17. A character class other than a single negated character that had a minimum
3201 but no maximum quantifier - for example [ab]{6,} - was not handled
3202 correctly by pce_dfa_exec(). It would match only one character.
3203
3204 18. A valid (though odd) pattern that looked like a POSIX character
3205 class but used an invalid character after [ (for example [[,abc,]]) caused
3206 pcre_compile() to give the error "Failed: internal error: code overflow" or
3207 in some cases to crash with a glibc free() error. This could even happen if
3208 the pattern terminated after [[ but there just happened to be a sequence of
3209 letters, a binary zero, and a closing ] in the memory that followed.
3210
3211 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3212 over the years. Originally (before any Unicode support), just the bottom 8
3213 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3214 output from "man perlunicode" includes this:
3215
3216 The regular expression compiler produces polymorphic opcodes. That
3217 is, the pattern adapts to the data and automatically switches to
3218 the Unicode character scheme when presented with Unicode data--or
3219 instead uses a traditional byte scheme when presented with byte
3220 data.
3221
3222 Sadly, a wide octal escape does not cause a switch, and in a string with
3223 no other multibyte characters, these octal escapes are treated as before.
3224 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3225 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3226 Unicode string.
3227
3228 I have not perpetrated such confusion in PCRE. Up till now, it took just
3229 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3230 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3231 translate to the appropriate multibyte character.
3232
3233 29. Applied some refactoring to reduce the number of warnings from Microsoft
3234 and Borland compilers. This has included removing the fudge introduced
3235 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3236 a warning about an unused variable.
3237
3238 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3239 characters since release 4.0, because Perl (from release 5.004) does not.
3240 [Or at least, is documented not to: some releases seem to be in conflict
3241 with the documentation.] However, when a pattern was studied with
3242 pcre_study() and all its branches started with \s, PCRE still included VT
3243 as a possible starting character. Of course, this did no harm; it just
3244 caused an unnecessary match attempt.
3245
3246 22. Removed a now-redundant internal flag bit that recorded the fact that case
3247 dependency changed within the pattern. This was once needed for "required
3248 byte" processing, but is no longer used. This recovers a now-scarce options
3249 bit. Also moved the least significant internal flag bit to the most-
3250 significant bit of the word, which was not previously used (hangover from
3251 the days when it was an int rather than a uint) to free up another bit for
3252 the future.
3253
3254 23. Added support for CRLF line endings as well as CR and LF. As well as the
3255 default being selectable at build time, it can now be changed at runtime
3256 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3257 specify that it is scanning data with non-default line endings.
3258
3259 24. Changed the definition of CXXLINK to make it agree with the definition of
3260 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3261
3262 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3263 recursions. This makes a big different to stack usage for some patterns.
3264
3265 26. If a subpattern containing a named recursion or subroutine reference such
3266 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3267 the space required for the compiled pattern went wrong and gave too small a
3268 value. Depending on the environment, this could lead to "Failed: internal
3269 error: code overflow at offset 49" or "glibc detected double free or
3270 corruption" errors.
3271
3272 27. Applied patches from Google (a) to support the new newline modes and (b) to
3273 advance over multibyte UTF-8 characters in GlobalReplace.
3274
3275 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3276 difference for some implementation of PCRE in some Windows version.
3277
3278 29. Added some extra testing facilities to pcretest:
3279
3280 \q<number> in a data line sets the "match limit" value
3281 \Q<number> in a data line sets the "match recursion limt" value
3282 -S <number> sets the stack size, where <number> is in megabytes
3283
3284 The -S option isn't available for Windows.
3285
3286
3287 Version 6.6 06-Feb-06
3288 ---------------------
3289
3290 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3291 in pcreposix.h. I have copied the definition from pcre.h.
3292
3293 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3294 because pcre.h is no longer a built file.
3295
3296 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3297 not normally included in the compiled code.
3298
3299
3300 Version 6.5 01-Feb-06
3301 ---------------------
3302
3303 1. When using the partial match feature with pcre_dfa_exec(), it was not
3304 anchoring the second and subsequent partial matches at the new starting
3305 point. This could lead to incorrect results. For example, with the pattern
3306 /1234/, partially matching against "123" and then "a4" gave a match.
3307
3308 2. Changes to pcregrep:
3309
3310 (a) All non-match returns from pcre_exec() were being treated as failures
3311 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3312 error message is output. Some extra information is given for the
3313 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3314 probably the only errors that are likely to be caused by users (by
3315 specifying a regex that has nested indefinite repeats, for instance).
3316 If there are more than 20 of these errors, pcregrep is abandoned.
3317
3318 (b) A binary zero was treated as data while matching, but terminated the
3319 output line if it was written out. This has been fixed: binary zeroes
3320 are now no different to any other data bytes.
3321
3322 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3323 used to set a locale for matching. The --locale=xxxx long option has
3324 been added (no short equivalent) to specify a locale explicitly on the
3325 pcregrep command, overriding the environment variables.
3326
3327 (d) When -B was used with -n, some line numbers in the output were one less
3328 than they should have been.
3329
3330 (e) Added the -o (--only-matching) option.
3331
3332 (f) If -A or -C was used with -c (count only), some lines of context were
3333 accidentally printed for the final match.
3334
3335 (g) Added the -H (--with-filename) option.
3336
3337 (h) The combination of options -rh failed to suppress file names for files
3338 that were found from directory arguments.
3339
3340 (i) Added the -D (--devices) and -d (--directories) options.
3341
3342 (j) Added the -F (--fixed-strings) option.
3343
3344 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3345
3346 (l) Added the --colo(u)r option.
3347
3348 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3349 is not present by default.
3350
3351 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3352 items such as (?R) or (?1), when the recursion could match a number of
3353 alternatives. If it matched one of the alternatives, but subsequently,
3354 outside the recursion, there was a failure, the code tried to back up into
3355 the recursion. However, because of the way PCRE is implemented, this is not
3356 possible, and the result was an incorrect result from the match.
3357
3358 In order to prevent this happening, the specification of recursion has
3359 been changed so that all such subpatterns are automatically treated as
3360 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3361
3362 4. I had overlooked the fact that, in some locales, there are characters for
3363 which isalpha() is true but neither isupper() nor islower() are true. In
3364 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3365 and ordfeminine) are like this. This affected the treatment of \w and \W
3366 when they appeared in character classes, but not when they appeared outside
3367 a character class. The bit map for "word" characters is now created
3368 separately from the results of isalnum() instead of just taking it from the
3369 upper, lower, and digit maps. (Plus the underscore character, of course.)
3370
3371 5. The above bug also affected the handling of POSIX character classes such as
3372 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3373 permanent tables. Instead, the bit maps for such a class were previously
3374 created as the appropriate unions of the upper, lower, and digit bitmaps.
3375 Now they are created by subtraction from the [[:word:]] class, which has
3376 its own bitmap.
3377
3378 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3379 It is created by subtracting the vertical space characters (\x09, \x0a,
3380 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3381 subtraction was done in the overall bitmap for a character class, meaning
3382 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3383 be recognized. This bug has been fixed.
3384
3385 7. Patches from the folks at Google:
3386
3387 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3388 real life, but is still worth protecting against".
3389
3390 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3391 regular expressions".
3392
3393 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3394 have it.
3395
3396 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3397 "configure" and the latter not, in order to fix a problem somebody had
3398 with compiling the Arg class on HP-UX.
3399
3400 (e) Improve the error-handling of the C++ wrapper a little bit.
3401
3402 (f) New tests for checking recursion limiting.
3403
3404 8. The pcre_memmove() function, which is used only if the environment does not
3405 have a standard memmove() function (and is therefore rarely compiled),
3406 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3407 returning a result (though PCRE never actually uses the result).
3408
3409 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3410 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3411 returned instead of calling malloc() with an overflowing number that would
3412 most likely cause subsequent chaos.
3413
3414 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3415
3416 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3417 with this option is matched, the nmatch and pmatch options of regexec() are
3418 ignored.
3419
3420 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3421 provided in case anyone wants to the the POSIX interface with UTF-8
3422 strings.
3423
3424 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3425 C++ linking (needed for some HP-UX environments).
3426
3427 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3428 (unused parameter) and in the pcre_printint() function (omitted "default"
3429 switch label when the default is to do nothing).
3430
3431 15. Added some code to make it possible, when PCRE is compiled as a C++
3432 library, to replace subject pointers for pcre_exec() with a smart pointer
3433 class, thus making it possible to process discontinuous strings.
3434
3435 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3436 much the same function. They were added by different people who were trying
3437 to make PCRE easy to compile on non-Unix systems. It has been suggested
3438 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3439 for compiling on Windows systems. I have therefore replaced it with
3440 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3441 defaults to "extern" for C or "extern C" for C++, which works fine on
3442 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3443 SCOPE with something explicit in config.h. In addition:
3444
3445 (a) pcreposix.h still had just "extern" instead of either of these macros;
3446 I have replaced it with PCRE_DATA_SCOPE.
3447
3448 (b) Functions such as _pcre_xclass(), which are internal to the library,
3449 but external in the C sense, all had PCRE_EXPORT in their definitions.
3450 This is apparently wrong for the Windows case, so I have removed it.
3451 (It makes no difference on Unix-like systems.)
3452
3453 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3454 of recursive calls to match(). This is different to MATCH_LIMIT because
3455 that limits the total number of calls to match(), not all of which increase
3456 the depth of recursion. Limiting the recursion depth limits the amount of
3457 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3458 when PCRE is compiled, and changed at run time. A patch from Google adds
3459 this functionality to the C++ interface.
3460
3461 18. Changes to the handling of Unicode character properties:
3462
3463 (a) Updated the table to Unicode 4.1.0.
3464
3465 (b) Recognize characters that are not in the table as "Cn" (undefined).
3466
3467 (c) I revised the way the table is implemented to a much improved format
3468 which includes recognition of ranges. It now supports the ranges that
3469 are defined in UnicodeData.txt, and it also amalgamates other
3470 characters into ranges. This has reduced the number of entries in the
3471 table from around 16,000 to around 3,000, thus reducing its size
3472 considerably. I realized I did not need to use a tree structure after
3473 all - a binary chop search is just as efficient. Having reduced the
3474 number of entries, I extended their size from 6 bytes to 8 bytes to
3475 allow for more data.
3476
3477 (d) Added support for Unicode script names via properties such as \p{Han}.
3478
3479 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3480 matching that character.
3481
3482 20. When matching a repeated Unicode property with a minimum greater than zero,
3483 (for example \pL{2,}), PCRE could look past the end of the subject if it
3484 reached it while seeking the minimum number of characters. This could
3485 happen only if some of the characters were more than one byte long, because
3486 there is a check for at least the minimum number of bytes.
3487
3488 21. Refactored the implementation of \p and \P so as to be more general, to
3489 allow for more different types of property in future. This has changed the
3490 compiled form incompatibly. Anybody with saved compiled patterns that use
3491 \p or \P will have to recompile them.
3492
3493 22. Added "Any" and "L&" to the supported property types.
3494
3495 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3496 but give a compile time error if the value is greater than 0xff.
3497
3498 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3499 accidentally not being installed or uninstalled.
3500
3501 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3502 made were to insert the current release number. This seemed silly, because
3503 it made things harder for people building PCRE on systems that don't run
3504 "configure". I have turned pcre.h into a distributed file, no longer built
3505 by "configure", with the version identification directly included. There is
3506 no longer a pcre.h.in file.
3507
3508 However, this change necessitated a change to the pcre-config script as
3509 well. It is built from pcre-config.in, and one of the substitutions was the
3510 release number. I have updated configure.ac so that ./configure now finds
3511 the release number by grepping pcre.h.
3512
3513 26. Added the ability to run the tests under valgrind.
3514
3515
3516 Version 6.4 05-Sep-05
3517 ---------------------
3518
3519 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3520 "--" to be printed when multiple files were scanned, even when none of the
3521 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3522 consider it to be a bug, and have restored the previous behaviour.
3523
3524 2. A couple of code tidies to get rid of compiler warnings.
3525
3526 3. The pcretest program used to cheat by referring to symbols in the library
3527 whose names begin with _pcre_. These are internal symbols that are not
3528 really supposed to be visible externally, and in some environments it is
3529 possible to suppress them. The cheating is now confined to including
3530 certain files from the library's source, which is a bit cleaner.
3531
3532 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3533 file's purpose clearer.
3534
3535 5. Reorganized pcre_ucp_findchar().
3536
3537
3538 Version 6.3 15-Aug-05
3539 ---------------------
3540
3541 1. The file libpcre.pc.in did not have general read permission in the tarball.
3542
3543 2. There were some problems when building without C++ support:
3544
3545 (a) If C++ support was not built, "make install" and "make test" still
3546 tried to test it.
3547
3548 (b) There were problems when the value of CXX was explicitly set. Some
3549 changes have been made to try to fix these, and ...
3550
3551 (c) --disable-cpp can now be used to explicitly disable C++ support.
3552
3553 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3554 backslash in a target when C++ was disabled. This confuses some
3555 versions of "make", apparently. Using an intermediate variable solves
3556 this. (Same for CPP_LOBJ.)
3557
3558 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3559 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3560 necessary on certain architectures.
3561
3562 4. Added a setting of -export-symbols-regex to the link command to remove
3563 those symbols that are exported in the C sense, but actually are local
3564 within the library, and not documented. Their names all begin with
3565 "_pcre_". This is not a perfect job, because (a) we have to except some
3566 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3567 available (and never for static libraries). I have made a note to try to
3568 find a way round (a) in the future.
3569
3570
3571 Version 6.2 01-Aug-05
3572 ---------------------
3573
3574 1. There was no test for integer overflow of quantifier values. A construction
3575 such as {1111111111111111} would give undefined results. What is worse, if
3576 a minimum quantifier for a parenthesized subpattern overflowed and became
3577 negative, the calculation of the memory size went wrong. This could have
3578 led to memory overwriting.
3579
3580 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3581
3582 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3583 operating environments where this matters.
3584
3585 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3586 PCRE options from within the C++ wrapper.
3587
3588 5. Named capturing subpatterns were not being correctly counted when a pattern
3589 was compiled. This caused two problems: (a) If there were more than 100
3590 such subpatterns, the calculation of the memory needed for the whole
3591 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3592 back references of the form \12, where the number was greater than 9, were
3593 not recognized as back references, even though there were sufficient
3594 previous subpatterns.
3595
3596 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3597 versions of gcc, e.g. 2.95.4.
3598
3599
3600 Version 6.1 21-Jun-05
3601 ---------------------
3602
3603 1. There was one reference to the variable "posix" in pcretest.c that was not
3604 surrounded by "#if !defined NOPOSIX".
3605
3606 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3607 the cross-check on the old pcre_info() function, for the benefit of the
3608 cut-down version of PCRE that is currently imported into Exim.
3609
3610 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3611 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3612 patterns that start (?i) but I don't think that matters. The use of (?i) is
3613 just an example; this all applies to the other options as well.
3614
3615 4. Since libtool seems to echo the compile commands it is issuing, the output
3616 from "make" can be reduced a bit by putting "@" in front of each libtool
3617 compile command.
3618
3619 5. Patch from the folks at Google for configure.in to be a bit more thorough
3620 in checking for a suitable C++ installation before trying to compile the
3621 C++ stuff. This should fix a reported problem when a compiler was present,
3622 but no suitable headers.
3623
3624 6. The man pages all had just "PCRE" as their title. I have changed them to
3625 be the relevant file name. I have also arranged that these names are
3626 retained in the file doc/pcre.txt, which is a concatenation in text format
3627 of all the man pages except the little individual ones for each function.
3628
3629 7. The NON-UNIX-USE file had not been updated for the different set of source
3630 files that come with release 6. I also added a few comments about the C++
3631 wrapper.
3632
3633
3634 Version 6.0 07-Jun-05
3635 ---------------------
3636
3637 1. Some minor internal re-organization to help with my DFA experiments.
3638
3639 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3640 didn't matter for the library itself when fully configured, but did matter
3641 when compiling without UCP support, or within Exim, where the ucp files are
3642 not imported.
3643
3644 3. Refactoring of the library code to split up the various functions into
3645 different source modules. The addition of the new DFA matching code (see
3646 below) to a single monolithic source would have made it really too
3647 unwieldy, quite apart from causing all the code to be include in a
3648 statically linked application, when only some functions are used. This is
3649 relevant even without the DFA addition now that patterns can be compiled in
3650 one application and matched in another.
3651
3652 The downside of splitting up is that there have to be some external
3653 functions and data tables that are used internally in different modules of
3654 the library but which are not part of the API. These have all had their
3655 names changed to start with "_pcre_" so that they are unlikely to clash
3656 with other external names.
3657
3658 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3659 a different (DFA) algorithm. Although it is slower than the original
3660 function, it does have some advantages for certain types of matching
3661 problem.
3662
3663 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3664 including restarting after a partial match.
3665
3666 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3667 defined when compiling for Windows was sent to me. I have put it into the
3668 code, though I have no means of testing or verifying it.
3669
3670 7. Added the pcre_refcount() auxiliary function.
3671
3672 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3673 match before or at the first newline in the subject string. In pcretest,
3674 the /f option on a pattern can be used to set this.
3675
3676 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3677 would behave wrongly. This has been present in PCRE since release 4.0.
3678
3679 10. A number of changes to the pcregrep command:
3680
3681 (a) Refactored how -x works; insert ^(...)$ instead of setting
3682 PCRE_ANCHORED and checking the length, in preparation for adding
3683 something similar for -w.
3684
3685 (b) Added the -w (match as a word) option.
3686
3687 (c) Refactored the way lines are read and buffered so as to have more
3688 than one at a time available.
3689
3690 (d) Implemented a pcregrep test script.
3691
3692 (e) Added the -M (multiline match) option. This allows patterns to match
3693 over several lines of the subject. The buffering ensures that at least
3694 8K, or the rest of the document (whichever is the shorter) is available
3695 for matching (and similarly the previous 8K for lookbehind assertions).
3696
3697 (f) Changed the --help output so that it now says
3698
3699 -w, --word-regex(p)
3700
3701 instead of two lines, one with "regex" and the other with "regexp"
3702 because that confused at least one person since the short forms are the
3703 same. (This required a bit of code, as the output is generated
3704 automatically from a table. It wasn't just a text change.)
3705
3706 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3707 option but starts with a hyphen. Could be a pattern or a path name
3708 starting with a hyphen, for instance.
3709
3710 (h) "-" can be given as a file name to represent stdin.
3711
3712 (i) When file names are being printed, "(standard input)" is used for
3713 the standard input, for compatibility with GNU grep. Previously
3714 "<stdin>" was used.
3715
3716 (j) The option --label=xxx can be used to supply a name to be used for
3717 stdin when file names are being printed. There is no short form.
3718
3719 (k) Re-factored the options decoding logic because we are going to add
3720 two more options that take data. Such options can now be given in four
3721 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3722
3723 (l) Added the -A, -B, and -C options for requesting that lines of context
3724 around matches be printed.
3725
3726 (m) Added the -L option to print the names of files that do not contain
3727 any matching lines, that is, the complement of -l.
3728
3729 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3730 continue to scan other files.
3731
3732 (o) The -s option was incorrectly implemented. For compatibility with other
3733 greps, it now suppresses the error message for a non-existent or non-
3734 accessible file (but not the return code). There is a new option called
3735 -q that suppresses the output of matching lines, which was what -s was
3736 previously doing.
3737
3738 (p) Added --include and --exclude options to specify files for inclusion
3739 and exclusion when recursing.
3740
3741 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3742 Hopefully, it now does.
3743
3744 12. Missing cast in pcre_study().
3745
3746 13. Added an "uninstall" target to the makefile.
3747
3748 14. Replaced "extern" in the function prototypes in Makefile.in with
3749 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3750 world, but is set differently for Windows.
3751
3752 15. Added a second compiling function called pcre_compile2(). The only
3753 difference is that it has an extra argument, which is a pointer to an
3754 integer error code. When there is a compile-time failure, this is set
3755 non-zero, in addition to the error test pointer being set to point to an
3756 error message. The new argument may be NULL if no error number is required
3757 (but then you may as well call pcre_compile(), which is now just a
3758 wrapper). This facility is provided because some applications need a
3759 numeric error indication, but it has also enabled me to tidy up the way
3760 compile-time errors are handled in the POSIX wrapper.
3761
3762 16. Added VPATH=.libs to the makefile; this should help when building with one
3763 prefix path and installing with another. (Or so I'm told by someone who
3764 knows more about this stuff than I do.)
3765
3766 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3767 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3768 match everything, including newlines. This is not POSIX-compatible, but
3769 somebody wanted the feature. From pcretest it can be activated by using
3770 both the P and the s flags.
3771
3772 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3773
3774 19. libpcre.pc was being incorrectly installed as executable.
3775
3776 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3777 it now also looks for '\r' so that it will work unmodified on Windows.
3778
3779 21. Added Google's contributed C++ wrapper to the distribution.
3780
3781 22. Added some untidy missing memory free() calls in pcretest, to keep
3782 Electric Fence happy when testing.
3783
3784
3785
3786 Version 5.0 13-Sep-04
3787 ---------------------
3788
3789 1. Internal change: literal characters are no longer packed up into items
3790 containing multiple characters in a single byte-string. Each character
3791 is now matched using a separate opcode. However, there may be more than one
3792 byte in the character in UTF-8 mode.
3793
3794 2. The pcre_callout_block structure has two new fields: pattern_position and
3795 next_item_length. These contain the offset in the pattern to the next match
3796 item, and its length, respectively.
3797
3798 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3799 insertion of callouts before each pattern item. Added the /C option to
3800 pcretest to make use of this.
3801
3802 4. On the advice of a Windows user, the lines
3803
3804 #if defined(_WIN32) || defined(WIN32)
3805 _setmode( _fileno( stdout ), 0x8000 );
3806 #endif /* defined(_WIN32) || defined(WIN32) */
3807
3808 have been added to the source of pcretest. This apparently does useful
3809 magic in relation to line terminators.
3810
3811 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3812 for the benefit of those environments where the "b" makes a difference.
3813
3814 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3815 to know about it. I have put a hack into configure.in that adds in code
3816 to set GCC=yes if CC=icc. This seems to end up at a point in the
3817 generated configure script that is early enough to affect the setting of
3818 compiler options, which is what is needed, but I have no means of testing
3819 whether it really works. (The user who reported this had patched the
3820 generated configure script, which of course I cannot do.)
3821
3822 LATER: After change 22 below (new libtool files), the configure script
3823 seems to know about icc (and also ecc). Therefore, I have commented out
3824 this hack in configure.in.
3825
3826 7. Added support for pkg-config (2 patches were sent in).
3827
3828 8. Negated POSIX character classes that used a combination of internal tables
3829 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3830 [[:^ascii]]. Typically, they would match almost any characters. The other
3831 POSIX classes were not broken in this way.
3832
3833 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3834 to find the match, as PCRE was deluded into thinking that the match had to
3835 start at the start point or following a newline. The same bug applied to
3836 patterns with negative forward assertions or any backward assertions
3837 preceding ".*" at the start, unless the pattern required a fixed first
3838 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3839
3840 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3841 starting at the last subject character, bytes beyond the end of the subject
3842 string were read.
3843
3844 11. Renamed the variable "class" as "classbits" to make life easier for C++
3845 users. (Previously there was a macro definition, but it apparently wasn't
3846 enough.)
3847
3848 12. Added the new field "tables" to the extra data so that tables can be passed
3849 in at exec time, or the internal tables can be re-selected. This allows
3850 a compiled regex to be saved and re-used at a later time by a different
3851 program that might have everything at different addresses.
3852
3853 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3854 -R library as well as a -L library.
3855
3856 14. The debugging options of pcretest (-d on the command line or D on a
3857 pattern) showed incorrect output for anything following an extended class
3858 that contained multibyte characters and which was followed by a quantifier.
3859
3860 15. Added optional support for general category Unicode character properties
3861 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3862 support. It adds about 90K to the size of the library. The meanings of the
3863 inbuilt class escapes such as \d and \s have NOT been changed.
3864
3865 16. Updated pcredemo.c to include calls to free() to release the memory for the
3866 compiled pattern.
3867
3868 17. The generated file chartables.c was being created in the source directory
3869 instead of in the building directory. This caused the build to fail if the
3870 source directory was different from the building directory, and was
3871 read-only.
3872
3873 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3874 file. No doubt somebody will tell me if they don't make sense... Also added
3875 Dan Mooney's comments about building on OpenVMS.
3876
3877 19. Added support for partial matching via the PCRE_PARTIAL option for
3878 pcre_exec() and the \P data escape in pcretest.
3879
3880 20. Extended pcretest with 3 new pattern features:
3881
3882 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3883 write the compiled pattern to the file whose name is "rest-of-line".
3884 This is a straight binary dump of the data, with the saved pointer to
3885 the character tables forced to be NULL. The study data, if any, is
3886 written too. After writing, pcretest reads a new pattern.
3887
3888 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3889 compiled pattern from the given file. There must not be any
3890 occurrences of "<" in the file name (pretty unlikely); if there are,
3891 pcretest will instead treat the initial "<" as a pattern delimiter.
3892 After reading in the pattern, pcretest goes on to read data lines as
3893 usual.
3894
3895 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3896 and 16-bit fields in a compiled pattern, to simulate a pattern that
3897 was compiled on a host of opposite endianness.
3898
3899 21. The pcre-exec() function can now cope with patterns that were compiled on
3900 hosts of opposite endianness, with this restriction:
3901
3902 As for any compiled expression that is saved and used later, the tables
3903 pointer field cannot be preserved; the extra_data field in the arguments
3904 to pcre_exec() should be used to pass in a tables address if a value
3905 other than the default internal tables were used at compile time.
3906
3907 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3908 now diagnosed as an error. Previously, most of the time, a negative number
3909 would have been treated as zero, but if in addition "ovector" was passed as
3910 NULL, a crash could occur.
3911
3912 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3913 new versions from the libtool 1.5 distribution (the last one is a copy of
3914 a file called libtool.m4). This seems to have fixed the need to patch
3915 "configure" to support Darwin 1.3 (which I used to do). However, I still
3916 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3917 workstation).
3918
3919 24. Changed the PCRE licence to be the more standard "BSD" licence.
3920
3921
3922 Version 4.5 01-Dec-03
3923 ---------------------
3924
3925 1. There has been some re-arrangement of the code for the match() function so
3926 that it can be compiled in a version that does not call itself recursively.
3927 Instead, it keeps those local variables that need separate instances for
3928 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3929 needs to "recurse". Keeping track of where control must go is done by means
3930 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3931 hide most of the details from the main code, and operates only if
3932 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3933 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3934 operating.
3935
3936 To make it easier for callers to provide specially tailored get/free
3937 functions for this usage, two new functions, pcre_stack_malloc, and
3938 pcre_stack_free, are used. They are always called in strict stacking order,
3939 and the size of block requested is always the same.
3940
3941 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3942 PCRE has been compiled to use the stack or the heap for recursion. The
3943 -C option of pcretest uses this to show which version is compiled.
3944
3945 A new data escape \S, is added to pcretest; it causes the amounts of store
3946 obtained and freed by both kinds of malloc/free at match time to be added
3947 to the output.
3948
3949 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3950 what's available on my current Linux desktop machine.
3951
3952 3. When matching a UTF-8 string, the test for a valid string at the start has
3953 been extended. If start_offset is not zero, PCRE now checks that it points
3954 to a byte that is the start of a UTF-8 character. If not, it returns
3955 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3956 this is necessary because there may be backward assertions in the pattern.
3957 When matching the same subject several times, it may save resources to use
3958 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3959
3960 4. The code for checking the validity of UTF-8 strings has been tightened so
3961 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3962 containing "overlong sequences".
3963
3964 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3965 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3966 should have been "&", but it just so happened that all the cases this let
3967 through by mistake were picked up later in the function.
3968
3969 6. I had used a variable called "isblank" - this is a C99 function, causing
3970 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3971
3972 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3973 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3974 so the version is shown; (c) stop "make test" echoing "./RunTest".
3975
3976 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3977
3978 9. The prototype for memmove() for systems that don't have it was using
3979 size_t, but the inclusion of the header that defines size_t was later. I've
3980 moved the #includes for the C headers earlier to avoid this.
3981
3982 10. Added some adjustments to the code to make it easier to compiler on certain
3983 special systems:
3984
3985 (a) Some "const" qualifiers were missing.
3986 (b) Added the macro EXPORT before all exported functions; by default this
3987 is defined to be empty.
3988 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3989 that it reads its output file name as an argument instead of writing
3990 to the standard output and assuming this can be redirected.
3991
3992 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3993 class containing characters with values greater than 255, PCRE compilation
3994 went into a loop.
3995
3996 12. A recursive reference to a subpattern that was within another subpattern
3997 that had a minimum quantifier of zero caused PCRE to crash. For example,
3998 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3999 recursion. If the recursively-called subpattern itself had a zero repeat,
4000 that was OK.
4001
4002 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4003 buffer into which it was copied (for escape processing) was still set at
4004 1024, so long lines caused crashes.
4005
4006 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4007 "internal error: code overflow...". This applied to any character class
4008 that was followed by a possessive quantifier.
4009
4010 15. Modified the Makefile to add libpcre.la as a prerequisite for
4011 libpcreposix.la because I was told this is needed for a parallel build to
4012 work.
4013
4014 16. If a pattern that contained .* following optional items at the start was
4015 studied, the wrong optimizing data was generated, leading to matching
4016 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4017 matching string must start with a or b or c. The correct conclusion for
4018 this pattern is that a match can start with any character.
4019
4020
4021 Version 4.4 13-Aug-03
4022 ---------------------
4023
4024 1. In UTF-8 mode, a character class containing characters with values between
4025 127 and 255 was not handled correctly if the compiled pattern was studied.
4026 In fixing this, I have also improved the studying algorithm for such
4027 classes (slightly).
4028
4029 2. Three internal functions had redundant arguments passed to them. Removal
4030 might give a very teeny performance improvement.
4031
4032 3. Documentation bug: the value of the capture_top field in a callout is *one
4033 more than* the number of the hightest numbered captured substring.
4034
4035 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4036 in incorrectly linking with a previously installed version. They now link
4037 explicitly with libpcre.la.
4038
4039 5. configure.in no longer needs to recognize Cygwin specially.
4040
4041 6. A problem in pcre.in for Windows platforms is fixed.
4042
4043 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4044 pcretest, it used to include the size of the study block as part of its
4045 output. Unfortunately, the structure contains a field that has a different
4046 size on different hardware architectures. This meant that the tests that
4047 showed this size failed. As the block is currently always of a fixed size,
4048 this information isn't actually particularly useful in pcretest output, so
4049 I have just removed it.
4050
4051 8. Three pre-processor statements accidentally did not start in column 1.
4052 Sadly, there are *still* compilers around that complain, even though
4053 standard C has not required this for well over a decade. Sigh.
4054
4055 9. In pcretest, the code for checking callouts passed small integers in the
4056 callout_data field, which is a void * field. However, some picky compilers
4057 complained about the casts involved for this on 64-bit systems. Now
4058 pcretest passes the address of the small integer instead, which should get
4059 rid of the warnings.
4060
4061 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4062 both compile and run time, and gives an error if an invalid UTF-8 sequence
4063 is found. There is a option for disabling this check in cases where the
4064 string is known to be correct and/or the maximum performance is wanted.
4065
4066 11. In response to a bug report, I changed one line in Makefile.in from
4067
4068 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4069 to
4070 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4071
4072 to look similar to other lines, but I have no way of telling whether this
4073 is the right thing to do, as I do not use Windows. No doubt I'll get told
4074 if it's wrong...
4075
4076
4077 Version 4.3 21-May-03
4078 ---------------------
4079
4080 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4081 Makefile.
4082
4083 2. Some refactoring to improve the quality of the code:
4084
4085 (i) The utf8_table... variables are now declared "const".
4086
4087 (ii) The code for \cx, which used the "case flipping" table to upper case
4088 lower case letters, now just substracts 32. This is ASCII-specific,
4089 but the whole concept of \cx is ASCII-specific, so it seems
4090 reasonable.
4091
4092 (iii) PCRE was using its character types table to recognize decimal and
4093 hexadecimal digits in the pattern. This is silly, because it handles
4094 only 0-9, a-f, and A-F, but the character types table is locale-
4095 specific, which means strange things might happen. A private
4096 table is now used for this - though it costs 256 bytes, a table is
4097 much faster than multiple explicit tests. Of course, the standard
4098 character types table is still used for matching digits in subject
4099 strings against \d.
4100
4101 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4102 ending in _t are). So I've renamed it as ESC_tee.
4103
4104 3. The first argument for regexec() in the POSIX wrapper should have been
4105 defined as "const".
4106
4107 4. Changed pcretest to use malloc() for its buffers so that they can be
4108 Electric Fenced for debugging.
4109
4110 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4111 to read one or more bytes before the start of the subject string. Often this
4112 had no effect on PCRE's behaviour, but in some circumstances it could
4113 provoke a segmentation fault.
4114
4115 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4116 to try to read one or more bytes before the start of the subject string.
4117
4118 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4119 UTF-8 support could misbehave in various ways if the subject string
4120 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4121 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4122 back over UTF-8 characters.)
4123
4124
4125 Version 4.2 14-Apr-03
4126 ---------------------
4127
4128 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4129
4130 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4131 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4132 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4133 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4134 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4135 and BUILD_EXEEXT
4136 Note: automatic setting of the BUILD variables is not yet working
4137 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4138 compile-time but not at link-time
4139 [LINK]: use for linking executables only
4140 make different versions for Windows and non-Windows
4141 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4142 libraries
4143 [LINK_FOR_BUILD]: new variable
4144 [OBJEXT]: use throughout
4145 [EXEEXT]: use throughout
4146 <winshared>: new target
4147 <wininstall>: new target
4148 <dftables.o>: use native compiler
4149 <dftables>: use native linker
4150 <install>: handle Windows platform correctly
4151 <clean>: ditto
4152 <check>: ditto
4153 copy DLL to top builddir before testing
4154
4155 As part of these changes, -no-undefined was removed again. This was reported
4156 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4157 in any case.
4158
4159 3. Some tidies to get rid of compiler warnings:
4160
4161 . In the match_data structure, match_limit was an unsigned long int, whereas
4162 match_call_count was an int. I've made them both unsigned long ints.
4163
4164 . In pcretest the fact that a const uschar * doesn't automatically cast to
4165 a void * provoked a warning.
4166
4167 . Turning on some more compiler warnings threw up some "shadow" variables
4168 and a few more missing casts.
4169
4170 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4171 option, a class that contained a single character with a value between 128
4172 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4173
4174 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4175 option, a class that contained several characters, but with at least one
4176 whose value was between 128 and 255 caused PCRE to crash.
4177
4178
4179 Version 4.1 12-Mar-03
4180 ---------------------
4181
4182 1. Compiling with gcc -pedantic found a couple of places where casts were
4183 needed, and a string in dftables.c that was longer than standard compilers are
4184 required to support.
4185
4186 2. Compiling with Sun's compiler found a few more places where the code could
4187 be tidied up in order to avoid warnings.
4188
4189 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4190 first of these names is deprecated in the latest Autoconf in favour of the name
4191 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4192 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4193 analogy I have changed it to CFLAGS_FOR_BUILD.
4194
4195 4. Added -no-undefined to the linking command in the Makefile, because this is
4196 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4197 linking step for the pcreposix library.
4198
4199 5. PCRE was failing to diagnose the case of two named groups with the same
4200 name.
4201
4202 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4203 literal character that is needed in the subject for a match, and scans along to
4204 ensure that it is present before embarking on the full matching process. This
4205 saves time in cases of nested unlimited repeats that are never going to match.
4206 Problem: the scan can take a lot of time if the subject is very long (e.g.
4207 megabytes), thus penalizing straightforward matches. It is now done only if the
4208 amount of subject to be scanned is less than 1000 bytes.
4209
4210 7. A lesser problem with the same optimization is that it was recording the
4211 first character of an anchored pattern as "needed", thus provoking a search
4212 right along the subject, even when the first match of the pattern was going to
4213 fail. The "needed" character is now not set for anchored patterns, unless it
4214 follows something in the pattern that is of non-fixed length. Thus, it still
4215 fulfils its original purpose of finding quick non-matches in cases of nested
4216 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4217
4218
4219 Version 4.0 17-Feb-03
4220 ---------------------
4221
4222 1. If a comment in an extended regex that started immediately after a meta-item
4223 extended to the end of string, PCRE compiled incorrect data. This could lead to
4224 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4225
4226 2. Moved to autoconf 2.53 and libtool 1.4.2.
4227
4228 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4229 the special perltest8 script is no longer needed - all the tests can be run
4230 from a single perltest script.
4231
4232 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4233 by \s. It has now been removed in PCRE. This means it isn't recognized as
4234 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4235 class [:space:] *does* include VT, thereby creating a mess.
4236
4237 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4238 space and tab.
4239
4240 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4241 its new features into the main test script, reducing the number of scripts.
4242
4243 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4244 were backward compatible, and made the (?i) apply to the whole pattern, as if
4245 /i were given. Now it behaves more logically, and applies the option setting
4246 only to what follows. PCRE has been changed to follow suit. However, if it
4247 finds options settings right at the start of the pattern, it extracts them into
4248 the global options, as before. Thus, they show up in the info data.
4249
4250 8. Added support for the \Q...\E escape sequence. Characters in between are
4251 treated as literals. This is slightly different from Perl in that $ and @ are
4252 also handled as literals inside the quotes. In Perl, they will cause variable
4253 interpolation. Note the following examples:
4254
4255 Pattern PCRE matches Perl matches
4256
4257 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4258 \Qabc\$xyz\E abc\$xyz abc\$xyz
4259 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4260
4261 For compatibility with Perl, \Q...\E sequences are recognized inside character
4262 classes as well as outside them.
4263
4264 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4265 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4266 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4267 signed/unsigned warnings.
4268
4269 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4270 option for pcretest, so I've replaced it by a simple function that does just
4271 that job.
4272
4273 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4274 "pcregrep -".
4275
4276 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4277 Java package. This provides some syntactic sugar for simple cases of what my
4278 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4279 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4280 item, you can use this simplified notation. Note that only makes sense with
4281 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4282 greediness, whatever the setting of the PCRE_UNGREEDY option.
4283
4284 13. A change of greediness default within a pattern was not taking effect at
4285 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4286 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4287 was abstracted outside.
4288
4289 14. PCRE now supports the \G assertion. It is true when the current matching
4290 position is at the start point of the match. This differs from \A when the
4291 starting offset is non-zero. Used with the /g option of pcretest (or similar
4292 code), it works in the same way as it does for Perl's /g option. If all
4293 alternatives of a regex begin with \G, the expression is anchored to the start
4294 match position, and the "anchored" flag is set in the compiled expression.
4295
4296 15. Some bugs concerning the handling of certain option changes within patterns
4297 have been fixed. These applied to options other than (?ims). For example,
4298 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4299 been the other way round. Some of this was related to change 7 above.
4300
4301 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4302 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4303 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4304 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4305
4306 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4307 mode. Unlike ".", it always matches newline, whatever the setting of
4308 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4309 assertions. Perl allows it, but it doesn't (in general) work because it can't
4310 calculate the length of the lookbehind. At least, that's the case for Perl
4311 5.8.0 - I've been told they are going to document that it doesn't work in
4312 future.
4313
4314 18. Added an error diagnosis for escapes that PCRE does not support: these are
4315 \L, \l, \N, \P, \p, \U, \u, and \X.
4316
4317 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4318 reading past the end of the pattern in cases such as /[abcd/.
4319
4320 20. PCRE was getting more memory than necessary for patterns with classes that
4321 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4322
4323 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4324 compiling PCRE for use with Virtual Pascal.
4325
4326 22. Small fix to the Makefile to make it work properly if the build is done
4327 outside the source tree.
4328
4329 23. Added a new extension: a condition to go with recursion. If a conditional
4330 subpattern starts with (?(R) the "true" branch is used if recursion has
4331 happened, whereas the "false" branch is used only at the top level.
4332
4333 24. When there was a very long string of literal characters (over 255 bytes
4334 without UTF support, over 250 bytes with UTF support), the computation of how
4335 much memory was required could be incorrect, leading to segfaults or other
4336 strange effects.
4337
4338 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4339 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4340 there was a subsequent back reference to those brackets. This meant that, for
4341 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4342 possible to check for precisely this case. All we can do is abandon the
4343 optimization if .* occurs inside capturing brackets when there are any back
4344 references whatsoever. (See below for a better fix that came later.)
4345
4346 26. The handling of the optimization for finding the first character of a
4347 non-anchored pattern, and for finding a character that is required later in the
4348 match were failing in some cases. This didn't break the matching; it just
4349 failed to optimize when it could. The way this is done has been re-implemented.
4350
4351 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4352
4353 28. Added a new feature that provides some of the functionality that Perl
4354 provides with (?{...}). The facility is termed a "callout". The way it is done
4355 in PCRE is for the caller to provide an optional function, by setting
4356 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4357 global variable. By default it is unset, which disables all calling out. To get
4358 the function called, the regex must include (?C) at appropriate points. This
4359 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4360 This provides a means of identifying different callout points. When PCRE
4361 reaches such a point in the regex, if pcre_callout has been set, the external
4362 function is called. It is provided with data in a structure called
4363 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4364 matching continues; if it returns a non-zero value, the match at the current
4365 point fails. However, backtracking will occur if possible. [This was changed
4366 later and other features added - see item 49 below.]
4367
4368 29. pcretest is upgraded to test the callout functionality. It provides a
4369 callout function that displays information. By default, it shows the start of
4370 the match and the current position in the text. There are some new data escapes
4371 to vary what happens:
4372
4373 \C+ in addition, show current contents of captured substrings
4374 \C- do not supply a callout function
4375 \C!n return 1 when callout number n is reached
4376 \C!n!m return 1 when callout number n is reached for the mth time
4377
4378 30. If pcregrep was called with the -l option and just a single file name, it
4379 output "<stdin>" if a match was found, instead of the file name.
4380
4381 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4382 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4383 pcre_exec(). This saves a malloc/free per call. The default value of
4384 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4385 when configuring.
4386
4387 32. The default maximum size of a compiled pattern is 64K. There have been a
4388 few cases of people hitting this limit. The code now uses macros to handle the
4389 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4390 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4391 configuring. Tests 2 and 5 work only with 2-byte links because they output
4392 debugging information about compiled patterns.
4393
4394 33. Internal code re-arrangements:
4395
4396 (a) Moved the debugging function for printing out a compiled regex into
4397 its own source file (printint.c) and used #include to pull it into
4398 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4399 separate copies.
4400
4401 (b) Defined the list of op-code names for debugging as a macro in
4402 internal.h so that it is next to the definition of the opcodes.
4403
4404 (c) Defined a table of op-code lengths for simpler skipping along compiled
4405 code. This is again a macro in internal.h so that it is next to the
4406 definition of the opcodes.
4407
4408 34. Added support for recursive calls to individual subpatterns, along the
4409 lines of Robin Houston's patch (but implemented somewhat differently).
4410
4411 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4412 allow it to read and process whole directories in Win32. This code was
4413 contributed by Lionel Fourquaux; it has not been tested by me.
4414
4415 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4416 used to name a group. Names consist of alphanumerics and underscores, and must
4417 be unique. Back references use the syntax (?P=name) and recursive calls use
4418 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4419 numbers. The function pcre_fullinfo() can be used after compilation to extract
4420 a name/number map. There are three relevant calls:
4421
4422 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4423 PCRE_INFO_NAMECOUNT yields the number of entries
4424 PCRE_INFO_NAMETABLE yields a pointer to the map.
4425
4426 The map is a vector of fixed-size entries. The size of each entry depends on
4427 the length of the longest name used. The first two bytes of each entry are the
4428 group number, most significant byte first. There follows the corresponding
4429 name, zero terminated. The names are in alphabetical order.
4430
4431 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4432 case instead of 255. Making it the same both with and without UTF-8 support
4433 means that the same test output works with both.
4434
4435 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4436 calling malloc() with a zero argument.
4437
4438 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4439 optimization. I've improved things by keeping a bitmap of backreferences with
4440 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4441 fact referenced, the optimization can be applied. It is unlikely that a
4442 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4443 the match to follow \n) will appear inside brackets with a number greater than
4444 31, but if it does, any back reference > 31 suppresses the optimization.
4445
4446 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4447 of disabling numbered capturing parentheses. Any opening parenthesis that is
4448 not followed by ? behaves as if it were followed by ?: but named parentheses
4449 can still be used for capturing (and they will acquire numbers in the usual
4450 way).
4451
4452 41. Redesigned the return codes from the match() function into yes/no/error so
4453 that errors can be passed back from deep inside the nested calls. A malloc
4454 failure while inside a recursive subpattern call now causes the
4455 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4456
4457 42. It is now possible to set a limit on the number of times the match()
4458 function is called in a call to pcre_exec(). This facility makes it possible to
4459 limit the amount of recursion and backtracking, though not in a directly
4460 obvious way, because the match() function is used in a number of different
4461 circumstances. The count starts from zero for each position in the subject
4462 string (for non-anchored patterns). The default limit is, for compatibility, a
4463 large number, namely 10 000 000. You can change this in two ways:
4464
4465 (a) When configuring PCRE before making, you can use --with-match-limit=n
4466 to set a default value for the compiled library.
4467
4468 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4469 a different value is set. See 45 below.
4470
4471 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4472
4473 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4474 of things that can be changed at compile time. The first argument specifies
4475 what is wanted and the second points to where the information is to be placed.
4476 The current list of available information is:
4477
4478 PCRE_CONFIG_UTF8
4479
4480 The output is an integer that is set to one if UTF-8 support is available;
4481 otherwise it is set to zero.
4482
4483 PCRE_CONFIG_NEWLINE
4484
4485 The output is an integer that it set to the value of the code that is used for
4486 newline. It is either LF (10) or CR (13).
4487
4488 PCRE_CONFIG_LINK_SIZE
4489
4490 The output is an integer that contains the number of bytes used for internal
4491 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4492
4493 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4494
4495 The output is an integer that contains the threshold above which the POSIX
4496 interface uses malloc() for output vectors. See item 31 above.
4497
4498 PCRE_CONFIG_MATCH_LIMIT
4499
4500 The output is an unsigned integer that contains the default limit of the number
4501 of match() calls in a pcre_exec() execution. See 42 above.
4502
4503 44. pcretest has been upgraded by the addition of the -C option. This causes it
4504 to extract all the available output from the new pcre_config() function, and to
4505 output it. The program then exits immediately.
4506
4507 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4508 order to support additional features. One way would have been to define
4509 pcre_exec2() (for example) with extra arguments, but this would not have been
4510 extensible, and would also have required all calls to the original function to
4511 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4512 is used for passing in "extra" data from pcre_study().
4513
4514 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4515 contains the following fields:
4516
4517 flags a bitmap indicating which of the following fields are set
4518 study_data opaque data from pcre_study()
4519 match_limit a way of specifying a limit on match() calls for a specific
4520 call to pcre_exec()
4521 callout_data data for callouts (see 49 below)
4522
4523 The flag bits are also defined in pcre.h, and are
4524
4525 PCRE_EXTRA_STUDY_DATA
4526 PCRE_EXTRA_MATCH_LIMIT
4527 PCRE_EXTRA_CALLOUT_DATA
4528
4529 The pcre_study() function now returns one of these new pcre_extra blocks, with
4530 the actual study data pointed to by the study_data field, and the
4531 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4532 before. That is, this change is entirely upwards-compatible and requires no
4533 change to existing code.
4534
4535 If you want to pass in additional data to pcre_exec(), you can either place it
4536 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4537 block.
4538
4539 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4540 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4541 times with different match limits, until it finds the minimum value needed for
4542 pcre_exec() to complete. The value is then output. This can be instructive; for
4543 most simple matches the number is quite small, but for pathological cases it
4544 gets very large very quickly.
4545
4546 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4547 returns the size of the data block pointed to by the study_data field in a
4548 pcre_extra block, that is, the value that was passed as the argument to
4549 pcre_malloc() when PCRE was getting memory in which to place the information
4550 created by pcre_study(). The fourth argument should point to a size_t variable.
4551 pcretest has been extended so that this information is shown after a successful
4552 pcre_study() call when information about the compiled regex is being displayed.
4553
4554 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4555 because what follows is always an absolute path. (Later: it turns out that this
4556 is more than cosmetic for MinGW, because it doesn't like empty path
4557 components.)
4558
4559 49. Some changes have been made to the callout feature (see 28 above):
4560
4561 (i) A callout function now has three choices for what it returns:
4562
4563 0 => success, carry on matching
4564 > 0 => failure at this point, but backtrack if possible
4565 < 0 => serious error, return this value from pcre_exec()
4566
4567 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4568 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4569 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4570 use by callout functions. It will never be used by PCRE itself.
4571
4572 (ii) The pcre_extra structure (see 45 above) has a void * field called
4573 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4574 pcre_callout_block structure has a field of the same name. The contents of
4575 the field passed in the pcre_extra structure are passed to the callout
4576 function in the corresponding field in the callout block. This makes it
4577 easier to use the same callout-containing regex from multiple threads. For
4578 testing, the pcretest program has a new data escape
4579
4580 \C*n pass the number n (may be negative) as callout_data
4581
4582 If the callout function in pcretest receives a non-zero value as
4583 callout_data, it returns that value.
4584
4585 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4586 there were some redundant $(CFLAGS) in commands that are now specified as
4587 $(LINK), which already includes $(CFLAGS).
4588
4589 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4590 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4591 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4592 one-byte characters throughout. Note that case-insensitive matching applies
4593 only to characters whose values are less than 256. PCRE doesn't support the
4594 notion of cases for higher-valued characters.
4595
4596 (i) A character class whose characters are all within 0-255 is handled as
4597 a bit map, and the map is inverted for negative classes. Previously, a
4598 character > 255 always failed to match such a class; however it should
4599 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4600
4601 (ii) A negated character class with a single character < 255 is coded as
4602 "not this character" (OP_NOT). This wasn't working properly when the test
4603 character was multibyte, either singly or repeated.
4604
4605 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4606 mode, for example: \x{100}{2,3}.
4607
4608 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4609 singly or repeated) now correctly test multibyte characters. However,
4610 PCRE doesn't recognize any characters with values greater than 255 as
4611 digits, spaces, or word characters. Such characters always match \D, \S,
4612 and \W, and never match \d, \s, or \w.
4613
4614 (v) Classes may now contain characters and character ranges with values
4615 greater than 255. For example: [ab\x{100}-\x{400}].
4616
4617 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4618 PCRE in UTF-8 mode.
4619
4620 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4621 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4622 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4623 value.)
4624
4625 53. The single man page has become too large. I have therefore split it up into
4626 a number of separate man pages. These also give rise to individual HTML pages;
4627 these are now put in a separate directory, and there is an index.html page that
4628 lists them all. Some hyperlinking between the pages has been installed.
4629
4630 54. Added convenience functions for handling named capturing parentheses.
4631
4632 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4633 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4634 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4635 are faulted.
4636
4637 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4638 calling configure. These values are used when compiling the dftables.c program
4639 which is run to generate the source of the default character tables. They
4640 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4641 you will need to set these values.
4642
4643 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4644
4645
4646 Version 3.9 02-Jan-02
4647 ---------------------
4648
4649 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4650
4651 2. If --disable-static was given, the building process failed when trying to
4652 build pcretest and pcregrep. (For some reason it was using libtool to compile
4653 them, which is not right, as they aren't part of the library.)
4654
4655
4656 Version 3.8 18-Dec-01
4657 ---------------------
4658
4659 1. The experimental UTF-8 code was completely screwed up. It was packing the
4660 bytes in the wrong order. How dumb can you get?
4661
4662
4663 Version 3.7 29-Oct-01
4664 ---------------------
4665
4666 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4667 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4668 this didn't happen under Solaris 8, where I normally test things.
4669
4670 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4671 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4672 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4673 this fix an hour or so after the initial 3.7 release.)
4674
4675
4676 Version 3.6 23-Oct-01
4677 ---------------------
4678
4679 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4680 offsets passed as NULL with zero offset count.
4681
4682 2. The config.guess and config.sub files had not been updated when I moved to
4683 the latest autoconf.
4684
4685
4686 Version 3.5 15-Aug-01
4687 ---------------------
4688
4689 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4690 had been forgotten.
4691
4692 2. By using declared but undefined structures, we can avoid using "void"
4693 definitions in pcre.h while keeping the internal definitions of the structures
4694 private.
4695
4696 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4697 user point of view, this means that both static and shared libraries are built
4698 by default, but this can be individually controlled. More of the work of
4699 handling this static/shared cases is now inside libtool instead of PCRE's make
4700 file.
4701
4702 4. The pcretest utility is now installed along with pcregrep because it is
4703 useful for users (to test regexs) and by doing this, it automatically gets
4704 relinked by libtool. The documentation has been turned into a man page, so
4705 there are now .1, .txt, and .html versions in /doc.
4706
4707 5. Upgrades to pcregrep:
4708 (i) Added long-form option names like gnu grep.
4709 (ii) Added --help to list all options with an explanatory phrase.
4710 (iii) Added -r, --recursive to recurse into sub-directories.
4711 (iv) Added -f, --file to read patterns from a file.
4712
4713 6. pcre_exec() was referring to its "code" argument before testing that
4714 argument for NULL (and giving an error if it was NULL).
4715
4716 7. Upgraded Makefile.in to allow for compiling in a different directory from
4717 the source directory.
4718
4719 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4720 options bits, the pointer it was passed was to an int instead of to an unsigned
4721 long int. This mattered only on 64-bit systems.
4722
4723 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4724 generated) instead of pcre.in, which it its source. Also made the same change
4725 in several of the .c files.
4726
4727 10. A new release of gcc defines printf() as a macro, which broke pcretest
4728 because it had an ifdef in the middle of a string argument for printf(). Fixed
4729 by using separate calls to printf().
4730
4731 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4732 script, to force use of CR or LF instead of \n in the source. On non-Unix
4733 systems, the value can be set in config.h.
4734
4735 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4736 absolute limit. Changed the text of the error message to make this clear, and
4737 likewise updated the man page.
4738
4739 13. The limit of 99 on the number of capturing subpatterns has been removed.
4740 The new limit is 65535, which I hope will not be a "real" limit.
4741
4742
4743 Version 3.4 22-Aug-00
4744 ---------------------
4745
4746 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4747
4748 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4749
4750
4751 Version 3.3 01-Aug-00
4752 ---------------------
4753
4754 1. If an octal character was given, but the value was greater than \377, it
4755 was not getting masked to the least significant bits, as documented. This could
4756 lead to crashes in some systems.
4757
4758 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4759 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4760
4761 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4762 These just pass their arguments on to (pcre_free)(), but they are provided
4763 because some uses of PCRE bind it to non-C systems that can call its functions,
4764 but cannot call free() or pcre_free() directly.
4765
4766 4. Add "make test" as a synonym for "make check". Corrected some comments in
4767 the Makefile.
4768
4769 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4770 Makefile.
4771
4772 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4773 command called pgrep for grepping around the active processes.
4774
4775 7. Added the beginnings of support for UTF-8 character strings.
4776
4777 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4778 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4779 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4780 out for the ar command.)
4781
4782
4783 Version 3.2 12-May-00
4784 ---------------------
4785
4786 This is purely a bug fixing release.
4787
4788 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4789 of ZA. This was just one example of several cases that could provoke this bug,
4790 which was introduced by change 9 of version 2.00. The code for breaking
4791 infinite loops after an iteration that matches an empty string was't working
4792 correctly.
4793
4794 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4795 when matched against abbab (for example). After matching an empty string, it
4796 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4797 caused it to match further down the string than it should.
4798
4799 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4800 was there because it doesn't seem to be needed, and it causes trouble on some
4801 systems, as it is not a Standard C header. It has been removed.
4802
4803 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4804 were reported on the Macintosh. The changes were from
4805
4806 while ((c = *(++ptr)) != 0 && c != '\n');
4807 to
4808 while ((c = *(++ptr)) != 0 && c != '\n') ;
4809
4810 Totally extraordinary, but if that's what it takes...
4811
4812 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4813 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4814 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4815 assumes the way PCRE uses memmove() (always moving upwards).
4816
4817 6. PCRE is being used in one environment where strchr() is not available. There
4818 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4819 faster code anyway.
4820
4821
4822 Version 3.1 09-Feb-00
4823 ---------------------
4824
4825 The only change in this release is the fixing of some bugs in Makefile.in for
4826 the "install" target:
4827
4828 (1) It was failing to install pcreposix.h.
4829
4830 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4831
4832
4833 Version 3.0 01-Feb-00
4834 ---------------------
4835
4836 1. Add support for the /+ modifier to perltest (to output $` like it does in
4837 pcretest).
4838
4839 2. Add support for the /g modifier to perltest.
4840
4841 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4842 matches null strings.
4843
4844 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4845 pattern. Perl treats empty patterns specially - it reuses the most recent
4846 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4847 effect.
4848
4849 5. The POSIX interface was broken in that it was just handing over the POSIX
4850 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4851 required a bigger vector, with some working space on the end. This means that
4852 the POSIX wrapper now has to get and free some memory, and copy the results.
4853
4854 6. Added some simple autoconf support, placing the test data and the
4855 documentation in separate directories, re-organizing some of the
4856 information files, and making it build pcre-config (a GNU standard). Also added
4857 libtool support for building PCRE as a shared library, which is now the
4858 default.
4859
4860 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4861 09 are not valid octal constants. Single digits will be used for minor values
4862 less than 10.
4863
4864 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4865 existing programs that set these in the POSIX interface can use PCRE without
4866 modification.
4867
4868 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4869 return all that pcre_info() returns, plus additional data. The pcre_info()
4870 function is retained for compatibility, but is considered to be obsolete.
4871
4872 10. Added experimental recursion feature (?R) to handle one common case that
4873 Perl 5.6 will be able to do with (?p{...}).
4874
4875 11. Added support for POSIX character classes like [:alpha:], which Perl is
4876 adopting.
4877
4878
4879 Version 2.08 31-Aug-99
4880 ----------------------
4881
4882 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4883 trying to match at the startoffset position, but instead was moving forward to
4884 the next newline as if a previous match had failed.
4885
4886 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4887 and could get into a loop if a null string was matched other than at the start
4888 of the subject.
4889
4890 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4891 be distinguished at compile time, and for completeness also added PCRE_DATE.
4892
4893 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4894 in GnuWin32 environments.
4895
4896
4897 Version 2.07 29-Jul-99
4898 ----------------------
4899
4900 1. The documentation is now supplied in plain text form and HTML as well as in
4901 the form of man page sources.
4902
4903 2. C++ compilers don't like assigning (void *) values to other pointer types.
4904 In particular this affects malloc(). Although there is no problem in Standard
4905 C, I've put in casts to keep C++ compilers happy.
4906
4907 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4908 should be (const char *).
4909
4910 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4911 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4912 However, I haven't made this a standard facility. The documentation doesn't
4913 mention it, and the Makefile doesn't support it.
4914
4915 5. The Makefile now contains an "install" target, with editable destinations at
4916 the top of the file. The pcretest program is not installed.
4917
4918 6. pgrep -V now gives the PCRE version number and date.
4919
4920 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4921 causing the entire string to be ignored, instead of just the last character.
4922
4923 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4924 non-matching string, it can take a very, very long time, even for strings of
4925 quite modest length, because of the nested recursion. PCRE now does better in
4926 some of these cases. It does this by remembering the last required literal
4927 character in the pattern, and pre-searching the subject to ensure it is present
4928 before running the real match. In other words, it applies a heuristic to detect
4929 some types of certain failure quickly, and in the above example, if presented
4930 with a string that has no trailing " it gives "no match" very quickly.
4931
4932 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4933 other alternatives are tried instead.
4934
4935
4936 Version 2.06 09-Jun-99
4937 ----------------------
4938
4939 1. Change pcretest's output for amount of store used to show just the code
4940 space, because the remainder (the data block) varies in size between 32-bit and
4941 64-bit systems.
4942
4943 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4944 start matching at. This allows lookbehinds to work when searching for multiple
4945 occurrences in a string.
4946
4947 3. Added additional options to pcretest for testing multiple occurrences:
4948
4949 /+ outputs the rest of the string that follows a match
4950 /g loops for multiple occurrences, using the new startoffset argument
4951 /G loops for multiple occurrences by passing an incremented pointer
4952
4953 4. PCRE wasn't doing the "first character" optimization for patterns starting
4954 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4955 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4956 the letter 'x'. On long subject strings, this gives a significant speed-up.
4957
4958
4959 Version 2.05 21-Apr-99
4960 ----------------------
4961
4962 1. Changed the type of magic_number from int to long int so that it works
4963 properly on 16-bit systems.
4964
4965 2. Fixed a bug which caused patterns starting with .* not to work correctly
4966 when the subject string contained newline characters. PCRE was assuming
4967 anchoring for such patterns in all cases, which is not correct because .* will
4968 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4969 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4970 must be retried after every newline in the subject.
4971
4972
4973 Version 2.04 18-Feb-99
4974 ----------------------
4975
4976 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4977 computation of the store needed to hold the pattern was incorrect (too large).
4978 If such patterns were nested a few deep, this could multiply and become a real
4979 problem.
4980
4981 2. Added /M option to pcretest to show the memory requirement of a specific
4982 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4983
4984 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4985 compiled in such a way that the backtracking after subsequent failure was
4986 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4987 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4988
4989
4990 Version 2.03 02-Feb-99
4991 ----------------------
4992
4993 1. Fixed typo and small mistake in man page.
4994
4995 2. Added 4th condition (GPL supersedes if conflict) and created separate
4996 LICENCE file containing the conditions.
4997
4998 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4999 Perl, that is the internal \ allows the delimiter to be included in the
5000 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5001 the final delimiter, add \ to the end of the pattern (to test the error).
5002
5003 4. Added the convenience functions for extracting substrings after a successful
5004 match. Updated pcretest to make it able to test these functions.
5005
5006
5007 Version 2.02 14-Jan-99
5008 ----------------------
5009
5010 1. Initialized the working variables associated with each extraction so that
5011 their saving and restoring doesn't refer to uninitialized store.
5012
5013 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5014 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5015 fix the problem.
5016
5017 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5018 calls, and wasn't printing the correct value for compiling calls. Increased the
5019 default value of LOOPREPEAT, and the number of significant figures in the
5020 times.
5021
5022 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5023
5024 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5025 a building problem on Windows NT with a FAT file system.
5026
5027
5028 Version 2.01 21-Oct-98
5029 ----------------------
5030
5031 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5032 to character tables built by pcre_maketables() in the current locale. If NULL
5033 is passed, the default tables are used.
5034
5035
5036 Version 2.00 24-Sep-98
5037 ----------------------
5038
5039 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5040 it any more.
5041
5042 2. Allow quantification of (?>) groups, and make it work correctly.
5043
5044 3. The first character computation wasn't working for (?>) groups.
5045
5046 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5047 end of the subject) and add 5.005's \z, which really does match only at the
5048 very end of the subject.
5049
5050 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5051
5052 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5053 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5054 localized options. All options to pcre_study() were also removed.
5055
5056 7. Add other new features from 5.005:
5057
5058 $(?<= positive lookbehind
5059 $(?<! negative lookbehind
5060 (?imsx-imsx) added the unsetting capability
5061 such a setting is global if at outer level; local otherwise
5062 (?imsx-imsx:) non-capturing groups with option setting
5063 (?(cond)re|re) conditional pattern matching
5064
5065 A backreference to itself in a repeated group matches the previous
5066 captured string.
5067
5068 8. General tidying up of studying (both automatic and via "study")
5069 consequential on the addition of new assertions.
5070
5071 9. As in 5.005, unlimited repeated groups that could match an empty substring
5072 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5073 runtime if any iteration does actually match an empty substring.
5074
5075 10. Include the RunTest script in the distribution.
5076
5077 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5078 discrepancies, some of which were old and were also with respect to 5.004. They
5079 have now been fixed.
5080
5081
5082 Version 1.09 28-Apr-98
5083 ----------------------
5084
5085 1. A negated single character class followed by a quantifier with a minimum
5086 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5087 program crashes, or just wrong answers. This did not apply to negated classes
5088 containing more than one character, or to minima other than one.
5089
5090
5091 Version 1.08 27-Mar-98
5092 ----------------------
5093
5094 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5095
5096 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5097 latter must appear before anything that relies on it in the pattern.
5098
5099
5100 Version 1.07 16-Feb-98
5101 ----------------------
5102
5103 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5104 repeat of a potentially empty string).
5105
5106
5107 Version 1.06 23-Jan-98
5108 ----------------------
5109
5110 1. Added Markus Oberhumer's little patches for C++.
5111
5112 2. Literal strings longer than 255 characters were broken.
5113
5114
5115 Version 1.05 23-Dec-97
5116 ----------------------
5117
5118 1. Negated character classes containing more than one character were failing if
5119 PCRE_CASELESS was set at run time.
5120
5121
5122 Version 1.04 19-Dec-97
5123 ----------------------
5124
5125 1. Corrected the man page, where some "const" qualifiers had been omitted.
5126
5127 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5128 input syntax.
5129
5130 3. Fixed memory leak which occurred when a regex with back references was
5131 matched with an offsets vector that wasn't big enough. The temporary memory
5132 that is used in this case wasn't being freed if the match failed.
5133
5134 4. Tidied pcretest to ensure it frees memory that it gets.
5135
5136 5. Temporary memory was being obtained in the case where the passed offsets
5137 vector was exactly big enough.
5138
5139 6. Corrected definition of offsetof() from change 5 below.
5140
5141 7. I had screwed up change 6 below and broken the rules for the use of
5142 setjmp(). Now fixed.
5143
5144
5145 Version 1.03 18-Dec-97
5146 ----------------------
5147
5148 1. A erroneous regex with a missing opening parenthesis was correctly
5149 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5150 on some systems.
5151
5152 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5153 it was reported that one broken compiler failed on the former because "code" is
5154 also an independent variable.
5155
5156 3. The erroneous regex a[]b caused an array overrun reference.
5157
5158 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5159 fail on data ending with that character. (It was going on too far, and checking
5160 the next character, typically a binary zero.) This was specific to the
5161 optimized code for single-character negative classes.
5162
5163 5. Added a contributed patch from the TIN world which does the following:
5164
5165 + Add an undef for memmove, in case the the system defines a macro for it.
5166
5167 + Add a definition of offsetof(), in case there isn't one. (I don't know
5168 the reason behind this - offsetof() is part of the ANSI standard - but
5169 it does no harm).
5170
5171 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5172 most of the places where whitespace preceded '#'. I have given up and
5173 allowed the remaining 2 cases to be at the margin.
5174
5175 + Rename some variables in pcre to eliminate shadowing. This seems very
5176 pedantic, but does no harm, of course.
5177
5178 6. Moved the call to setjmp() into its own function, to get rid of warnings
5179 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5180
5181 7. Constructs such as \d{8,} were compiling into the equivalent of
5182 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5183 outcome, but in this particular case used more store than had been allocated,
5184 which caused the bug to be discovered because it threw up an internal error.
5185
5186 8. The debugging code in both pcre and pcretest for outputting the compiled
5187 form of a regex was going wrong in the case of back references followed by
5188 curly-bracketed repeats.
5189
5190
5191 Version 1.02 12-Dec-97
5192 ----------------------
5193
5194 1. Typos in pcre.3 and comments in the source fixed.
5195
5196 2. Applied a contributed patch to get rid of places where it used to remove
5197 'const' from variables, and fixed some signed/unsigned and uninitialized
5198 variable warnings.
5199
5200 3. Added the "runtest" target to Makefile.
5201
5202 4. Set default compiler flag to -O2 rather than just -O.
5203
5204
5205 Version 1.01 19-Nov-97
5206 ----------------------
5207
5208 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5209 like /([ab]*)*/, that is, for classes with more than one character in them.
5210
5211 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5212 as /((?>a*))*/ (a PCRE_EXTRA facility).
5213
5214
5215 Version 1.00 18-Nov-97
5216 ----------------------
5217
5218 1. Added compile-time macros to support systems such as SunOS4 which don't have
5219 memmove() or strerror() but have other things that can be used instead.
5220
5221 2. Arranged that "make clean" removes the executables.
5222
5223
5224 Version 0.99 27-Oct-97
5225 ----------------------
5226
5227 1. Fixed bug in code for optimizing classes with only one character. It was
5228 initializing a 32-byte map regardless, which could cause it to run off the end
5229 of the memory it had got.
5230
5231 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5232
5233
5234 Version 0.98 22-Oct-97
5235 ----------------------
5236
5237 1. Fixed bug in code for handling temporary memory usage when there are more
5238 back references than supplied space in the ovector. This could cause segfaults.
5239
5240
5241 Version 0.97 21-Oct-97
5242 ----------------------
5243
5244 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5245
5246 2. Optimized negated single characters not to use a bit map.
5247
5248 3. Brought error texts together as macro definitions; clarified some of them;
5249 fixed one that was wrong - it said "range out of order" when it meant "invalid
5250 escape sequence".
5251
5252 4. Changed some char * arguments to const char *.
5253
5254 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5255
5256 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5257 pcretest.
5258
5259
5260 Version 0.96 16-Oct-97
5261 ----------------------
5262
5263 1. Added a simple "pgrep" utility to the distribution.
5264
5265 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5266 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5267 where "ddd" means "one or more decimal digits".
5268
5269 3. Fixed serious bug. If a pattern had a back reference, but the call to
5270 pcre_exec() didn't supply a large enough ovector to record the related
5271 identifying subpattern, the match always failed. PCRE now remembers the number
5272 of the largest back reference, and gets some temporary memory in which to save
5273 the offsets during matching if necessary, in order to ensure that
5274 backreferences always work.
5275
5276 4. Increased the compatibility with Perl in a number of ways:
5277
5278 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5279 to request this handling. The option can be set at compile or exec time.
5280
5281 (b) $ matches before a terminating newline by default; an option
5282 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5283 mode). The option can be set at compile or exec time.
5284
5285 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5286 the same as Perl's. If the decimal number it represents is less than 10
5287 or there aren't that many previous left capturing parentheses, an octal
5288 escape is read. Inside a character class, it's always an octal escape,
5289 even if it is a single digit.
5290
5291 (d) An escaped but undefined alphabetic character is taken as a literal,
5292 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5293 escapes.
5294
5295 (e) {0} is now permitted. (The previous item is removed from the compiled
5296 pattern).
5297
5298 5. Changed all the names of code files so that the basic parts are no longer
5299 than 10 characters, and abolished the teeny "globals.c" file.
5300
5301 6. Changed the handling of character classes; they are now done with a 32-byte
5302 bit map always.
5303
5304 7. Added the -d and /D options to pcretest to make it possible to look at the
5305 internals of compilation without having to recompile pcre.
5306
5307
5308 Version 0.95 23-Sep-97
5309 ----------------------
5310
5311 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5312 \x20 at the start of a run of normal characters. These were being treated as
5313 real characters, instead of the source characters being re-checked.
5314
5315
5316 Version 0.94 18-Sep-97
5317 ----------------------
5318
5319 1. The functions are now thread-safe, with the caveat that the global variables
5320 containing pointers to malloc() and free() or alternative functions are the
5321 same for all threads.
5322
5323 2. Get pcre_study() to generate a bitmap of initial characters for non-
5324 anchored patterns when this is possible, and use it if passed to pcre_exec().
5325
5326
5327 Version 0.93 15-Sep-97
5328 ----------------------
5329
5330 1. /(b)|(:+)/ was computing an incorrect first character.
5331
5332 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5333 but not actually doing anything yet.
5334
5335 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5336 as Perl does (e.g. [-az] or [az-]).
5337
5338 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5339 all possible positions.
5340
5341 5. Split up into different modules to avoid including unneeded functions in a
5342 compiled binary. However, compile and exec are still in one module. The "study"
5343 function is split off.
5344
5345 6. The character tables are now in a separate module whose source is generated
5346 by an auxiliary program - but can then be edited by hand if required. There are
5347 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5348 toupper() in the code.
5349
5350 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5351 make them global. Abolish the function for setting them, as the caller can now
5352 set them directly.
5353
5354
5355 Version 0.92 11-Sep-97
5356 ----------------------
5357
5358 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5359 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5360
5361 2. Caseless matching was not working in character classes if the characters in
5362 the pattern were in upper case.
5363
5364 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5365
5366 4. Make PCRE_ANCHORED public and accept as a compile option.
5367
5368 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5369 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5370 pass them.
5371
5372 6. Give an error if bad option bits passed at compile or run time.
5373
5374 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5375 pcretest to cause it to pass that flag.
5376
5377 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5378 options, and the first character, if set.
5379
5380 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5381
5382
5383 Version 0.91 10-Sep-97
5384 ----------------------
5385
5386 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5387 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5388
5389 2. PCRE was looping on encountering an indefinitely repeated back reference to
5390 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5391 Perl does - treats the match as successful.
5392
5393 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5