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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1369 - (show annotations)
Tue Oct 8 15:06:46 2013 UTC (6 years, 1 month ago) by ph10
File size: 253317 byte(s)
Update \8 and \9 handling to match most recent Perl.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.34 xx-xxxx-201x
5 --------------------------
6
7 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
8 executable memory. Patch inspired by Carsten Klein.
9
10 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
11 this macro is never tested and has no effect, because the work to support
12 coverage involves only compiling and linking options and special targets in
13 the Makefile. The comment in config.h implied that defining the macro would
14 enable coverage support, which is totally false. There was also support for
15 setting this macro in the CMake files (my fault, I just copied it from
16 configure). SUPPORT_GCOV has now been removed.
17
18 3. Make a small performance improvement in strlen16() and strlen32() in
19 pcretest.
20
21 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
22 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
23 these cases). There was also one in pcretest.c.
24
25 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
26
27 6. In UTF mode, the code for checking whether a group could match an empty
28 string (which is used for indefinitely repeated groups to allow for
29 breaking an infinite loop) was broken when the group contained a repeated
30 negated single-character class with a character that occupied more than one
31 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
32 UTF-8 mode). The effect was undefined: the group might or might not be
33 deemed as matching an empty string, or the program might have crashed.
34
35 7. The code for checking whether a group could match an empty string was not
36 recognizing that \h, \H, \v, \V, and \R must match a character.
37
38 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
39 an empty string. If it can, pcretest shows this in its information output.
40
41 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
42 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
43 matched by pcre_exec() without using JIT:
44
45 (a) If the rest of the pattern did not match after a maximal run of
46 grapheme clusters, the code for backing up to try with fewer of them
47 did not always back up over a full grapheme when characters that do not
48 have the modifier quality were involved, e.g. Hangul syllables.
49
50 (b) If the match point in a subject started with modifier character, and
51 there was no match, the code could incorrectly back up beyond the match
52 point, and potentially beyond the first character in the subject,
53 leading to a segfault or an incorrect match result.
54
55 10. A conditional group with an assertion condition could lead to PCRE
56 recording an incorrect first data item for a match if no other first data
57 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
58 first data item, and therefore matched "ca" after "c" instead of at the
59 start.
60
61 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
62 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
63
64 12. The source of pcregrep now includes z/OS-specific code so that it can be
65 compiled for z/OS as part of the special z/OS distribution.
66
67 13. Added the -T and -TM options to pcretest.
68
69 14. The code in pcre_compile.c for creating the table of named capturing groups
70 has been refactored. Instead of creating the table dynamically during the
71 actual compiling pass, the information is remembered during the pre-compile
72 pass (on the stack unless there are more than 20 named groups, in which
73 case malloc() is used) and the whole table is created before the actual
74 compile happens. This has simplified the code (it is now nearly 150 lines
75 shorter) and prepared the way for better handling of references to groups
76 with duplicate names.
77
78 15. A back reference to a named subpattern when there is more than one of the
79 same name now checks them in the order in which they appear in the pattern.
80 The first one that is set is used for the reference. Previously only the
81 first one was inspected. This change makes PCRE more compatible with Perl.
82
83 16. Unicode character properties were updated from Unicode 6.3.0.
84
85 17. The compile-time code for auto-possessification has been refactored, based
86 on a patch by Zoltan Herczeg. It now happens after instead of during
87 compilation. The code is cleaner, and more cases are handled. The option
88 PCRE_NO_AUTO_POSSESSIFY is added for testing purposes, and the -O and /O
89 options in pcretest are provided to set it.
90
91 18. The character VT has been added to the set of characters that match \s and
92 are generally treated as white space, following this same change in Perl
93 5.18. There is now no difference between "Perl space" and "POSIX space".
94
95 19. The code for checking named groups as conditions, either for being set or
96 for being recursed, has been refactored (this is related to 14 and 15
97 above). Processing unduplicated named groups should now be as fast at
98 numerical groups, and processing duplicated groups should be faster than
99 before.
100
101 20. Two patches to the CMake build system, by Alexander Barkov:
102
103 (1) Replace the "source" command by "." in CMakeLists.txt because
104 "source" is a bash-ism.
105
106 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
107 without these the CMake build does not work on Solaris.
108
109 21. Perl has changed its handling of \8 and \9. If there is no previously
110 encountered capturing group of those numbers, they are treated as the
111 literal characters 8 and 9 instead of a binary zero followed by the
112 literals. PCRE now does the same.
113
114
115 Version 8.33 28-May-2013
116 ------------------------
117
118 1. Added 'U' to some constants that are compared to unsigned integers, to
119 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
120 variables that are added to signed variables, to ensure the result is
121 signed and can be negated.
122
123 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
124 CMake config files.
125
126 3. Revise the creation of config.h.generic so that all boolean macros are
127 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
128 overriding via -D on the command line possible.
129
130 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
131 to unsigned int is reported to make a quite noticeable speed difference in
132 a specific Windows environment. Testing on Linux did also appear to show
133 some benefit (and it is clearly not harmful). Also fixed the definition of
134 Xop which should be unsigned.
135
136 5. Related to (4), changing the definition of the intermediate variable cc
137 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
138 improvements.
139
140 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
141 unnecessary spaces.
142
143 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
144 and later.
145
146 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
147
148 9. Optimizing fast_forward_start_bits in JIT.
149
150 10. Adding support for callouts in JIT, and fixing some issues revealed
151 during this work. Namely:
152
153 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
154
155 (b) Minimum length was not checked before the matching is started.
156
157 11. The value of capture_last that is passed to callouts was incorrect in some
158 cases when there was a capture on one path that was subsequently abandoned
159 after a backtrack. Also, the capture_last value is now reset after a
160 recursion, since all captures are also reset in this case.
161
162 12. The interpreter no longer returns the "too many substrings" error in the
163 case when an overflowing capture is in a branch that is subsequently
164 abandoned after a backtrack.
165
166 13. In the pathological case when an offset vector of size 2 is used, pcretest
167 now prints out the matched string after a yield of 0 or 1.
168
169 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
170 Only supported by the JIT compiler at the moment.
171
172 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
173
174 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
175 offset of the starting point of the matching process, provided the offsets
176 vector is large enough.
177
178 17. The \A escape now records a lookbehind value of 1, though its execution
179 does not actually inspect the previous character. This is to ensure that,
180 in partial multi-segment matching, at least one character from the old
181 segment is retained when a new segment is processed. Otherwise, if there
182 are no lookbehinds in the pattern, \A might match incorrectly at the start
183 of a new segment.
184
185 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
186
187 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
188 gives some modest performance improvement in 8-bit mode.
189
190 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
191 be expressed in certain programming languages using Universal Character
192 Names.
193
194 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
195 which points out that "non characters" are not "characters that may not
196 appear in Unicode strings" but rather "characters that are reserved for
197 internal use and have only local meaning".
198
199 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
200 there was a conditional group that depended on an assertion, if the
201 assertion was false, the callout that immediately followed the alternation
202 in the condition was skipped when pcre_exec() was used for matching.
203
204 23. Allow an explicit callout to be inserted before an assertion that is the
205 condition for a conditional group, for compatibility with automatic
206 callouts, which always insert a callout at this point.
207
208 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
209 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
210
211 25. (*PRUNE) is now supported by the JIT compiler.
212
213 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
214
215 27. Fix the case where there are two or more SKIPs with arguments that may be
216 ignored.
217
218 28. (*SKIP) is now supported by the JIT compiler.
219
220 29. (*THEN) is now supported by the JIT compiler.
221
222 30. Update RunTest with additional test selector options.
223
224 31. The way PCRE handles backtracking verbs has been changed in two ways.
225
226 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
227 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
228 backtracking. In some cases this makes it more Perl-compatible, but Perl's
229 rather obscure rules do not always do the same thing.
230
231 (2) Previously, backtracking verbs were confined within assertions. This is
232 no longer the case for positive assertions, except for (*ACCEPT). Again,
233 this sometimes improves Perl compatibility, and sometimes does not.
234
235 32. A number of tests that were in test 2 because Perl did things differently
236 have been moved to test 1, because either Perl or PCRE has changed, and
237 these tests are now compatible.
238
239 32. Backtracking control verbs are now handled in the same way in JIT and
240 interpreter.
241
242 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
243 contained a forward subroutine reference caused a compile error.
244
245 34. Auto-detect and optimize limited repetitions in JIT.
246
247 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
248 blocking (*UTF) etc.
249
250 36. In the interpreter, maximizing pattern repetitions for characters and
251 character types now use tail recursion, which reduces stack usage.
252
253 37. The value of the max lookbehind was not correctly preserved if a compiled
254 and saved regex was reloaded on a host of different endianness.
255
256 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
257 of the compiled pattern block, expand the flags field from 16 to 32 bits
258 because it was almost full.
259
260 39. Try madvise first before posix_madvise.
261
262 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
263 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
264 original change locked out all matches of empty strings. This has been
265 changed so that one match of an empty string per line is recognized.
266 Subsequent searches on the same line (for colouring or for --only-matching,
267 for example) do not recognize empty strings.
268
269 41. Applied a user patch to fix a number of spelling mistakes in comments.
270
271 42. Data lines longer than 65536 caused pcretest to crash.
272
273 43. Clarified the data type for length and startoffset arguments for pcre_exec
274 and pcre_dfa_exec in the function-specific man pages, where they were
275 explicitly stated to be in bytes, never having been updated. I also added
276 some clarification to the pcreapi man page.
277
278 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
279 a segmentation fault.
280
281
282 Version 8.32 30-November-2012
283 -----------------------------
284
285 1. Improved JIT compiler optimizations for first character search and single
286 character iterators.
287
288 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
289 Patch by Daniel Richard G.
290
291 3. Single character iterator optimizations in the JIT compiler.
292
293 4. Improved JIT compiler optimizations for character ranges.
294
295 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
296 Reported by Giuseppe D'Angelo.
297
298 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
299 of a line, was being set incorrectly in cases where .* appeared inside
300 atomic brackets at the start of a pattern, or where there was a subsequent
301 *PRUNE or *SKIP.
302
303 7. Improved instruction cache flush for POWER/PowerPC.
304 Patch by Daniel Richard G.
305
306 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
307 grep:
308
309 (a) There is now no limit to the number of patterns to be matched.
310
311 (b) An error is given if a pattern is too long.
312
313 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
314 are now supported.
315
316 (d) --exclude-from and --include-from (multiple use) have been added.
317
318 (e) Exclusions and inclusions now apply to all files and directories, not
319 just to those obtained from scanning a directory recursively.
320
321 (f) Multiple uses of -f and --file-list are now supported.
322
323 (g) In a Windows environment, the default for -d has been changed from
324 "read" (the GNU grep default) to "skip", because otherwise the presence
325 of a directory in the file list provokes an error.
326
327 (h) The documentation has been revised and clarified in places.
328
329 9. Improve the matching speed of capturing brackets.
330
331 10. Changed the meaning of \X so that it now matches a Unicode extended
332 grapheme cluster.
333
334 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
335 out POSIX threads when JIT support is configured.
336
337 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
338
339 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
340 structure could go wrong in environments where size_t is not the same size
341 as int.
342
343 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
344 set.
345
346 15. The EBCDIC support had decayed; later updates to the code had included
347 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
348 general tidy up of EBCDIC-related issues, and the documentation was also
349 not quite right. There is now a test that can be run on ASCII systems to
350 check some of the EBCDIC-related things (but is it not a full test).
351
352 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
353 in a small tidy to the code.
354
355 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
356
357 18. If the --only-matching (-o) option in pcregrep is specified multiple
358 times, each one causes appropriate output. For example, -o1 -o2 outputs the
359 substrings matched by the 1st and 2nd capturing parentheses. A separating
360 string can be specified by --om-separator (default empty).
361
362 19. Improving the first n character searches.
363
364 20. Turn case lists for horizontal and vertical white space into macros so that
365 they are defined only once.
366
367 21. This set of changes together give more compatible Unicode case-folding
368 behaviour for characters that have more than one other case when UCP
369 support is available.
370
371 (a) The Unicode property table now has offsets into a new table of sets of
372 three or more characters that are case-equivalent. The MultiStage2.py
373 script that generates these tables (the pcre_ucd.c file) now scans
374 CaseFolding.txt instead of UnicodeData.txt for character case
375 information.
376
377 (b) The code for adding characters or ranges of characters to a character
378 class has been abstracted into a generalized function that also handles
379 case-independence. In UTF-mode with UCP support, this uses the new data
380 to handle characters with more than one other case.
381
382 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
383 whose other case is greater than 256 are now correctly matched
384 caselessly. Previously, the high codepoint matched the low one, but not
385 vice versa.
386
387 (d) The processing of \h, \H, \v, and \ in character classes now makes use
388 of the new class addition function, using character lists defined as
389 macros alongside the case definitions of 20 above.
390
391 (e) Caseless back references now work with characters that have more than
392 one other case.
393
394 (f) General caseless matching of characters with more than one other case
395 is supported.
396
397 22. Unicode character properties were updated from Unicode 6.2.0
398
399 23. Improved CMake support under Windows. Patch by Daniel Richard G.
400
401 24. Add support for 32-bit character strings, and UTF-32
402
403 25. Major JIT compiler update (code refactoring and bugfixing).
404 Experimental Sparc 32 support is added.
405
406 26. Applied a modified version of Daniel Richard G's patch to create
407 pcre.h.generic and config.h.generic by "make" instead of in the
408 PrepareRelease script.
409
410 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
411 pcre_compile.c when checking for a zero character.
412
413 28. Introducing a native interface for JIT. Through this interface, the compiled
414 machine code can be directly executed. The purpose of this interface is to
415 provide fast pattern matching, so several sanity checks are not performed.
416 However, feature tests are still performed. The new interface provides
417 1.4x speedup compared to the old one.
418
419 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
420 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
421 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
422
423 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
424 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
425 the "old" RFC 2279). Instead, it ended up passing a negative length to
426 pcre_exec().
427
428 31. Add support for GCC's visibility feature to hide internal functions.
429
430 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
431 "unknown -C option" after outputting 0 or 1.
432
433 33. There is now support for generating a code coverage report for the test
434 suite in environments where gcc is the compiler and lcov is installed. This
435 is mainly for the benefit of the developers.
436
437 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
438 unaddressable using valgrind annotations, allowing valgrind to detect
439 invalid memory accesses. This is mainly for the benefit of the developers.
440
441 25. (*UTF) can now be used to start a pattern in any of the three libraries.
442
443 26. Give configure error if --enable-cpp but no C++ compiler found.
444
445
446 Version 8.31 06-July-2012
447 -------------------------
448
449 1. Fixing a wrong JIT test case and some compiler warnings.
450
451 2. Removed a bashism from the RunTest script.
452
453 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
454 to unsigned type, result still unsigned" that was given by an MS compiler
455 on encountering the code "-sizeof(xxx)".
456
457 4. Partial matching support is added to the JIT compiler.
458
459 5. Fixed several bugs concerned with partial matching of items that consist
460 of more than one character:
461
462 (a) /^(..)\1/ did not partially match "aba" because checking references was
463 done on an "all or nothing" basis. This also applied to repeated
464 references.
465
466 (b) \R did not give a hard partial match if \r was found at the end of the
467 subject.
468
469 (c) \X did not give a hard partial match after matching one or more
470 characters at the end of the subject.
471
472 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
473 a partial match for the string "\r".
474
475 (e) When newline was set to CRLF, the metacharacter "." did not recognize
476 a partial match for a CR character at the end of the subject string.
477
478 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
479 running pcretest, the text "(JIT)" added to the output whenever JIT is
480 actually used to run the match.
481
482 7. Individual JIT compile options can be set in pcretest by following -s+[+]
483 or /S+[+] with a digit between 1 and 7.
484
485 8. OP_NOT now supports any UTF character not just single-byte ones.
486
487 9. (*MARK) control verb is now supported by the JIT compiler.
488
489 10. The command "./RunTest list" lists the available tests without actually
490 running any of them. (Because I keep forgetting what they all are.)
491
492 11. Add PCRE_INFO_MAXLOOKBEHIND.
493
494 12. Applied a (slightly modified) user-supplied patch that improves performance
495 when the heap is used for recursion (compiled with --disable-stack-for-
496 recursion). Instead of malloc and free for each heap frame each time a
497 logical recursion happens, frames are retained on a chain and re-used where
498 possible. This sometimes gives as much as 30% improvement.
499
500 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
501 call.
502
503 14. As documented, (*COMMIT) is now confined to within a positive assertion.
504
505 15. It is now possible to link pcretest with libedit as an alternative to
506 libreadline.
507
508 16. (*COMMIT) control verb is now supported by the JIT compiler.
509
510 17. The Unicode data tables have been updated to Unicode 6.1.0.
511
512 18. Added --file-list option to pcregrep.
513
514 19. Added binary file support to pcregrep, including the -a, --binary-files,
515 -I, and --text options.
516
517 20. The madvise function is renamed for posix_madvise for QNX compatibility
518 reasons. Fixed by Giuseppe D'Angelo.
519
520 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
521
522 22. Changed the option for creating symbolic links for 16-bit man pages from
523 -s to -sf so that re-installing does not cause issues.
524
525 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
526
527 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
528 matches in certain environments (the workspace was not being correctly
529 retained). Also added to pcre_dfa_exec() a simple plausibility check on
530 some of the workspace data at the beginning of a restart.
531
532 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
533 was not doing so when it should - probably a typo introduced by SVN 528
534 (change 8.10/14).
535
536 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
537 \w+ when the character tables indicated that \x{c4} was a word character.
538 There were several related cases, all because the tests for doing a table
539 lookup were testing for characters less than 127 instead of 255.
540
541 27. If a pattern contains capturing parentheses that are not used in a match,
542 their slots in the ovector are set to -1. For those that are higher than
543 any matched groups, this happens at the end of processing. In the case when
544 there were back references that the ovector was too small to contain
545 (causing temporary malloc'd memory to be used during matching), and the
546 highest capturing number was not used, memory off the end of the ovector
547 was incorrectly being set to -1. (It was using the size of the temporary
548 memory instead of the true size.)
549
550 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
551 ovector size, it uses memory at the end of the block that it has got.
552
553 29. Check for an overlong MARK name and give an error at compile time. The
554 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
555
556 30. JIT compiler update.
557
558 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
559 Rill for the patch.
560
561 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
562
563 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
564
565 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
566 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
567 was enabled.
568
569 35. Improve JIT code generation for greedy plus quantifier.
570
571 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
572 1 to "aa" instead of to an empty string. The bug affected repeated groups
573 that could potentially match an empty string.
574
575 37. Optimizing single character iterators in JIT.
576
577 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
578 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
579 codepoints that are too big for the mode are faulted, and in a UTF mode,
580 disallowed codepoints are also faulted.
581
582 39. If PCRE was compiled with UTF support, in three places in the DFA
583 matcher there was code that should only have been obeyed in UTF mode, but
584 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
585 processing when bytes with values greater than 127 were present. In 16-bit
586 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
587 both cases the values are those that cannot be the first data item in a UTF
588 character. The three items that might have provoked this were recursions,
589 possessively repeated groups, and atomic groups.
590
591 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
592 and pcregrep, because some OS require shared objects to be explicitly
593 passed to ld, causing the link step to fail if they are not.
594
595 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
596 mode, patterns that started with \h* or \R* might be incorrectly matched.
597
598
599 Version 8.30 04-February-2012
600 -----------------------------
601
602 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
603 name is defined in ctype.h.
604
605 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
606 only in quite long subpatterns.
607
608 3. Removed the function pcre_info(), which has been obsolete and deprecated
609 since it was replaced by pcre_fullinfo() in February 2000.
610
611 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
612 match a (*MARK), and the match failed at the start of the subject, a
613 reference to memory before the start of the subject could occur. This bug
614 was introduced by fix 17 of release 8.21.
615
616 5. A reference to an unset group with zero minimum repetition was giving
617 totally wrong answers (in non-JavaScript-compatibility mode). For example,
618 /(another)?(\1?)test/ matched against "hello world test". This bug was
619 introduced in release 8.13.
620
621 6. Add support for 16-bit character strings (a large amount of work involving
622 many changes and refactorings).
623
624 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
625 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
626 from a file.
627
628 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
629 rounding is not applied in this particular case).
630
631 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
632 if they appear, or are escaped, in patterns.
633
634 10. Get rid of a number of -Wunused-but-set-variable warnings.
635
636 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
637 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
638 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
639 also returns the mark "x". This bug applied to capturing parentheses,
640 non-capturing parentheses, and atomic parentheses. It also applied to some
641 assertions.
642
643 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
644 information out of configure.ac instead of relying on pcre.h.generic, which
645 is not stored in the repository.
646
647 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
648 -lreadline.
649
650 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
651
652 15. Applied Graycode's patch to put the top-level frame on the stack rather
653 than the heap when not using the stack for recursion. This gives a
654 performance improvement in many cases when recursion is not deep.
655
656 16. Experimental code added to "pcretest -C" to output the stack frame size.
657
658
659 Version 8.21 12-Dec-2011
660 ------------------------
661
662 1. Updating the JIT compiler.
663
664 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
665 are added as well.
666
667 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
668 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
669 calling _pcre_jit_exec. Some extra comments are added.
670
671 4. (*MARK) settings inside atomic groups that do not contain any capturing
672 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
673 was introduced by change 18 for 8.20.
674
675 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
676 ECMA-262 standard.
677
678 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
679 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
680 This bug was probably introduced by change 9 of 8.13.
681
682 7. While fixing 6 above, I noticed that a number of other items were being
683 incorrectly rejected as "not fixed length". This arose partly because newer
684 opcodes had not been added to the fixed-length checking code. I have (a)
685 corrected the bug and added tests for these items, and (b) arranged for an
686 error to occur if an unknown opcode is encountered while checking for fixed
687 length instead of just assuming "not fixed length". The items that were
688 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
689 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
690 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
691
692 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
693 being incorrectly compiled and would have given unpredicatble results.
694
695 9. A possessively repeated subpattern with minimum repeat count greater than
696 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
697 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
698 the first (A) could occur when it should not.
699
700 10. Add a cast and remove a redundant test from the code.
701
702 11. JIT should use pcre_malloc/pcre_free for allocation.
703
704 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
705 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
706 is anything other than /usr, -L is still shown).
707
708 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
709
710 14. Perl does not support \N without a following name in a [] class; PCRE now
711 also gives an error.
712
713 15. If a forward reference was repeated with an upper limit of around 2000,
714 it caused the error "internal error: overran compiling workspace". The
715 maximum number of forward references (including repeats) was limited by the
716 internal workspace, and dependent on the LINK_SIZE. The code has been
717 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
718 the default depends on LINK_SIZE. There is a new upper limit (for safety)
719 of around 200,000 forward references. While doing this, I also speeded up
720 the filling in of repeated forward references.
721
722 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
723 incorrectly expecting the subject to contain another "a" after the start.
724
725 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
726 in the match, the SKIP should be ignored. This was not happening; instead
727 the SKIP was being treated as NOMATCH. For patterns such as
728 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
729 tested.
730
731 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
732 now much more compatible with Perl, in particular in cases where the result
733 is a non-match for a non-anchored pattern. For example, if
734 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
735 "m", where previously it did not return a name. A side effect of this
736 change is that for partial matches, the last encountered mark name is
737 returned, as for non matches. A number of tests that were previously not
738 Perl-compatible have been moved into the Perl-compatible test files. The
739 refactoring has had the pleasing side effect of removing one argument from
740 the match() function, thus reducing its stack requirements.
741
742 19. If the /S+ option was used in pcretest to study a pattern using JIT,
743 subsequent uses of /S (without +) incorrectly behaved like /S+.
744
745 21. Retrieve executable code size support for the JIT compiler and fixing
746 some warnings.
747
748 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
749 not work when the shorter character appeared right at the end of the
750 subject string.
751
752 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
753 systems.
754
755 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
756 output it when the /M option is used in pcretest.
757
758 25. The CheckMan script was not being included in the distribution. Also, added
759 an explicit "perl" to run Perl scripts from the PrepareRelease script
760 because this is reportedly needed in Windows.
761
762 26. If study data was being save in a file and studying had not found a set of
763 "starts with" bytes for the pattern, the data written to the file (though
764 never used) was taken from uninitialized memory and so caused valgrind to
765 complain.
766
767 27. Updated RunTest.bat as provided by Sheri Pierce.
768
769 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
770
771 29. Computation of memory usage for the table of capturing group names was
772 giving an unnecessarily large value.
773
774
775 Version 8.20 21-Oct-2011
776 ------------------------
777
778 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
779 a POSIX class. After further experiments with Perl, which convinced me that
780 Perl has bugs and confusions, a closing square bracket is no longer allowed
781 in a POSIX name. This bug also affected patterns with classes that started
782 with full stops.
783
784 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
785 captured substring, but while checking the failing first alternative,
786 substring 1 is temporarily captured. If the output vector supplied to
787 pcre_exec() was not big enough for this capture, the yield of the function
788 was still zero ("insufficient space for captured substrings"). This cannot
789 be totally fixed without adding another stack variable, which seems a lot
790 of expense for a edge case. However, I have improved the situation in cases
791 such as /(a)(b)x|abc/ matched against "abc", where the return code
792 indicates that fewer than the maximum number of slots in the ovector have
793 been set.
794
795 3. Related to (2) above: when there are more back references in a pattern than
796 slots in the output vector, pcre_exec() uses temporary memory during
797 matching, and copies in the captures as far as possible afterwards. It was
798 using the entire output vector, but this conflicts with the specification
799 that only 2/3 is used for passing back captured substrings. Now it uses
800 only the first 2/3, for compatibility. This is, of course, another edge
801 case.
802
803 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
804 main code base, and can be used by building with --enable-jit. When this is
805 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
806 runtime --no-jit option is given.
807
808 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
809 ovector, the return from the function was zero, implying that there were
810 other matches that did not fit. The correct "exactly full" value is now
811 returned.
812
813 6. If a subpattern that was called recursively or as a subroutine contained
814 (*PRUNE) or any other control that caused it to give a non-standard return,
815 invalid errors such as "Error -26 (nested recursion at the same subject
816 position)" or even infinite loops could occur.
817
818 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
819 computing the minimum length on reaching *ACCEPT, and so ended up with the
820 wrong value of 1 rather than 0. Further investigation indicates that
821 computing a minimum subject length in the presence of *ACCEPT is difficult
822 (think back references, subroutine calls), and so I have changed the code
823 so that no minimum is registered for a pattern that contains *ACCEPT.
824
825 8. If (*THEN) was present in the first (true) branch of a conditional group,
826 it was not handled as intended. [But see 16 below.]
827
828 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
829 Sheri Pierce.
830
831 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
832 the first byte in a match must be "a".
833
834 11. Change 17 for 8.13 increased the recursion depth for patterns like
835 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
836 pattern contains any instances of (*THEN). If it does not, the old
837 optimizations are restored. It would be nice to do this on a per-group
838 basis, but at the moment that is not feasible.
839
840 12. In some environments, the output of pcretest -C is CRLF terminated. This
841 broke RunTest's code that checks for the link size. A single white space
842 character after the value is now allowed for.
843
844 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
845 For "fr", it uses the Windows-specific input and output files.
846
847 14. If (*THEN) appeared in a group that was called recursively or as a
848 subroutine, it did not work as intended. [But see next item.]
849
850 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
851 pattern fragments (but not containing any | characters). If A and B are
852 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
853 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
854 D. In other words, Perl considers parentheses that do not contain any |
855 characters to be part of a surrounding alternative, whereas PCRE was
856 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
857 differently. PCRE now behaves in the same way as Perl, except in the case
858 of subroutine/recursion calls such as (?1) which have in any case always
859 been different (but PCRE had them first :-).
860
861 16. Related to 15 above: Perl does not treat the | in a conditional group as
862 creating alternatives. Such a group is treated in the same way as an
863 ordinary group without any | characters when processing (*THEN). PCRE has
864 been changed to match Perl's behaviour.
865
866 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
867 RunGrepTest script failed.
868
869 18. Change 22 for version 13 caused atomic groups to use more stack. This is
870 inevitable for groups that contain captures, but it can lead to a lot of
871 stack use in large patterns. The old behaviour has been restored for atomic
872 groups that do not contain any capturing parentheses.
873
874 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
875 suppress the check for a minimum subject length at run time. (If it was
876 given to pcre_exec() or pcre_dfa_exec() it did work.)
877
878 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
879 fail to work when decoding hex characters in data strings in EBCDIC
880 environments.
881
882 21. It appears that in at least one Mac OS environment, the isxdigit() function
883 is implemented as a macro that evaluates to its argument more than once,
884 contravening the C 90 Standard (I haven't checked a later standard). There
885 was an instance in pcretest which caused it to go wrong when processing
886 \x{...} escapes in subject strings. The has been rewritten to avoid using
887 things like p++ in the argument of isxdigit().
888
889
890 Version 8.13 16-Aug-2011
891 ------------------------
892
893 1. The Unicode data tables have been updated to Unicode 6.0.0.
894
895 2. Two minor typos in pcre_internal.h have been fixed.
896
897 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
898 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
899 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
900
901 4. There were a number of related bugs in the code for matching backrefences
902 caselessly in UTF-8 mode when codes for the characters concerned were
903 different numbers of bytes. For example, U+023A and U+2C65 are an upper
904 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
905 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
906 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
907 2-byte code at the end of the subject (it thought there wasn't enough data
908 left).
909
910 5. Comprehensive information about what went wrong is now returned by
911 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
912 as the output vector has at least 2 elements. The offset of the start of
913 the failing character and a reason code are placed in the vector.
914
915 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
916 now returned is for the first byte of the failing character, instead of the
917 last byte inspected. This is an incompatible change, but I hope it is small
918 enough not to be a problem. It makes the returned offset consistent with
919 pcre_exec() and pcre_dfa_exec().
920
921 7. pcretest now gives a text phrase as well as the error number when
922 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
923 failure, the offset and reason code are output.
924
925 8. When \R was used with a maximizing quantifier it failed to skip backwards
926 over a \r\n pair if the subsequent match failed. Instead, it just skipped
927 back over a single character (\n). This seems wrong (because it treated the
928 two characters as a single entity when going forwards), conflicts with the
929 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
930 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
931 has been changed.
932
933 9. Some internal refactoring has changed the processing so that the handling
934 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
935 time (the PCRE_DOTALL option was changed this way some time ago: version
936 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
937 which was always a bit of a fudge. It also means that there is one less
938 argument for the match() function, which reduces its stack requirements
939 slightly. This change also fixes an incompatibility with Perl: the pattern
940 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
941
942 10. More internal refactoring has drastically reduced the number of recursive
943 calls to match() for possessively repeated groups such as (abc)++ when
944 using pcre_exec().
945
946 11. While implementing 10, a number of bugs in the handling of groups were
947 discovered and fixed:
948
949 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
950 (a|)*(?1) gave a compile-time internal error.
951 ((a|)+)+ did not notice that the outer group could match an empty string.
952 (^a|^)+ was not marked as anchored.
953 (.*a|.*)+ was not marked as matching at start or after a newline.
954
955 12. Yet more internal refactoring has removed another argument from the match()
956 function. Special calls to this function are now indicated by setting a
957 value in a variable in the "match data" data block.
958
959 13. Be more explicit in pcre_study() instead of relying on "default" for
960 opcodes that mean there is no starting character; this means that when new
961 ones are added and accidentally left out of pcre_study(), testing should
962 pick them up.
963
964 14. The -s option of pcretest has been documented for ages as being an old
965 synonym of -m (show memory usage). I have changed it to mean "force study
966 for every regex", that is, assume /S for every regex. This is similar to -i
967 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
968 using it. It makes it easier to run collections of tests with and without
969 study enabled, and thereby test pcre_study() more easily. All the standard
970 tests are now run with and without -s (but some patterns can be marked as
971 "never study" - see 20 below).
972
973 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
974 restoration of the capturing data to the outer values was not happening
975 correctly.
976
977 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
978 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
979 pattern had matched an empty string, and so incorrectly returned a no
980 match.
981
982 17. There was optimizing code for the last branch of non-capturing parentheses,
983 and also for the obeyed branch of a conditional subexpression, which used
984 tail recursion to cut down on stack usage. Unfortunately, now that there is
985 the possibility of (*THEN) occurring in these branches, tail recursion is
986 no longer possible because the return has to be checked for (*THEN). These
987 two optimizations have therefore been removed. [But see 8.20/11 above.]
988
989 18. If a pattern containing \R was studied, it was assumed that \R always
990 matched two bytes, thus causing the minimum subject length to be
991 incorrectly computed because \R can also match just one byte.
992
993 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
994 was incorrectly computed.
995
996 20. If /S is present twice on a test pattern in pcretest input, it now
997 *disables* studying, thereby overriding the use of -s on the command line
998 (see 14 above). This is necessary for one or two tests to keep the output
999 identical in both cases.
1000
1001 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1002 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1003
1004 22. When an atomic group that contained a capturing parenthesis was
1005 successfully matched, but the branch in which it appeared failed, the
1006 capturing was not being forgotten if a higher numbered group was later
1007 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1008 group 1 to "a", when in fact it should be unset. This applied to multi-
1009 branched capturing and non-capturing groups, repeated or not, and also to
1010 positive assertions (capturing in negative assertions does not happen
1011 in PCRE) and also to nested atomic groups.
1012
1013 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1014 subject after a captured substring, to make it easier to tell which of a
1015 number of identical substrings has been captured.
1016
1017 24. The way atomic groups are processed by pcre_exec() has been changed so that
1018 if they are repeated, backtracking one repetition now resets captured
1019 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1020 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1021 "aaa". Previously, it would have been "a". As part of this code
1022 refactoring, the way recursive calls are handled has also been changed.
1023
1024 25. If an assertion condition captured any substrings, they were not passed
1025 back unless some other capturing happened later. For example, if
1026 (?(?=(a))a) was matched against "a", no capturing was returned.
1027
1028 26. When studying a pattern that contained subroutine calls or assertions,
1029 the code for finding the minimum length of a possible match was handling
1030 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1031 group 1 called group 2 while simultaneously a separate group 2 called group
1032 1). A stack overflow occurred in this case. I have fixed this by limiting
1033 the recursion depth to 10.
1034
1035 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1036 Fortmann. This supports explicit test numbers on the command line, and has
1037 argument validation and error reporting.
1038
1039 28. An instance of \X with an unlimited repeat could fail if at any point the
1040 first character it looked at was a mark character.
1041
1042 29. Some minor code refactoring concerning Unicode properties and scripts
1043 should reduce the stack requirement of match() slightly.
1044
1045 30. Added the '=' option to pcretest to check the setting of unused capturing
1046 slots at the end of the pattern, which are documented as being -1, but are
1047 not included in the return count.
1048
1049 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1050 compiled something random. Now it gives a compile-time error (as does
1051 Perl).
1052
1053 32. A *MARK encountered during the processing of a positive assertion is now
1054 recorded and passed back (compatible with Perl).
1055
1056 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1057 had alternative anchored branches, the search for a second match in a line
1058 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1059 matched the line "0102" twice. The same bug affected patterns that started
1060 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1061 twice.
1062
1063 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1064 does, and because of capturing effects, quantifying parenthesized
1065 assertions may at times be useful. Quantifiers are now allowed for
1066 parenthesized assertions.
1067
1068 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1069
1070 36. \g was being checked for fancy things in a character class, when it should
1071 just be a literal "g".
1072
1073 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1074 appearance of a nested POSIX class supersedes an apparent external class.
1075 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1076 unescaped square brackets may also appear as part of class names. For
1077 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1078 more like Perl. (But see 8.20/1 above.)
1079
1080 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1081 was because it thought it was \N{name}, which is not supported).
1082
1083 39. Add minix to OS list not supporting the -S option in pcretest.
1084
1085 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1086 cannot analyze patterns in sufficient detail to catch mutual recursions
1087 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1088 subgroup is called recursively as a subpattern for a second time at the
1089 same position in the subject string. In previous releases this might have
1090 been caught by the recursion limit, or it might have run out of stack.
1091
1092 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1093 happen only once. PCRE was, however incorrectly giving a compile time error
1094 "recursive call could loop indefinitely" because it cannot analyze the
1095 pattern in sufficient detail. The compile time test no longer happens when
1096 PCRE is compiling a conditional subpattern, but actual runaway loops are
1097 now caught at runtime (see 40 above).
1098
1099 42. It seems that Perl allows any characters other than a closing parenthesis
1100 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1101 has been changed to be the same.
1102
1103 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1104 as not to get warnings when autogen.sh is called. Also changed
1105 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1106
1107 44. To help people who use pcregrep to scan files containing exceedingly long
1108 lines, the following changes have been made:
1109
1110 (a) The default value of the buffer size parameter has been increased from
1111 8K to 20K. (The actual buffer used is three times this size.)
1112
1113 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1114 PCRE is built.
1115
1116 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1117 to be set at run time.
1118
1119 (d) Numerical values in pcregrep options can be followed by K or M, for
1120 example --buffer-size=50K.
1121
1122 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1123 given and the return code is set to 2.
1124
1125 45. Add a pointer to the latest mark to the callout data block.
1126
1127 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1128 partial match of an empty string instead of no match. This was specific to
1129 the use of ".".
1130
1131 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1132 complete match instead of a partial match. This bug was dependent on both
1133 the PCRE_UTF8 and PCRE_DOTALL options being set.
1134
1135 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1136 starting byte set, because \b was not being ignored.
1137
1138
1139 Version 8.12 15-Jan-2011
1140 ------------------------
1141
1142 1. Fixed some typos in the markup of the man pages, and wrote a script that
1143 checks for such things as part of the documentation building process.
1144
1145 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1146 --match-limit and --recursion-limit options (added for 8.11). In
1147 particular, this made one of the standard tests fail. (The integer value
1148 went into the wrong half of a long int.)
1149
1150 3. If the --colour option was given to pcregrep with -v (invert match), it
1151 did strange things, either producing crazy output, or crashing. It should,
1152 of course, ignore a request for colour when reporting lines that do not
1153 match.
1154
1155 4. Another pcregrep bug caused similar problems if --colour was specified with
1156 -M (multiline) and the pattern match finished with a line ending.
1157
1158 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1159 matched in multiline mode, the following line was shown as part of the
1160 match. This seems wrong, so I have changed it.
1161
1162 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1163 the check for further matches in the same line (so they could be coloured)
1164 to overrun the end of the current line. If another match was found, it was
1165 incorrectly shown (and then shown again when found in the next line).
1166
1167 7. If pcregrep was compiled under Windows, there was a reference to the
1168 function pcregrep_exit() before it was defined. I am assuming this was
1169 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1170 reported by a user. I've moved the definition above the reference.
1171
1172
1173 Version 8.11 10-Dec-2010
1174 ------------------------
1175
1176 1. (*THEN) was not working properly if there were untried alternatives prior
1177 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1178 backtracked to try for "b" instead of moving to the next alternative branch
1179 at the same level (in this case, to look for "c"). The Perl documentation
1180 is clear that when (*THEN) is backtracked onto, it goes to the "next
1181 alternative in the innermost enclosing group".
1182
1183 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1184 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1185 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1186 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1187 (*THEN).
1188
1189 3. If \s appeared in a character class, it removed the VT character from
1190 the class, even if it had been included by some previous item, for example
1191 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1192 of \s, but is part of the POSIX "space" class.)
1193
1194 4. A partial match never returns an empty string (because you can always
1195 match an empty string at the end of the subject); however the checking for
1196 an empty string was starting at the "start of match" point. This has been
1197 changed to the "earliest inspected character" point, because the returned
1198 data for a partial match starts at this character. This means that, for
1199 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1200 (previously it gave "no match").
1201
1202 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1203 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1204 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1205 has an implication that the given string is incomplete (because a partial
1206 match is preferred over a full match). For this reason, these items now
1207 give a partial match in this situation. [Aside: previously, the one case
1208 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1209 match rather than a full match, which was wrong by the old rules, but is
1210 now correct.]
1211
1212 6. There was a bug in the handling of #-introduced comments, recognized when
1213 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1214 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1215 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1216 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1217 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1218 places in pcre_compile().
1219
1220 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1221 comments when looking ahead for named forward references to subpatterns,
1222 the only newline sequence it recognized was NL. It now handles newlines
1223 according to the set newline convention.
1224
1225 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1226 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1227 cater for a lack of strerror(). These oversights have been fixed.
1228
1229 9. Added --match-limit and --recursion-limit to pcregrep.
1230
1231 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1232
1233 11. When the -o option was used, pcregrep was setting a return code of 1, even
1234 when matches were found, and --line-buffered was not being honoured.
1235
1236 12. Added an optional parentheses number to the -o and --only-matching options
1237 of pcregrep.
1238
1239 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1240 can match an empty string. The code to do it in pcretest and pcredemo
1241 needed fixing:
1242
1243 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1244 only one byte after an empty string match just before CRLF (this case
1245 just got forgotten; "any" and "anycrlf" were OK).
1246
1247 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1248 mode when an empty string match preceded an ASCII character followed by
1249 a non-ASCII character. (The code for advancing by one character rather
1250 than one byte was nonsense.)
1251
1252 (c) The pcredemo.c sample program did not have any code at all to handle
1253 the cases when CRLF is a valid newline sequence.
1254
1255 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1256 as a starting offset was within the subject string. There is now a new
1257 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1258 negative or greater than the length of the string. In order to test this,
1259 pcretest is extended to allow the setting of negative starting offsets.
1260
1261 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1262 starting offset points to the beginning of a UTF-8 character was
1263 unnecessarily clumsy. I tidied it up.
1264
1265 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1266 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1267
1268 17. Nobody had reported that the --include_dir option, which was added in
1269 release 7.7 should have been called --include-dir (hyphen, not underscore)
1270 for compatibility with GNU grep. I have changed it to --include-dir, but
1271 left --include_dir as an undocumented synonym, and the same for
1272 --exclude-dir, though that is not available in GNU grep, at least as of
1273 release 2.5.4.
1274
1275 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1276 characters from a string of bytes) have been redefined so as not to use
1277 loops, in order to improve performance in some environments. At the same
1278 time, I abstracted some of the common code into auxiliary macros to save
1279 repetition (this should not affect the compiled code).
1280
1281 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1282 compile-time error is now given if \c is not followed by an ASCII
1283 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1284 different, and any byte value is allowed.)
1285
1286 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1287 START_OPTIMIZE option, which is now allowed at compile time - but just
1288 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1289 to pcregrep and other applications that have no direct access to PCRE
1290 options. The new /Y option in pcretest sets this option when calling
1291 pcre_compile().
1292
1293 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1294 back references. Groups containing recursive back references were forced to
1295 be atomic by that change, but in the case of named groups, the amount of
1296 memory required was incorrectly computed, leading to "Failed: internal
1297 error: code overflow". This has been fixed.
1298
1299 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1300 pcretest.c, to avoid build problems in some Borland environments.
1301
1302
1303 Version 8.10 25-Jun-2010
1304 ------------------------
1305
1306 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1307 THEN.
1308
1309 2. (*ACCEPT) was not working when inside an atomic group.
1310
1311 3. Inside a character class, \B is treated as a literal by default, but
1312 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1313 causes the error). The code is unchanged, but I tidied the documentation.
1314
1315 4. Inside a character class, PCRE always treated \R and \X as literals,
1316 whereas Perl faults them if its -w option is set. I have changed PCRE so
1317 that it faults them when PCRE_EXTRA is set.
1318
1319 5. Added support for \N, which always matches any character other than
1320 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1321
1322 6. When compiling pcregrep with newer versions of gcc which may have
1323 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1324 declared with attribute warn_unused_result" were given. Just casting the
1325 result to (void) does not stop the warnings; a more elaborate fudge is
1326 needed. I've used a macro to implement this.
1327
1328 7. Minor change to pcretest.c to avoid a compiler warning.
1329
1330 8. Added four artifical Unicode properties to help with an option to make
1331 \s etc use properties (see next item). The new properties are: Xan
1332 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1333
1334 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1335 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1336 this option. Modified pcretest to add /W to test this facility. Added
1337 REG_UCP to make it available via the POSIX interface.
1338
1339 10. Added --line-buffered to pcregrep.
1340
1341 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1342 studied, and the match started with a letter with a code point greater than
1343 127 whose first byte was different to the first byte of the other case of
1344 the letter, the other case of this starting letter was not recognized
1345 (#976).
1346
1347 12. If a pattern that was studied started with a repeated Unicode property
1348 test, for example, \p{Nd}+, there was the theoretical possibility of
1349 setting up an incorrect bitmap of starting bytes, but fortunately it could
1350 not have actually happened in practice until change 8 above was made (it
1351 added property types that matched character-matching opcodes).
1352
1353 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1354 possible starting bytes for non-anchored patterns.
1355
1356 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1357 \R, and also a number of cases that involve Unicode properties, both
1358 explicit and implicit when PCRE_UCP is set.
1359
1360 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1361 input, it could crash or give wrong results if characters with values
1362 greater than 0xc0 were present in the subject string. (Detail: it assumed
1363 UTF-8 input when processing these items.)
1364
1365 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1366 size_t is 64-bit (#991).
1367
1368 17. Added a check for running out of memory when PCRE is compiled with
1369 --disable-stack-for-recursion (#990).
1370
1371 18. If the last data line in a file for pcretest does not have a newline on
1372 the end, a newline was missing in the output.
1373
1374 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1375 less than 128) in its various bitmaps. However, there is a facility for
1376 generating tables according to the current locale when PCRE is compiled. It
1377 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1378 characters, are recognized by isspace() and therefore were getting set in
1379 these tables, and indeed these tables seem to approximate to ISO 8859. This
1380 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1381 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1382 which of course cannot start UTF-8 characters. I have changed the code so
1383 that only real ASCII characters (less than 128) and the correct starting
1384 bytes for UTF-8 encodings are set for characters greater than 127 when in
1385 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1386 altogether.)
1387
1388 20. Added the /T option to pcretest so as to be able to run tests with non-
1389 standard character tables, thus making it possible to include the tests
1390 used for 19 above in the standard set of tests.
1391
1392 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1393 reference to a subpattern the other side of a comment that contains an
1394 opening parenthesis caused either an internal compiling error, or a
1395 reference to the wrong subpattern.
1396
1397
1398 Version 8.02 19-Mar-2010
1399 ------------------------
1400
1401 1. The Unicode data tables have been updated to Unicode 5.2.0.
1402
1403 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1404 configured.
1405
1406 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1407 original author of that file, following a query about its status.
1408
1409 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1410 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1411
1412 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1413 quantifier applied to a forward-referencing subroutine call, could compile
1414 incorrect code or give the error "internal error: previously-checked
1415 referenced subpattern not found".
1416
1417 6. Both MS Visual Studio and Symbian OS have problems with initializing
1418 variables to point to external functions. For these systems, therefore,
1419 pcre_malloc etc. are now initialized to local functions that call the
1420 relevant global functions.
1421
1422 7. There were two entries missing in the vectors called coptable and poptable
1423 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1424 I've fixed the data, and added a kludgy way of testing at compile time that
1425 the lengths are correct (equal to the number of opcodes).
1426
1427 8. Following on from 7, I added a similar kludge to check the length of the
1428 eint vector in pcreposix.c.
1429
1430 9. Error texts for pcre_compile() are held as one long string to avoid too
1431 much relocation at load time. To find a text, the string is searched,
1432 counting zeros. There was no check for running off the end of the string,
1433 which could happen if a new error number was added without updating the
1434 string.
1435
1436 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1437
1438 11. \K was not working if it appeared in an atomic group or in a group that
1439 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1440 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1441 the assertion is positive, but not if it is negative.
1442
1443 12. Change 11 fortuitously reduced the size of the stack frame used in the
1444 "match()" function of pcre_exec.c by one pointer. Forthcoming
1445 implementation of support for (*MARK) will need an extra pointer on the
1446 stack; I have reserved it now, so that the stack frame size does not
1447 decrease.
1448
1449 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1450 item in branch that calls a recursion is a subroutine call - as in the
1451 second branch in the above example - was incorrectly given the compile-
1452 time error "recursive call could loop indefinitely" because pcre_compile()
1453 was not correctly checking the subroutine for matching a non-empty string.
1454
1455 14. The checks for overrunning compiling workspace could trigger after an
1456 overrun had occurred. This is a "should never occur" error, but it can be
1457 triggered by pathological patterns such as hundreds of nested parentheses.
1458 The checks now trigger 100 bytes before the end of the workspace.
1459
1460 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1461
1462
1463 Version 8.01 19-Jan-2010
1464 ------------------------
1465
1466 1. If a pattern contained a conditional subpattern with only one branch (in
1467 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1468 computed the wrong minimum data length (which is of course zero for such
1469 subpatterns). This could cause incorrect "no match" results.
1470
1471 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1472 the pattern is reset in the first branch, pcre_compile() failed with
1473 "internal error: code overflow at offset...". This happened only when
1474 the reset was to the original external option setting. (An optimization
1475 abstracts leading options settings into an external setting, which was the
1476 cause of this.)
1477
1478 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1479 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1480 assertion pattern did not match (meaning that the assertion was true), it
1481 was incorrectly treated as false if the SKIP had been reached during the
1482 matching. This also applied to assertions used as conditions.
1483
1484 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1485 assertion subpattern, including such a pattern used as a condition,
1486 unpredictable results occurred, instead of the error return
1487 PCRE_ERROR_DFA_UITEM.
1488
1489 5. The C++ GlobalReplace function was not working like Perl for the special
1490 situation when an empty string is matched. It now does the fancy magic
1491 stuff that is necessary.
1492
1493 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1494 removed. (These were left over from very, very early versions of PCRE.)
1495
1496 7. Some cosmetic changes to the code to make life easier when compiling it
1497 as part of something else:
1498
1499 (a) Change DEBUG to PCRE_DEBUG.
1500
1501 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1502 called "current" as "current_branch", to prevent a collision with the
1503 Linux macro when compiled as a kernel module.
1504
1505 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1506 prevent a collision with the Linux macro when compiled as a kernel
1507 module.
1508
1509 8. In pcre_compile() there are some checks for integer overflows that used to
1510 cast potentially large values to (double). This has been changed to that
1511 when building, a check for int64_t is made, and if it is found, it is used
1512 instead, thus avoiding the use of floating point arithmetic. (There is no
1513 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1514 double.
1515
1516 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1517 2005 (difference between two addresses compared to an unsigned value).
1518
1519 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1520 custom one, because of the following reported problem in Windows:
1521
1522 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1523 under Win32.
1524 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1525 therefore missing the function definition.
1526 - The compiler thus generates a "C" signature for the test function.
1527 - The linker fails to find the "C" function.
1528 - PCRE fails to configure if asked to do so against libbz2.
1529
1530 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1531 messages were output:
1532
1533 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1534 rerunning libtoolize, to keep the correct libtool macros in-tree.
1535 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1536
1537 I have done both of these things.
1538
1539 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1540 most of the time, it *can* run out if it is given a pattern that contains a
1541 runaway infinite recursion. I updated the discussion in the pcrestack man
1542 page.
1543
1544 13. Now that we have gone to the x.xx style of version numbers, the minor
1545 version may start with zero. Using 08 or 09 is a bad idea because users
1546 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1547 interpreted as invalid octal numbers. I've updated the previous comment in
1548 configure.ac, and also added a check that gives an error if 08 or 09 are
1549 used.
1550
1551 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1552 causing partial matching to fail when the end of the subject matched \W
1553 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1554
1555 15. There were some discrepancies between the declarations in pcre_internal.h
1556 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1557 their definitions. The declarations used "const uschar *" and the
1558 definitions used USPTR. Even though USPTR is normally defined as "const
1559 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1560 reported that: "This difference in casting confuses some C++ compilers, for
1561 example, SunCC recognizes above declarations as different functions and
1562 generates broken code for hbpcre." I have changed the declarations to use
1563 USPTR.
1564
1565 16. GNU libtool is named differently on some systems. The autogen.sh script now
1566 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1567 (FreeBSD).
1568
1569 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1570 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1571 comment: "Figure out how to create a longlong from a string: strtoll and
1572 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1573 instance, but it only takes 2 args instead of 3!"
1574
1575 18. A subtle bug concerned with back references has been fixed by a change of
1576 specification, with a corresponding code fix. A pattern such as
1577 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1578 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1579 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1580 same bug. Such groups have to be quantified to be useful, or contained
1581 inside another quantified group. (If there's no repetition, the reference
1582 can never match.) The problem arises because, having left the group and
1583 moved on to the rest of the pattern, a later failure that backtracks into
1584 the group uses the captured value from the final iteration of the group
1585 rather than the correct earlier one. I have fixed this in PCRE by forcing
1586 any group that contains a reference to itself to be an atomic group; that
1587 is, there cannot be any backtracking into it once it has completed. This is
1588 similar to recursive and subroutine calls.
1589
1590
1591 Version 8.00 19-Oct-09
1592 ----------------------
1593
1594 1. The table for translating pcre_compile() error codes into POSIX error codes
1595 was out-of-date, and there was no check on the pcre_compile() error code
1596 being within the table. This could lead to an OK return being given in
1597 error.
1598
1599 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1600 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1601 in a Windows environment.
1602
1603 3. The pcregrep --count option prints the count for each file even when it is
1604 zero, as does GNU grep. However, pcregrep was also printing all files when
1605 --files-with-matches was added. Now, when both options are given, it prints
1606 counts only for those files that have at least one match. (GNU grep just
1607 prints the file name in this circumstance, but including the count seems
1608 more useful - otherwise, why use --count?) Also ensured that the
1609 combination -clh just lists non-zero counts, with no names.
1610
1611 4. The long form of the pcregrep -F option was incorrectly implemented as
1612 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1613 but it seems right to fix it, and I didn't think it was worth preserving
1614 the old behaviour.
1615
1616 5. The command line items --regex=pattern and --regexp=pattern were not
1617 recognized by pcregrep, which required --regex pattern or --regexp pattern
1618 (with a space rather than an '='). The man page documented the '=' forms,
1619 which are compatible with GNU grep; these now work.
1620
1621 6. No libpcreposix.pc file was created for pkg-config; there was just
1622 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1623
1624 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1625 when UCP support is not needed, by modifying the Python script that
1626 generates it from Unicode data files. This should not matter if the module
1627 is correctly used as a library, but I received one complaint about 50K of
1628 unwanted data. My guess is that the person linked everything into his
1629 program rather than using a library. Anyway, it does no harm.
1630
1631 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1632 was a minimum greater than 1 for a wide character in a possessive
1633 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1634 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1635 character. Chaos in the form of incorrect output or a compiling loop could
1636 result.
1637
1638 9. The restrictions on what a pattern can contain when partial matching is
1639 requested for pcre_exec() have been removed. All patterns can now be
1640 partially matched by this function. In addition, if there are at least two
1641 slots in the offset vector, the offset of the earliest inspected character
1642 for the match and the offset of the end of the subject are set in them when
1643 PCRE_ERROR_PARTIAL is returned.
1644
1645 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1646 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1647 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1648 and may be more useful for multi-segment matching.
1649
1650 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1651 used to be given if ever the end of the subject was reached; now it is
1652 given only if matching could not proceed because another character was
1653 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1654 string "Z", which now yields "no match" instead of "partial match". In the
1655 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1656 final character ended with (*FAIL).
1657
1658 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1659 if the pattern had a "must contain" character that was already found in the
1660 earlier partial match, unless partial matching was again requested. For
1661 example, with the pattern /dog.(body)?/, the "must contain" character is
1662 "g". If the first part-match was for the string "dog", restarting with
1663 "sbody" failed. This bug has been fixed.
1664
1665 13. The string returned by pcre_dfa_exec() after a partial match has been
1666 changed so that it starts at the first inspected character rather than the
1667 first character of the match. This makes a difference only if the pattern
1668 starts with a lookbehind assertion or \b or \B (\K is not supported by
1669 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1670 matching functions compatible, and I think it's the right thing to do.
1671
1672 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1673 so that the demonstration program is easily available in environments where
1674 PCRE has not been installed from source.
1675
1676 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1677 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1678 library.
1679
1680 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1681 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1682 is not the first non-POSIX option to be added. Clearly some people find
1683 these options useful.
1684
1685 17. If a caller to the POSIX matching function regexec() passes a non-zero
1686 value for nmatch with a NULL value for pmatch, the value of
1687 nmatch is forced to zero.
1688
1689 18. RunGrepTest did not have a test for the availability of the -u option of
1690 the diff command, as RunTest does. It now checks in the same way as
1691 RunTest, and also checks for the -b option.
1692
1693 19. If an odd number of negated classes containing just a single character
1694 interposed, within parentheses, between a forward reference to a named
1695 subpattern and the definition of the subpattern, compilation crashed with
1696 an internal error, complaining that it could not find the referenced
1697 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1698 [The bug was that it was starting one character too far in when skipping
1699 over the character class, thus treating the ] as data rather than
1700 terminating the class. This meant it could skip too much.]
1701
1702 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1703 /g option in pcretest when the pattern contains \K, which makes it possible
1704 to have an empty string match not at the start, even when the pattern is
1705 anchored. Updated pcretest and pcredemo to use this option.
1706
1707 21. If the maximum number of capturing subpatterns in a recursion was greater
1708 than the maximum at the outer level, the higher number was returned, but
1709 with unset values at the outer level. The correct (outer level) value is
1710 now given.
1711
1712 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1713 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1714 make it do so. The string so far is captured, making this feature
1715 compatible with Perl.
1716
1717 23. The tests have been re-organized, adding tests 11 and 12, to make it
1718 possible to check the Perl 5.10 features against Perl 5.10.
1719
1720 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1721 pattern matches a fixed length string. PCRE did not allow this; now it
1722 does. Neither allows recursion.
1723
1724 25. I finally figured out how to implement a request to provide the minimum
1725 length of subject string that was needed in order to match a given pattern.
1726 (It was back references and recursion that I had previously got hung up
1727 on.) This code has now been added to pcre_study(); it finds a lower bound
1728 to the length of subject needed. It is not necessarily the greatest lower
1729 bound, but using it to avoid searching strings that are too short does give
1730 some useful speed-ups. The value is available to calling programs via
1731 pcre_fullinfo().
1732
1733 26. While implementing 25, I discovered to my embarrassment that pcretest had
1734 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1735 study optimizations had never been tested with that matching function.
1736 Oops. What is worse, even when it was passed study data, there was a bug in
1737 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1738 were also very few tests of studied patterns with pcre_dfa_exec().
1739
1740 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1741 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1742 on the other side of the coin, they are no longer allowed to have different
1743 names, because these cannot be distinguished in PCRE, and this has caused
1744 confusion. (This is a difference from Perl.)
1745
1746 28. When duplicate subpattern names are present (necessarily with different
1747 numbers, as required by 27 above), and a test is made by name in a
1748 conditional pattern, either for a subpattern having been matched, or for
1749 recursion in such a pattern, all the associated numbered subpatterns are
1750 tested, and the overall condition is true if the condition is true for any
1751 one of them. This is the way Perl works, and is also more like the way
1752 testing by number works.
1753
1754
1755 Version 7.9 11-Apr-09
1756 ---------------------
1757
1758 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1759 (pcretest), all targets were linked against these libraries. This included
1760 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1761 libraries. This caused unwanted dependencies to be created. This problem
1762 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1763 pcretest is linked with readline.
1764
1765 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1766 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1767 moved outside it again, because FALSE and TRUE are already defined in AIX,
1768 but BOOL is not.
1769
1770 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1771 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1772
1773 4. The pcregrep documentation said spaces were inserted as well as colons (or
1774 hyphens) following file names and line numbers when outputting matching
1775 lines. This is not true; no spaces are inserted. I have also clarified the
1776 wording for the --colour (or --color) option.
1777
1778 5. In pcregrep, when --colour was used with -o, the list of matching strings
1779 was not coloured; this is different to GNU grep, so I have changed it to be
1780 the same.
1781
1782 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1783 each matching line was coloured. Now it goes on to look for further matches
1784 of any of the test patterns, which is the same behaviour as GNU grep.
1785
1786 7. A pattern that could match an empty string could cause pcregrep to loop; it
1787 doesn't make sense to accept an empty string match in pcregrep, so I have
1788 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1789 seems to be how GNU grep behaves. [But see later change 40 for release
1790 8.33.]
1791
1792 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1793 start or after a newline", because the conditional assertion was not being
1794 correctly handled. The rule now is that both the assertion and what follows
1795 in the first alternative must satisfy the test.
1796
1797 9. If auto-callout was enabled in a pattern with a conditional group whose
1798 condition was an assertion, PCRE could crash during matching, both with
1799 pcre_exec() and pcre_dfa_exec().
1800
1801 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1802 used for matching.
1803
1804 11. Unicode property support in character classes was not working for
1805 characters (bytes) greater than 127 when not in UTF-8 mode.
1806
1807 12. Added the -M command line option to pcretest.
1808
1809 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1810
1811 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1812
1813 16. Added comments and documentation about mis-use of no_arg in the C++
1814 wrapper.
1815
1816 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1817 from Martin Jerabek that uses macro names for all relevant character and
1818 string constants.
1819
1820 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1821 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1822 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1823 these, but not everybody uses configure.
1824
1825 19. A conditional group that had only one branch was not being correctly
1826 recognized as an item that could match an empty string. This meant that an
1827 enclosing group might also not be so recognized, causing infinite looping
1828 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1829 with the subject "ab", where knowledge that the repeated group can match
1830 nothing is needed in order to break the loop.
1831
1832 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1833 exec(), but without supplying a callout function, matching went wrong.
1834
1835 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1836 leak if the size of the offset vector was greater than 30. When the vector
1837 is smaller, the saved offsets during recursion go onto a local stack
1838 vector, but for larger vectors malloc() is used. It was failing to free
1839 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1840 error, in fact).
1841
1842 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1843 heapframe that is used only when UTF-8 support is enabled. This caused no
1844 problem, but was untidy.
1845
1846 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1847 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1848 included within another project.
1849
1850 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1851 slightly modified by me:
1852
1853 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1854 not building pcregrep.
1855
1856 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1857 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1858
1859 25. Forward references, both numeric and by name, in patterns that made use of
1860 duplicate group numbers, could behave incorrectly or give incorrect errors,
1861 because when scanning forward to find the reference group, PCRE was not
1862 taking into account the duplicate group numbers. A pattern such as
1863 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1864
1865 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1866 the feature of a custom pointer more persuasive (as requested by a user).
1867
1868 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1869 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1870 pre-defined.
1871
1872 28. Added support for (*UTF8) at the start of a pattern.
1873
1874 29. Arrange for flags added by the "release type" setting in CMake to be shown
1875 in the configuration summary.
1876
1877
1878 Version 7.8 05-Sep-08
1879 ---------------------
1880
1881 1. Replaced UCP searching code with optimized version as implemented for Ad
1882 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1883 stage table and inline lookup instead of a function, giving speed ups of 2
1884 to 5 times on some simple patterns that I tested. Permission was given to
1885 distribute the MultiStage2.py script that generates the tables (it's not in
1886 the tarball, but is in the Subversion repository).
1887
1888 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1889 scripts.
1890
1891 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1892 a group with a zero qualifier. The result of the study could be incorrect,
1893 or the function might crash, depending on the pattern.
1894
1895 4. Caseless matching was not working for non-ASCII characters in back
1896 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1897 It now works when Unicode Property Support is available.
1898
1899 5. In pcretest, an escape such as \x{de} in the data was always generating
1900 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1901 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1902 truncation.
1903
1904 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1905
1906 7. Added two (int) casts to pcregrep when printing the difference of two
1907 pointers, in case they are 64-bit values.
1908
1909 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1910 test 2 if it fails.
1911
1912 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1913 and a #define of that name to empty if it is not externally set. This is to
1914 allow users of MSVC to set it if necessary.
1915
1916 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1917 the convenience functions in the pcre_get.c source file.
1918
1919 11. An option change at the start of a pattern that had top-level alternatives
1920 could cause overwriting and/or a crash. This command provoked a crash in
1921 some environments:
1922
1923 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1924
1925 This potential security problem was recorded as CVE-2008-2371.
1926
1927 12. For a pattern where the match had to start at the beginning or immediately
1928 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1929 pcre_dfa_exec() could read past the end of the passed subject if there was
1930 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1931 pcretest so that it places the subject at the end of its malloc-ed buffer.
1932
1933 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1934 exec() might read past the end of the data buffer in UTF-8 mode.
1935
1936 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1937 the data contained the byte 0x85 as part of a UTF-8 character within its
1938 first line. This applied both to normal and DFA matching.
1939
1940 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1941 /^[^d]*?$/8 failed to match "abc".
1942
1943 16. Added a missing copyright notice to pcrecpp_internal.h.
1944
1945 17. Make it more clear in the documentation that values returned from
1946 pcre_exec() in ovector are byte offsets, not character counts.
1947
1948 18. Tidied a few places to stop certain compilers from issuing warnings.
1949
1950 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1951 supplied by Stefan Weber. I made a further small update for 7.8 because
1952 there is a change of source arrangements: the pcre_searchfuncs.c module is
1953 replaced by pcre_ucd.c.
1954
1955
1956 Version 7.7 07-May-08
1957 ---------------------
1958
1959 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1960 a string to a long long, pretend we don't even have a long long." This is
1961 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1962
1963 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1964 pre-7.6 versions, which defined a global no_arg variable instead of putting
1965 it in the RE class. (See also #8 below.)
1966
1967 3. Remove a line of dead code, identified by coverity and reported by Nuno
1968 Lopes.
1969
1970 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1971
1972 (1) The include/exclude patterns were being applied to the whole pathnames
1973 of files, instead of just to the final components.
1974
1975 (2) If there was more than one level of directory, the subdirectories were
1976 skipped unless they satisfied the include/exclude conditions. This is
1977 inconsistent with GNU grep (and could even be seen as contrary to the
1978 pcregrep specification - which I improved to make it absolutely clear).
1979 The action now is always to scan all levels of directory, and just
1980 apply the include/exclude patterns to regular files.
1981
1982 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1983 --exclude_dir in the tests to avoid scanning .svn directories.
1984
1985 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1986 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1987 doesn't support NULs in patterns.
1988
1989 7. Added some missing "const"s to declarations of static tables in
1990 pcre_compile.c and pcre_dfa_exec.c.
1991
1992 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1993 caused by fix #2 above. (Subsequently also a second patch to fix the
1994 first patch. And a third patch - this was a messy problem.)
1995
1996 9. Applied Craig's patch to remove the use of push_back().
1997
1998 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1999 matching function regexec().
2000
2001 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2002 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2003 references. PCRE supports relative numbers with this syntax (I don't think
2004 Oniguruma does).
2005
2006 12. Previously, a group with a zero repeat such as (...){0} was completely
2007 omitted from the compiled regex. However, this means that if the group
2008 was called as a subroutine from elsewhere in the pattern, things went wrong
2009 (an internal error was given). Such groups are now left in the compiled
2010 pattern, with a new opcode that causes them to be skipped at execution
2011 time.
2012
2013 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2014 to the way PCRE behaves:
2015
2016 (a) A lone ] character is dis-allowed (Perl treats it as data).
2017
2018 (b) A back reference to an unmatched subpattern matches an empty string
2019 (Perl fails the current match path).
2020
2021 (c) A data ] in a character class must be notated as \] because if the
2022 first data character in a class is ], it defines an empty class. (In
2023 Perl it is not possible to have an empty class.) The empty class []
2024 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2025 The negative empty class [^] matches any one character, independently
2026 of the DOTALL setting.
2027
2028 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2029 non-existent subpattern following a character class starting with ']' and
2030 containing () gave an internal compiling error instead of "reference to
2031 non-existent subpattern". Fortunately, when the pattern did exist, the
2032 compiled code was correct. (When scanning forwards to check for the
2033 existence of the subpattern, it was treating the data ']' as terminating
2034 the class, so got the count wrong. When actually compiling, the reference
2035 was subsequently set up correctly.)
2036
2037 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2038 it was being rejected as not supported by pcre_dfa_exec(), even though
2039 other assertions are supported. I have made pcre_dfa_exec() support
2040 (*FAIL).
2041
2042 16. The implementation of 13c above involved the invention of a new opcode,
2043 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2044 cannot be changed at match time, I realized I could make a small
2045 improvement to matching performance by compiling OP_ALLANY instead of
2046 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2047 on the OP_ANY path.
2048
2049 17. Compiling pcretest on Windows with readline support failed without the
2050 following two fixes: (1) Make the unistd.h include conditional on
2051 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2052
2053 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2054 ncurses library to be included for pcretest when ReadLine support is
2055 requested, but also to allow for it to be overridden. This patch came from
2056 Daniel Bergström.
2057
2058 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2059 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2060 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2061 spotting this.
2062
2063
2064 Version 7.6 28-Jan-08
2065 ---------------------
2066
2067 1. A character class containing a very large number of characters with
2068 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2069 overflow.
2070
2071 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2072 HAVE_LONG_LONG is not defined.
2073
2074 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2075 bring them up to date and include new features. This patch includes:
2076
2077 - Fixed PH's badly added libz and libbz2 support.
2078 - Fixed a problem with static linking.
2079 - Added pcredemo. [But later removed - see 7 below.]
2080 - Fixed dftables problem and added an option.
2081 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2082 HAVE_LONG_LONG.
2083 - Added readline support for pcretest.
2084 - Added an listing of the option settings after cmake has run.
2085
2086 4. A user submitted a patch to Makefile that makes it easy to create
2087 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2088 Makefile.am that cause it to include this special target, without
2089 affecting anything else. Note that the same mingw target plus all
2090 the other distribution libraries and programs are now supported
2091 when configuring with CMake (see 6 below) instead of with
2092 Configure/Make.
2093
2094 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2095 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2096 exported in the Windows port". It has not yet been confirmed that the patch
2097 solves the problem, but it does no harm.
2098
2099 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2100 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2101 with CMake, and also correct the comment about stack recursion.
2102
2103 7. Remove the automatic building of pcredemo from the ./configure system and
2104 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2105 of a program that users should build themselves after PCRE is installed, so
2106 building it automatically is not really right. What is more, it gave
2107 trouble in some build environments.
2108
2109 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2110
2111
2112 Version 7.5 10-Jan-08
2113 ---------------------
2114
2115 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2116 values in parens when parsing an RE using the C++ wrapper."
2117
2118 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2119 Characters greater than 255 were excluded from the class instead of being
2120 included.
2121
2122 3. The same bug as (2) above applied to negated POSIX classes such as
2123 [:^space:].
2124
2125 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2126 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2127 I have changed it.
2128
2129 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2130 first named subpattern) and a construct such as (?&a) would reference the
2131 first named subpattern whose name started with "a" (in other words, the
2132 length check was missing). Both these problems are fixed. "Subpattern name
2133 expected" is now given for (?&) (a zero-length name), and this patch also
2134 makes it give the same error for \k'' (previously it complained that that
2135 was a reference to a non-existent subpattern).
2136
2137 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2138 this is right because (?- can be followed by option settings as well as by
2139 digits. I have, however, made the messages clearer.
2140
2141 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2142 than the number used in the conditional) now cause a compile-time error.
2143 This is actually not compatible with Perl, which accepts such patterns, but
2144 treats the conditional as always being FALSE (as PCRE used to), but it
2145 seems to me that giving a diagnostic is better.
2146
2147 8. Change "alphameric" to the more common word "alphanumeric" in comments
2148 and messages.
2149
2150 9. Fix two occurrences of "backslash" in comments that should have been
2151 "backspace".
2152
2153 10. Remove two redundant lines of code that can never be obeyed (their function
2154 was moved elsewhere).
2155
2156 11. The program that makes PCRE's Unicode character property table had a bug
2157 which caused it to generate incorrect table entries for sequences of
2158 characters that have the same character type, but are in different scripts.
2159 It amalgamated them into a single range, with the script of the first of
2160 them. In other words, some characters were in the wrong script. There were
2161 thirteen such cases, affecting characters in the following ranges:
2162
2163 U+002b0 - U+002c1
2164 U+0060c - U+0060d
2165 U+0061e - U+00612
2166 U+0064b - U+0065e
2167 U+0074d - U+0076d
2168 U+01800 - U+01805
2169 U+01d00 - U+01d77
2170 U+01d9b - U+01dbf
2171 U+0200b - U+0200f
2172 U+030fc - U+030fe
2173 U+03260 - U+0327f
2174 U+0fb46 - U+0fbb1
2175 U+10450 - U+1049d
2176
2177 12. The -o option (show only the matching part of a line) for pcregrep was not
2178 compatible with GNU grep in that, if there was more than one match in a
2179 line, it showed only the first of them. It now behaves in the same way as
2180 GNU grep.
2181
2182 13. If the -o and -v options were combined for pcregrep, it printed a blank
2183 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2184 does the same. The return code can be used to tell if there were any
2185 non-matching lines.
2186
2187 14. Added --file-offsets and --line-offsets to pcregrep.
2188
2189 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2190 infinitely looping recursion. The bug was that positive lookaheads were not
2191 being skipped when checking for a possible empty match (negative lookaheads
2192 and both kinds of lookbehind were skipped).
2193
2194 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2195 inclusion of <windows.h> to before rather than after the definition of
2196 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2197
2198 17. Specifying a possessive quantifier with a specific limit for a Unicode
2199 character property caused pcre_compile() to compile bad code, which led at
2200 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2201 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2202 caused the error; without that there was no problem.
2203
2204 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2205
2206 19. Added --enable-pcretest-libreadline.
2207
2208 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2209 RE::GlobalReplace(). As a result, the number of replacements returned was
2210 double what it should be. I removed one of the increments, but Craig sent a
2211 later patch that removed the other one (the right fix) and added unit tests
2212 that check the return values (which was not done before).
2213
2214 21. Several CMake things:
2215
2216 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2217 the names libpcre and libpcreposix, not just pcre and pcreposix.
2218
2219 (2) The above change means that pcretest and pcregrep are now correctly
2220 linked with the newly-built libraries, not previously installed ones.
2221
2222 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2223
2224 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2225 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2226 UTF-8 newline character). The key issue is that the pattern starts .*;
2227 this means that the match must be either at the beginning, or after a
2228 newline. The bug was in the code for advancing after a failed match and
2229 checking that the new position followed a newline. It was not taking
2230 account of UTF-8 characters correctly.
2231
2232 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2233 character classes. PCRE was not treating the sequence [:...:] as a
2234 character class unless the ... were all letters. Perl, however, seems to
2235 allow any characters between [: and :], though of course it rejects as
2236 unknown any "names" that contain non-letters, because all the known class
2237 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2238 for example, whereas PCRE did not - it did not recognize a POSIX character
2239 class. This seemed a bit dangerous, so the code has been changed to be
2240 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2241 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2242 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2243 Perl does, and where it didn't before.
2244
2245 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2246 Windows environments %n is disabled by default.
2247
2248
2249 Version 7.4 21-Sep-07
2250 ---------------------
2251
2252 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2253 means that a class such as [\s] counted as "explicit reference to CR or
2254 LF". That isn't really right - the whole point of the change was to try to
2255 help when there was an actual mention of one of the two characters. So now
2256 the change happens only if \r or \n (or a literal CR or LF) character is
2257 encountered.
2258
2259 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2260 of both had grown to the point where there were only 3 bits left.
2261 Fortunately, there was spare space in the data structure, and so I have
2262 moved the internal flags into a new 16-bit field to free up more option
2263 bits.
2264
2265 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2266 but did not set the internal JCHANGED flag - either of these is enough to
2267 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2268 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2269 start sets both bits.
2270
2271 4. Added options (at build time, compile time, exec time) to change \R from
2272 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2273
2274 5. doc/pcresyntax.html was missing from the distribution.
2275
2276 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2277 compatibility, even though it is no longer used.
2278
2279 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2280 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2281 windows.h file is present (where different names are used). [This was
2282 reversed later after testing - see 16 below.]
2283
2284 8. Changed all #include <config.h> to #include "config.h". There were also
2285 some further <pcre.h> cases that I changed to "pcre.h".
2286
2287 9. When pcregrep was used with the --colour option, it missed the line ending
2288 sequence off the lines that it output.
2289
2290 10. It was pointed out to me that arrays of string pointers cause lots of
2291 relocations when a shared library is dynamically loaded. A technique of
2292 using a single long string with a table of offsets can drastically reduce
2293 these. I have refactored PCRE in four places to do this. The result is
2294 dramatic:
2295
2296 Originally: 290
2297 After changing UCP table: 187
2298 After changing error message table: 43
2299 After changing table of "verbs" 36
2300 After changing table of Posix names 22
2301
2302 Thanks to the folks working on Gregex for glib for this insight.
2303
2304 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2305 unicode-properties was also set.
2306
2307 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2308
2309 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2310 checked only for CRLF.
2311
2312 14. Added casts to pcretest.c to avoid compiler warnings.
2313
2314 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2315
2316 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2317 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2318 entirely. This removes changes made in 7 above.
2319
2320 17. The CMake files have been updated, and there is now more information about
2321 building with CMake in the NON-UNIX-USE document.
2322
2323
2324 Version 7.3 28-Aug-07
2325 ---------------------
2326
2327 1. In the rejigging of the build system that eventually resulted in 7.1, the
2328 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2329 brackets there is not right, since it causes compilers to look for an
2330 installed pcre.h, not the version that is in the source that is being
2331 compiled (which of course may be different). I have changed it back to:
2332
2333 #include "pcre.h"
2334
2335 I have a vague recollection that the change was concerned with compiling in
2336 different directories, but in the new build system, that is taken care of
2337 by the VPATH setting the Makefile.
2338
2339 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2340 when the subject happened to end in the byte 0x85 (e.g. if the last
2341 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2342 characters but of course it shouldn't be taken as a newline when it is part
2343 of another character. The bug was that, for an unlimited repeat of . in
2344 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2345 characters when looking for a newline.
2346
2347 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2348
2349 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2350 in debug output.
2351
2352 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2353 long printing in the pcrecpp unittest when running under MinGW.
2354
2355 6. ESC_K was left out of the EBCDIC table.
2356
2357 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2358 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2359 limit also applies to "virtual nesting" when a pattern is recursive, and in
2360 this case 1000 isn't so big. I have been able to remove this limit at the
2361 expense of backing off one optimization in certain circumstances. Normally,
2362 when pcre_exec() would call its internal match() function recursively and
2363 immediately return the result unconditionally, it uses a "tail recursion"
2364 feature to save stack. However, when a subpattern that can match an empty
2365 string has an unlimited repetition quantifier, it no longer makes this
2366 optimization. That gives it a stack frame in which to save the data for
2367 checking that an empty string has been matched. Previously this was taken
2368 from the 1000-entry workspace that had been reserved. So now there is no
2369 explicit limit, but more stack is used.
2370
2371 8. Applied Daniel's patches to solve problems with the import/export magic
2372 syntax that is required for Windows, and which was going wrong for the
2373 pcreposix and pcrecpp parts of the library. These were overlooked when this
2374 problem was solved for the main library.
2375
2376 9. There were some crude static tests to avoid integer overflow when computing
2377 the size of patterns that contain repeated groups with explicit upper
2378 limits. As the maximum quantifier is 65535, the maximum group length was
2379 set at 30,000 so that the product of these two numbers did not overflow a
2380 32-bit integer. However, it turns out that people want to use groups that
2381 are longer than 30,000 bytes (though not repeat them that many times).
2382 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2383 made it possible to implement the integer overflow checks in a much more
2384 dynamic way, which I have now done. The artificial limitation on group
2385 length has been removed - we now have only the limit on the total length of
2386 the compiled pattern, which depends on the LINK_SIZE setting.
2387
2388 10. Fixed a bug in the documentation for get/copy named substring when
2389 duplicate names are permitted. If none of the named substrings are set, the
2390 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2391 empty string.
2392
2393 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2394 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2395 because the ] is interpreted as the first data character and the
2396 terminating ] is not found. PCRE has been made compatible with Perl in this
2397 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2398 cause memory overwriting.
2399
2400 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2401 string has been matched (to stop an infinite loop). It was not recognizing
2402 a conditional subpattern that could match an empty string if that
2403 subpattern was within another subpattern. For example, it looped when
2404 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2405 condition was not nested. This bug has been fixed.
2406
2407 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2408 past the start of the subject in the presence of bytes with the top bit
2409 set, for example "\x8aBCD".
2410
2411 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2412 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2413
2414 14. Optimized (?!) to (*FAIL).
2415
2416 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2417 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2418 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2419 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2420 does: it's just the validity check that is more restrictive.
2421
2422 16. Inserted checks for integer overflows during escape sequence (backslash)
2423 processing, and also fixed erroneous offset values for syntax errors during
2424 backslash processing.
2425
2426 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2427 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2428
2429 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2430 caused an overrun.
2431
2432 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2433 something other than just ASCII characters) inside a group that had an
2434 unlimited repeat caused a loop at compile time (while checking to see
2435 whether the group could match an empty string).
2436
2437 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2438 [\P{Any}] did so. (Error in the code for printing property names.)
2439
2440 21. An orphan \E inside a character class could cause a crash.
2441
2442 22. A repeated capturing bracket such as (A)? could cause a wild memory
2443 reference during compilation.
2444
2445 23. There are several functions in pcre_compile() that scan along a compiled
2446 expression for various reasons (e.g. to see if it's fixed length for look
2447 behind). There were bugs in these functions when a repeated \p or \P was
2448 present in the pattern. These operators have additional parameters compared
2449 with \d, etc, and these were not being taken into account when moving along
2450 the compiled data. Specifically:
2451
2452 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2453 length.
2454
2455 (b) An item such as \pL+ within a repeated group could cause crashes or
2456 loops.
2457
2458 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2459 "reference to non-existent subpattern" error.
2460
2461 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2462
2463 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2464 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2465
2466 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2467
2468 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2469 character were causing crashes (broken optimization).
2470
2471 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2472 \p or \P) caused a compile-time loop.
2473
2474 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2475 break. For example, the unstudied pattern [\r\n]A does not match the string
2476 "\r\nA" because change 7.0/46 below moves the current point on by two
2477 characters after failing to match at the start. However, the pattern \nA
2478 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2479 the same is true. There doesn't seem any very clean way out of this, but
2480 what I have chosen to do makes the common cases work: PCRE now takes note
2481 of whether there can be an explicit match for \r or \n anywhere in the
2482 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2483 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2484 pattern has explicit CR or LF references.
2485
2486 29. Added (*CR) etc for changing newline setting at start of pattern.
2487
2488
2489 Version 7.2 19-Jun-07
2490 ---------------------
2491
2492 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2493 which is apparently normally available under Windows.
2494
2495 2. Re-jig the pcregrep tests with different newline settings in an attempt
2496 to make them independent of the local environment's newline setting.
2497
2498 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2499
2500 4. Some of the "internals" tests were previously cut out when the link size
2501 was not 2, because the output contained actual offsets. The recent new
2502 "Z" feature of pcretest means that these can be cut out, making the tests
2503 usable with all link sizes.
2504
2505 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2506 stack recursion. This gives a massive performance boost under BSD, but just
2507 a small improvement under Linux. However, it saves one field in the frame
2508 in all cases.
2509
2510 6. Added more features from the forthcoming Perl 5.10:
2511
2512 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2513 recursion call. It refers to the nth most recently opened parentheses.
2514
2515 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2516 to be opened parentheses.
2517
2518 (c) Conditions that refer to capturing parentheses can be specified
2519 relatively, for example, (?(-2)... or (?(+3)...
2520
2521 (d) \K resets the start of the current match so that everything before
2522 is not part of it.
2523
2524 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2525
2526 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2527 reference syntax.
2528
2529 (g) (?| introduces a group in which the numbering of parentheses in each
2530 alternative starts with the same number.
2531
2532 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2533
2534 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2535 PCRE_INFO_JCHANGED.
2536
2537 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2538 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2539 for detecting groups that can match an empty string.
2540
2541 9. A pattern with a very large number of alternatives (more than several
2542 hundred) was running out of internal workspace during the pre-compile
2543 phase, where pcre_compile() figures out how much memory will be needed. A
2544 bit of new cunning has reduced the workspace needed for groups with
2545 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2546 workspace instead of running out of the 4096 that are available.
2547
2548 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2549
2550 11. Applied patch from Google to remove an optimization that didn't quite work.
2551 The report of the bug said:
2552
2553 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2554 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2555 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2556
2557 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2558 it matched the wrong number of bytes.
2559
2560
2561 Version 7.1 24-Apr-07
2562 ---------------------
2563
2564 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2565 that is more "standard", making use of automake and other Autotools. There
2566 is some re-arrangement of the files and adjustment of comments consequent
2567 on this.
2568
2569 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2570 for recursive directory scanning broke on some systems because the files
2571 are not scanned in any specific order and on different systems the order
2572 was different. A call to "sort" has been inserted into RunGrepTest for the
2573 approprate test as a short-term fix. In the longer term there may be an
2574 alternative.
2575
2576 3. I had an email from Eric Raymond about problems translating some of PCRE's
2577 man pages to HTML (despite the fact that I distribute HTML pages, some
2578 people do their own conversions for various reasons). The problems
2579 concerned the use of low-level troff macros .br and .in. I have therefore
2580 removed all such uses from the man pages (some were redundant, some could
2581 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2582 HTML has been updated to handle .nf/.fi and to complain if it encounters
2583 .br or .in.
2584
2585 4. Updated comments in configure.ac that get placed in config.h.in and also
2586 arranged for config.h to be included in the distribution, with the name
2587 config.h.generic, for the benefit of those who have to compile without
2588 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2589
2590 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2591 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2592 makevp.bat for the current PCRE, using the additional files
2593 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2594
2595 6. A Windows user reported a minor discrepancy with test 2, which turned out
2596 to be caused by a trailing space on an input line that had got lost in his
2597 copy. The trailing space was an accident, so I've just removed it.
2598
2599 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2600 that is needed.
2601
2602 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2603 as "const" (a) because they are and (b) because it helps the PHP
2604 maintainers who have recently made a script to detect big data structures
2605 in the php code that should be moved to the .rodata section. I remembered
2606 to update Builducptable as well, so it won't revert if ucptable.h is ever
2607 re-created.
2608
2609 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2610 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2611 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2612 support is not required. This saves 1.5-2K of code, which is important in
2613 some applications.
2614
2615 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2616 so as not to refer to the tables, even though these functions will never be
2617 called when UTF-8 support is disabled. Otherwise there are problems with a
2618 shared library.
2619
2620 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2621
2622 (a) It was defining its arguments as char * instead of void *.
2623
2624 (b) It was assuming that all moves were upwards in memory; this was true
2625 a long time ago when I wrote it, but is no longer the case.
2626
2627 The emulated memove() is provided for those environments that have neither
2628 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2629 is clearly not the case, as these two bugs were recently reported.
2630
2631 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2632 and Detrail to create the HTML documentation, the .txt form of the man
2633 pages, and it removes trailing spaces from listed files. It also creates
2634 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2635 case, it wraps all the #defines with #ifndefs. This script should be run
2636 before "make dist".
2637
2638 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2639 with Unicode property support.
2640
2641 (a) For a maximizing quantifier, if the two different cases of the
2642 character were of different lengths in their UTF-8 codings (there are
2643 some cases like this - I found 11), and the matching function had to
2644 back up over a mixture of the two cases, it incorrectly assumed they
2645 were both the same length.
2646
2647 (b) When PCRE was configured to use the heap rather than the stack for
2648 recursion during matching, it was not correctly preserving the data for
2649 the other case of a UTF-8 character when checking ahead for a match
2650 while processing a minimizing repeat. If the check also involved
2651 matching a wide character, but failed, corruption could cause an
2652 erroneous result when trying to check for a repeat of the original
2653 character.
2654
2655 13. Some tidying changes to the testing mechanism:
2656
2657 (a) The RunTest script now detects the internal link size and whether there
2658 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2659 values substituted by "configure". (The RunGrepTest script already did
2660 this for UTF-8.) The configure.ac script no longer substitutes the
2661 relevant variables.
2662
2663 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2664 with length and offset values. This means that the output is different
2665 for different internal link sizes. Test 2 is skipped for link sizes
2666 other than 2 because of this, bypassing the problem. Unfortunately,
2667 there was also a test in test 3 (the locale tests) that used /B and
2668 failed for link sizes other than 2. Rather than cut the whole test out,
2669 I have added a new /Z option to pcretest that replaces the length and
2670 offset values with spaces. This is now used to make test 3 independent
2671 of link size. (Test 2 will be tidied up later.)
2672
2673 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2674 segmentation fault instead of returning the appropriate error message.
2675
2676 15. In multiline mode when the newline sequence was set to "any", the pattern
2677 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2678 This doesn't seem right; it now treats the CRLF combination as the line
2679 ending, and so does not match in that case. It's only a pattern such as ^$
2680 that would hit this one: something like ^ABC$ would have failed after \r
2681 and then tried again after \r\n.
2682
2683 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2684 in an attempt to make files that differ only in their line terminators
2685 compare equal. This works on Linux.
2686
2687 17. Under certain error circumstances pcregrep might try to free random memory
2688 as it exited. This is now fixed, thanks to valgrind.
2689
2690 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2691 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2692 was because its rules for how to advance for /g after matching an empty
2693 string at the end of a line did not allow for this case. They now check for
2694 it specially.
2695
2696 20. pcretest is supposed to handle patterns and data of any length, by
2697 extending its buffers when necessary. It was getting this wrong when the
2698 buffer for a data line had to be extended.
2699
2700 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2701 CRLF as a newline sequence.
2702
2703 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2704 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2705 I have nevertheless tidied it up.
2706
2707 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2708
2709 24. Added a man page for pcre-config.
2710
2711
2712 Version 7.0 19-Dec-06
2713 ---------------------
2714
2715 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2716 moving to gcc 4.1.1.
2717
2718 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2719 sys/time.h, which is documented as needed for this function. It doesn't
2720 seem to matter on Linux, but it showed up on some releases of OS X.
2721
2722 3. It seems that there are systems where bytes whose values are greater than
2723 127 match isprint() in the "C" locale. The "C" locale should be the
2724 default when a C program starts up. In most systems, only ASCII printing
2725 characters match isprint(). This difference caused the output from pcretest
2726 to vary, making some of the tests fail. I have changed pcretest so that:
2727
2728 (a) When it is outputting text in the compiled version of a pattern, bytes
2729 other than 32-126 are always shown as hex escapes.
2730
2731 (b) When it is outputting text that is a matched part of a subject string,
2732 it does the same, unless a different locale has been set for the match
2733 (using the /L modifier). In this case, it uses isprint() to decide.
2734
2735 4. Fixed a major bug that caused incorrect computation of the amount of memory
2736 required for a compiled pattern when options that changed within the
2737 pattern affected the logic of the preliminary scan that determines the
2738 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2739 that the computed length was too small. The symptoms of this bug were
2740 either the PCRE error "internal error: code overflow" from pcre_compile(),
2741 or a glibc crash with a message such as "pcretest: free(): invalid next
2742 size (fast)". Examples of patterns that provoked this bug (shown in
2743 pcretest format) are:
2744
2745 /(?-x: )/x
2746 /(?x)(?-x: \s*#\s*)/
2747 /((?i)[\x{c0}])/8
2748 /(?i:[\x{c0}])/8
2749
2750 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2751 is now done differently.
2752
2753 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2754 wrapper classes; (b) implement a new function in the C++ scanner that is
2755 more efficient than the old way of doing things because it avoids levels of
2756 recursion in the regex matching; (c) add a paragraph to the documentation
2757 for the FullMatch() function.
2758
2759 6. The escape sequence \n was being treated as whatever was defined as
2760 "newline". Not only was this contrary to the documentation, which states
2761 that \n is character 10 (hex 0A), but it also went horribly wrong when
2762 "newline" was defined as CRLF. This has been fixed.
2763
2764 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2765 was being set to -1 for the "end of line" case (supposedly a value that no
2766 character can have). Though this value is never used (the check for end of
2767 line is "zero bytes in current character"), it caused compiler complaints.
2768 I've changed it to 0xffffffff.
2769
2770 8. In pcre_version.c, the version string was being built by a sequence of
2771 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2772 string (as it is for production releases) called a macro with an empty
2773 argument. The C standard says the result of this is undefined. The gcc
2774 compiler treats it as an empty string (which was what was wanted) but it is
2775 reported that Visual C gives an error. The source has been hacked around to
2776 avoid this problem.
2777
2778 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2779 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2780 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2781 of them did).
2782
2783 10. Originally, pcretest opened its input and output without "b"; then I was
2784 told that "b" was needed in some environments, so it was added for release
2785 5.0 to both the input and output. (It makes no difference on Unix-like
2786 systems.) Later I was told that it is wrong for the input on Windows. I've
2787 now abstracted the modes into two macros, to make it easier to fiddle with
2788 them, and removed "b" from the input mode under Windows.
2789
2790 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2791
2792 12. Added -help and --help to pcretest as an official way of being reminded
2793 of the options.
2794
2795 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2796 and pcrecpp.cc because they annoy compilers at high warning levels.
2797
2798 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2799
2800 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2801 scripts are not C programs :-) and which was not noticed because it works
2802 on Linux.
2803
2804 16. pcretest is supposed to handle any length of pattern and data line (as one
2805 line or as a continued sequence of lines) by extending its input buffer if
2806 necessary. This feature was broken for very long pattern lines, leading to
2807 a string of junk being passed to pcre_compile() if the pattern was longer
2808 than about 50K.
2809
2810 17. I have done a major re-factoring of the way pcre_compile() computes the
2811 amount of memory needed for a compiled pattern. Previously, there was code
2812 that made a preliminary scan of the pattern in order to do this. That was
2813 OK when PCRE was new, but as the facilities have expanded, it has become
2814 harder and harder to keep it in step with the real compile phase, and there
2815 have been a number of bugs (see for example, 4 above). I have now found a
2816 cunning way of running the real compile function in a "fake" mode that
2817 enables it to compute how much memory it would need, while actually only
2818 ever using a few hundred bytes of working memory and without too many
2819 tests of the mode. This should make future maintenance and development
2820 easier. A side effect of this work is that the limit of 200 on the nesting
2821 depth of parentheses has been removed (though this was never a serious
2822 limitation, I suspect). However, there is a downside: pcre_compile() now
2823 runs more slowly than before (30% or more, depending on the pattern). I
2824 hope this isn't a big issue. There is no effect on runtime performance.
2825
2826 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2827 newline (only possible for the last line of a file) and it was a
2828 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2829
2830 19. Added additional timing features to pcretest. (1) The -tm option now times
2831 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2832 separate command line item, by a number that specifies the number of
2833 repeats to use when timing. The default is 50000; this gives better
2834 precision, but takes uncomfortably long for very large patterns.
2835
2836 20. Extended pcre_study() to be more clever in cases where a branch of a
2837 subpattern has no definite first character. For example, (a*|b*)[cd] would
2838 previously give no result from pcre_study(). Now it recognizes that the
2839 first character must be a, b, c, or d.
2840
2841 21. There was an incorrect error "recursive call could loop indefinitely" if
2842 a subpattern (or the entire pattern) that was being tested for matching an
2843 empty string contained only one non-empty item after a nested subpattern.
2844 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2845 incorrectly, because the \d was being skipped in the check.
2846
2847 22. The pcretest program now has a new pattern option /B and a command line
2848 option -b, which is equivalent to adding /B to every pattern. This causes
2849 it to show the compiled bytecode, without the additional information that
2850 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2851 is the same as /B/I).
2852
2853 23. A new optimization is now able automatically to treat some sequences such
2854 as a*b as a*+b. More specifically, if something simple (such as a character
2855 or a simple class like \d) has an unlimited quantifier, and is followed by
2856 something that cannot possibly match the quantified thing, the quantifier
2857 is automatically "possessified".
2858
2859 24. A recursive reference to a subpattern whose number was greater than 39
2860 went wrong under certain circumstances in UTF-8 mode. This bug could also
2861 have affected the operation of pcre_study().
2862
2863 25. Realized that a little bit of performance could be had by replacing
2864 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2865
2866 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2867
2868 27. Possessive quantifiers such as a++ were previously implemented by turning
2869 them into atomic groups such as ($>a+). Now they have their own opcodes,
2870 which improves performance. This includes the automatically created ones
2871 from 23 above.
2872
2873 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2874 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2875 the first matched character to be a colon. This applied both to named and
2876 numbered groups.
2877
2878 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2879
2880 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2881 building PCRE on Windows easier, so I have included it in the distribution.
2882
2883 31. There is now a check in pcretest against a ridiculously large number being
2884 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2885 loop, the loop is abandoned.
2886
2887 32. Forward references to subpatterns in conditions such as (?(2)...) where
2888 subpattern 2 is defined later cause pcre_compile() to search forwards in
2889 the pattern for the relevant set of parentheses. This search went wrong
2890 when there were unescaped parentheses in a character class, parentheses
2891 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2892
2893 33. "Subroutine" calls and backreferences were previously restricted to
2894 referencing subpatterns earlier in the regex. This restriction has now
2895 been removed.
2896
2897 34. Added a number of extra features that are going to be in Perl 5.10. On the
2898 whole, these are just syntactic alternatives for features that PCRE had
2899 previously implemented using the Python syntax or my own invention. The
2900 other formats are all retained for compatibility.
2901
2902 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2903 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2904 also .NET compatible.
2905
2906 (b) A recursion or subroutine call to a named group can now be defined as
2907 (?&name) as well as (?P>name).
2908
2909 (c) A backreference to a named group can now be defined as \k<name> or
2910 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2911 5.10, are also .NET compatible.
2912
2913 (d) A conditional reference to a named group can now use the syntax
2914 (?(<name>) or (?('name') as well as (?(name).
2915
2916 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2917 groups (named and numbered) that are never evaluated inline, but can be
2918 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2919 is always false. There may be only one alternative in such a group.
2920
2921 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2922 as the simple (?(R). The condition is true only if the most recent
2923 recursion is that of the given number or name. It does not search out
2924 through the entire recursion stack.
2925
2926 (g) The escape \gN or \g{N} has been added, where N is a positive or
2927 negative number, specifying an absolute or relative reference.
2928
2929 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2930 some "unreachable code" warnings.
2931
2932 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2933 things, this adds five new scripts.
2934
2935 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2936 There were also incompatibilities regarding the handling of \Q..\E inside
2937 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2938 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2939
2940 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2941 matches an empty string, and forcibly breaks the loop. There were bugs in
2942 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2943 against aaaa the result was just "a" rather than "aaaa", for example. Two
2944 separate and independent bugs (that affected different cases) have been
2945 fixed.
2946
2947 39. Refactored the code to abolish the use of different opcodes for small
2948 capturing bracket numbers. This is a tidy that I avoided doing when I
2949 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2950 The new approach is not only tidier, it makes it possible to reduce the
2951 memory needed to fix the previous bug (38).
2952
2953 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2954 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2955 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2956 mode.
2957
2958 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2959 report.
2960
2961 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2962 copy construction and assignment in the C++ wrapper.
2963
2964 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2965 couple of bugs that could have given wrong results in the "--newline=crlf"
2966 case.
2967
2968 44. Added a number of casts and did some reorganization of signed/unsigned int
2969 variables following suggestions from Dair Grant. Also renamed the variable
2970 "this" as "item" because it is a C++ keyword.
2971
2972 45. Arranged for dftables to add
2973
2974 #include "pcre_internal.h"
2975
2976 to pcre_chartables.c because without it, gcc 4.x may remove the array
2977 definition from the final binary if PCRE is built into a static library and
2978 dead code stripping is activated.
2979
2980 46. For an unanchored pattern, if a match attempt fails at the start of a
2981 newline sequence, and the newline setting is CRLF or ANY, and the next two
2982 characters are CRLF, advance by two characters instead of one.
2983
2984
2985 Version 6.7 04-Jul-06
2986 ---------------------
2987
2988 1. In order to handle tests when input lines are enormously long, pcretest has
2989 been re-factored so that it automatically extends its buffers when
2990 necessary. The code is crude, but this _is_ just a test program. The
2991 default size has been increased from 32K to 50K.
2992
2993 2. The code in pcre_study() was using the value of the re argument before
2994 testing it for NULL. (Of course, in any sensible call of the function, it
2995 won't be NULL.)
2996
2997 3. The memmove() emulation function in pcre_internal.h, which is used on
2998 systems that lack both memmove() and bcopy() - that is, hardly ever -
2999 was missing a "static" storage class specifier.
3000
3001 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3002 containing an extended class (one that cannot be represented by a bitmap
3003 because it contains high-valued characters or Unicode property items, e.g.
3004 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3005 pattern, but PCRE should not loop if you do not (it no longer does).
3006 [Detail: two cases were found: (a) a repeated subpattern containing an
3007 extended class; (b) a recursive reference to a subpattern that followed a
3008 previous extended class. It wasn't skipping over the extended class
3009 correctly when UTF-8 mode was not set.]
3010
3011 5. A negated single-character class was not being recognized as fixed-length
3012 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3013 compile error "lookbehind assertion is not fixed length".
3014
3015 6. The RunPerlTest auxiliary script was showing an unexpected difference
3016 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3017 write a Perl script that can interpret lines of an input file either as
3018 byte characters or as UTF-8, which is what "perltest" was being required to
3019 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3020 can't do is switch easily at run time between having the "use utf8;" pragma
3021 or not. In the end, I fudged it by using the RunPerlTest script to insert
3022 "use utf8;" explicitly for the UTF-8 tests.
3023
3024 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3025 the end of the subject string, contrary to the documentation and to what
3026 Perl does. This was true of both matching functions. Now it matches only at
3027 the start of the subject and immediately after *internal* newlines.
3028
3029 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3030 a pointer to an int instead of a pointer to an unsigned long int. This
3031 caused problems on 64-bit systems.
3032
3033 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3034 instance of the 'standard' template library not being so standard".
3035
3036 10. There was no check on the number of named subpatterns nor the maximum
3037 length of a subpattern name. The product of these values is used to compute
3038 the size of the memory block for a compiled pattern. By supplying a very
3039 long subpattern name and a large number of named subpatterns, the size
3040 computation could be caused to overflow. This is now prevented by limiting
3041 the length of names to 32 characters, and the number of named subpatterns
3042 to 10,000.
3043
3044 11. Subpatterns that are repeated with specific counts have to be replicated in
3045 the compiled pattern. The size of memory for this was computed from the
3046 length of the subpattern and the repeat count. The latter is limited to
3047 65535, but there was no limit on the former, meaning that integer overflow
3048 could in principle occur. The compiled length of a repeated subpattern is
3049 now limited to 30,000 bytes in order to prevent this.
3050
3051 12. Added the optional facility to have named substrings with the same name.
3052
3053 13. Added the ability to use a named substring as a condition, using the
3054 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3055 are numbers (not recommended). Forward references are permitted.
3056
3057 14. Added forward references in named backreferences (if you see what I mean).
3058
3059 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3060 pattern could run off the end of the subject. For example, the pattern
3061 "(?s)(.{1,5})"8 did this with the subject "ab".
3062
3063 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3064 PCRE_CASELESS was set when matching characters that were quantified with ?
3065 or *.
3066
3067 17. A character class other than a single negated character that had a minimum
3068 but no maximum quantifier - for example [ab]{6,} - was not handled
3069 correctly by pce_dfa_exec(). It would match only one character.
3070
3071 18. A valid (though odd) pattern that looked like a POSIX character
3072 class but used an invalid character after [ (for example [[,abc,]]) caused
3073 pcre_compile() to give the error "Failed: internal error: code overflow" or
3074 in some cases to crash with a glibc free() error. This could even happen if
3075 the pattern terminated after [[ but there just happened to be a sequence of
3076 letters, a binary zero, and a closing ] in the memory that followed.
3077
3078 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3079 over the years. Originally (before any Unicode support), just the bottom 8
3080 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3081 output from "man perlunicode" includes this:
3082
3083 The regular expression compiler produces polymorphic opcodes. That
3084 is, the pattern adapts to the data and automatically switches to
3085 the Unicode character scheme when presented with Unicode data--or
3086 instead uses a traditional byte scheme when presented with byte
3087 data.
3088
3089 Sadly, a wide octal escape does not cause a switch, and in a string with
3090 no other multibyte characters, these octal escapes are treated as before.
3091 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3092 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3093 Unicode string.
3094
3095 I have not perpetrated such confusion in PCRE. Up till now, it took just
3096 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3097 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3098 translate to the appropriate multibyte character.
3099
3100 29. Applied some refactoring to reduce the number of warnings from Microsoft
3101 and Borland compilers. This has included removing the fudge introduced
3102 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3103 a warning about an unused variable.
3104
3105 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3106 characters since release 4.0, because Perl (from release 5.004) does not.
3107 [Or at least, is documented not to: some releases seem to be in conflict
3108 with the documentation.] However, when a pattern was studied with
3109 pcre_study() and all its branches started with \s, PCRE still included VT
3110 as a possible starting character. Of course, this did no harm; it just
3111 caused an unnecessary match attempt.
3112
3113 22. Removed a now-redundant internal flag bit that recorded the fact that case
3114 dependency changed within the pattern. This was once needed for "required
3115 byte" processing, but is no longer used. This recovers a now-scarce options
3116 bit. Also moved the least significant internal flag bit to the most-
3117 significant bit of the word, which was not previously used (hangover from
3118 the days when it was an int rather than a uint) to free up another bit for
3119 the future.
3120
3121 23. Added support for CRLF line endings as well as CR and LF. As well as the
3122 default being selectable at build time, it can now be changed at runtime
3123 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3124 specify that it is scanning data with non-default line endings.
3125
3126 24. Changed the definition of CXXLINK to make it agree with the definition of
3127 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3128
3129 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3130 recursions. This makes a big different to stack usage for some patterns.
3131
3132 26. If a subpattern containing a named recursion or subroutine reference such
3133 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3134 the space required for the compiled pattern went wrong and gave too small a
3135 value. Depending on the environment, this could lead to "Failed: internal
3136 error: code overflow at offset 49" or "glibc detected double free or
3137 corruption" errors.
3138
3139 27. Applied patches from Google (a) to support the new newline modes and (b) to
3140 advance over multibyte UTF-8 characters in GlobalReplace.
3141
3142 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3143 difference for some implementation of PCRE in some Windows version.
3144
3145 29. Added some extra testing facilities to pcretest:
3146
3147 \q<number> in a data line sets the "match limit" value
3148 \Q<number> in a data line sets the "match recursion limt" value
3149 -S <number> sets the stack size, where <number> is in megabytes
3150
3151 The -S option isn't available for Windows.
3152
3153
3154 Version 6.6 06-Feb-06
3155 ---------------------
3156
3157 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3158 in pcreposix.h. I have copied the definition from pcre.h.
3159
3160 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3161 because pcre.h is no longer a built file.
3162
3163 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3164 not normally included in the compiled code.
3165
3166
3167 Version 6.5 01-Feb-06
3168 ---------------------
3169
3170 1. When using the partial match feature with pcre_dfa_exec(), it was not
3171 anchoring the second and subsequent partial matches at the new starting
3172 point. This could lead to incorrect results. For example, with the pattern
3173 /1234/, partially matching against "123" and then "a4" gave a match.
3174
3175 2. Changes to pcregrep:
3176
3177 (a) All non-match returns from pcre_exec() were being treated as failures
3178 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3179 error message is output. Some extra information is given for the
3180 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3181 probably the only errors that are likely to be caused by users (by
3182 specifying a regex that has nested indefinite repeats, for instance).
3183 If there are more than 20 of these errors, pcregrep is abandoned.
3184
3185 (b) A binary zero was treated as data while matching, but terminated the
3186 output line if it was written out. This has been fixed: binary zeroes
3187 are now no different to any other data bytes.
3188
3189 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3190 used to set a locale for matching. The --locale=xxxx long option has
3191 been added (no short equivalent) to specify a locale explicitly on the
3192 pcregrep command, overriding the environment variables.
3193
3194 (d) When -B was used with -n, some line numbers in the output were one less
3195 than they should have been.
3196
3197 (e) Added the -o (--only-matching) option.
3198
3199 (f) If -A or -C was used with -c (count only), some lines of context were
3200 accidentally printed for the final match.
3201
3202 (g) Added the -H (--with-filename) option.
3203
3204 (h) The combination of options -rh failed to suppress file names for files
3205 that were found from directory arguments.
3206
3207 (i) Added the -D (--devices) and -d (--directories) options.
3208
3209 (j) Added the -F (--fixed-strings) option.
3210
3211 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3212
3213 (l) Added the --colo(u)r option.
3214
3215 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3216 is not present by default.
3217
3218 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3219 items such as (?R) or (?1), when the recursion could match a number of
3220 alternatives. If it matched one of the alternatives, but subsequently,
3221 outside the recursion, there was a failure, the code tried to back up into
3222 the recursion. However, because of the way PCRE is implemented, this is not
3223 possible, and the result was an incorrect result from the match.
3224
3225 In order to prevent this happening, the specification of recursion has
3226 been changed so that all such subpatterns are automatically treated as
3227 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3228
3229 4. I had overlooked the fact that, in some locales, there are characters for
3230 which isalpha() is true but neither isupper() nor islower() are true. In
3231 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3232 and ordfeminine) are like this. This affected the treatment of \w and \W
3233 when they appeared in character classes, but not when they appeared outside
3234 a character class. The bit map for "word" characters is now created
3235 separately from the results of isalnum() instead of just taking it from the
3236 upper, lower, and digit maps. (Plus the underscore character, of course.)
3237
3238 5. The above bug also affected the handling of POSIX character classes such as
3239 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3240 permanent tables. Instead, the bit maps for such a class were previously
3241 created as the appropriate unions of the upper, lower, and digit bitmaps.
3242 Now they are created by subtraction from the [[:word:]] class, which has
3243 its own bitmap.
3244
3245 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3246 It is created by subtracting the vertical space characters (\x09, \x0a,
3247 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3248 subtraction was done in the overall bitmap for a character class, meaning
3249 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3250 be recognized. This bug has been fixed.
3251
3252 7. Patches from the folks at Google:
3253
3254 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3255 real life, but is still worth protecting against".
3256
3257 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3258 regular expressions".
3259
3260 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3261 have it.
3262
3263 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3264 "configure" and the latter not, in order to fix a problem somebody had
3265 with compiling the Arg class on HP-UX.
3266
3267 (e) Improve the error-handling of the C++ wrapper a little bit.
3268
3269 (f) New tests for checking recursion limiting.
3270
3271 8. The pcre_memmove() function, which is used only if the environment does not
3272 have a standard memmove() function (and is therefore rarely compiled),
3273 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3274 returning a result (though PCRE never actually uses the result).
3275
3276 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3277 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3278 returned instead of calling malloc() with an overflowing number that would
3279 most likely cause subsequent chaos.
3280
3281 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3282
3283 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3284 with this option is matched, the nmatch and pmatch options of regexec() are
3285 ignored.
3286
3287 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3288 provided in case anyone wants to the the POSIX interface with UTF-8
3289 strings.
3290
3291 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3292 C++ linking (needed for some HP-UX environments).
3293
3294 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3295 (unused parameter) and in the pcre_printint() function (omitted "default"
3296 switch label when the default is to do nothing).
3297
3298 15. Added some code to make it possible, when PCRE is compiled as a C++
3299 library, to replace subject pointers for pcre_exec() with a smart pointer
3300 class, thus making it possible to process discontinuous strings.
3301
3302 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3303 much the same function. They were added by different people who were trying
3304 to make PCRE easy to compile on non-Unix systems. It has been suggested
3305 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3306 for compiling on Windows systems. I have therefore replaced it with
3307 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3308 defaults to "extern" for C or "extern C" for C++, which works fine on
3309 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3310 SCOPE with something explicit in config.h. In addition:
3311
3312 (a) pcreposix.h still had just "extern" instead of either of these macros;
3313 I have replaced it with PCRE_DATA_SCOPE.
3314
3315 (b) Functions such as _pcre_xclass(), which are internal to the library,
3316 but external in the C sense, all had PCRE_EXPORT in their definitions.
3317 This is apparently wrong for the Windows case, so I have removed it.
3318 (It makes no difference on Unix-like systems.)
3319
3320 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3321 of recursive calls to match(). This is different to MATCH_LIMIT because
3322 that limits the total number of calls to match(), not all of which increase
3323 the depth of recursion. Limiting the recursion depth limits the amount of
3324 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3325 when PCRE is compiled, and changed at run time. A patch from Google adds
3326 this functionality to the C++ interface.
3327
3328 18. Changes to the handling of Unicode character properties:
3329
3330 (a) Updated the table to Unicode 4.1.0.
3331
3332 (b) Recognize characters that are not in the table as "Cn" (undefined).
3333
3334 (c) I revised the way the table is implemented to a much improved format
3335 which includes recognition of ranges. It now supports the ranges that
3336 are defined in UnicodeData.txt, and it also amalgamates other
3337 characters into ranges. This has reduced the number of entries in the
3338 table from around 16,000 to around 3,000, thus reducing its size
3339 considerably. I realized I did not need to use a tree structure after
3340 all - a binary chop search is just as efficient. Having reduced the
3341 number of entries, I extended their size from 6 bytes to 8 bytes to
3342 allow for more data.
3343
3344 (d) Added support for Unicode script names via properties such as \p{Han}.
3345
3346 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3347 matching that character.
3348
3349 20. When matching a repeated Unicode property with a minimum greater than zero,
3350 (for example \pL{2,}), PCRE could look past the end of the subject if it
3351 reached it while seeking the minimum number of characters. This could
3352 happen only if some of the characters were more than one byte long, because
3353 there is a check for at least the minimum number of bytes.
3354
3355 21. Refactored the implementation of \p and \P so as to be more general, to
3356 allow for more different types of property in future. This has changed the
3357 compiled form incompatibly. Anybody with saved compiled patterns that use
3358 \p or \P will have to recompile them.
3359
3360 22. Added "Any" and "L&" to the supported property types.
3361
3362 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3363 but give a compile time error if the value is greater than 0xff.
3364
3365 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3366 accidentally not being installed or uninstalled.
3367
3368 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3369 made were to insert the current release number. This seemed silly, because
3370 it made things harder for people building PCRE on systems that don't run
3371 "configure". I have turned pcre.h into a distributed file, no longer built
3372 by "configure", with the version identification directly included. There is
3373 no longer a pcre.h.in file.
3374
3375 However, this change necessitated a change to the pcre-config script as
3376 well. It is built from pcre-config.in, and one of the substitutions was the
3377 release number. I have updated configure.ac so that ./configure now finds
3378 the release number by grepping pcre.h.
3379
3380 26. Added the ability to run the tests under valgrind.
3381
3382
3383 Version 6.4 05-Sep-05
3384 ---------------------
3385
3386 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3387 "--" to be printed when multiple files were scanned, even when none of the
3388 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3389 consider it to be a bug, and have restored the previous behaviour.
3390
3391 2. A couple of code tidies to get rid of compiler warnings.
3392
3393 3. The pcretest program used to cheat by referring to symbols in the library
3394 whose names begin with _pcre_. These are internal symbols that are not
3395 really supposed to be visible externally, and in some environments it is
3396 possible to suppress them. The cheating is now confined to including
3397 certain files from the library's source, which is a bit cleaner.
3398
3399 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3400 file's purpose clearer.
3401
3402 5. Reorganized pcre_ucp_findchar().
3403
3404
3405 Version 6.3 15-Aug-05
3406 ---------------------
3407
3408 1. The file libpcre.pc.in did not have general read permission in the tarball.
3409
3410 2. There were some problems when building without C++ support:
3411
3412 (a) If C++ support was not built, "make install" and "make test" still
3413 tried to test it.
3414
3415 (b) There were problems when the value of CXX was explicitly set. Some
3416 changes have been made to try to fix these, and ...
3417
3418 (c) --disable-cpp can now be used to explicitly disable C++ support.
3419
3420 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3421 backslash in a target when C++ was disabled. This confuses some
3422 versions of "make", apparently. Using an intermediate variable solves
3423 this. (Same for CPP_LOBJ.)
3424
3425 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3426 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3427 necessary on certain architectures.
3428
3429 4. Added a setting of -export-symbols-regex to the link command to remove
3430 those symbols that are exported in the C sense, but actually are local
3431 within the library, and not documented. Their names all begin with
3432 "_pcre_". This is not a perfect job, because (a) we have to except some
3433 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3434 available (and never for static libraries). I have made a note to try to
3435 find a way round (a) in the future.
3436
3437
3438 Version 6.2 01-Aug-05
3439 ---------------------
3440
3441 1. There was no test for integer overflow of quantifier values. A construction
3442 such as {1111111111111111} would give undefined results. What is worse, if
3443 a minimum quantifier for a parenthesized subpattern overflowed and became
3444 negative, the calculation of the memory size went wrong. This could have
3445 led to memory overwriting.
3446
3447 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3448
3449 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3450 operating environments where this matters.
3451
3452 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3453 PCRE options from within the C++ wrapper.
3454
3455 5. Named capturing subpatterns were not being correctly counted when a pattern
3456 was compiled. This caused two problems: (a) If there were more than 100
3457 such subpatterns, the calculation of the memory needed for the whole
3458 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3459 back references of the form \12, where the number was greater than 9, were
3460 not recognized as back references, even though there were sufficient
3461 previous subpatterns.
3462
3463 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3464 versions of gcc, e.g. 2.95.4.
3465
3466
3467 Version 6.1 21-Jun-05
3468 ---------------------
3469
3470 1. There was one reference to the variable "posix" in pcretest.c that was not
3471 surrounded by "#if !defined NOPOSIX".
3472
3473 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3474 the cross-check on the old pcre_info() function, for the benefit of the
3475 cut-down version of PCRE that is currently imported into Exim.
3476
3477 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3478 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3479 patterns that start (?i) but I don't think that matters. The use of (?i) is
3480 just an example; this all applies to the other options as well.
3481
3482 4. Since libtool seems to echo the compile commands it is issuing, the output
3483 from "make" can be reduced a bit by putting "@" in front of each libtool
3484 compile command.
3485
3486 5. Patch from the folks at Google for configure.in to be a bit more thorough
3487 in checking for a suitable C++ installation before trying to compile the
3488 C++ stuff. This should fix a reported problem when a compiler was present,
3489 but no suitable headers.
3490
3491 6. The man pages all had just "PCRE" as their title. I have changed them to
3492 be the relevant file name. I have also arranged that these names are
3493 retained in the file doc/pcre.txt, which is a concatenation in text format
3494 of all the man pages except the little individual ones for each function.
3495
3496 7. The NON-UNIX-USE file had not been updated for the different set of source
3497 files that come with release 6. I also added a few comments about the C++
3498 wrapper.
3499
3500
3501 Version 6.0 07-Jun-05
3502 ---------------------
3503
3504 1. Some minor internal re-organization to help with my DFA experiments.
3505
3506 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3507 didn't matter for the library itself when fully configured, but did matter
3508 when compiling without UCP support, or within Exim, where the ucp files are
3509 not imported.
3510
3511 3. Refactoring of the library code to split up the various functions into
3512 different source modules. The addition of the new DFA matching code (see
3513 below) to a single monolithic source would have made it really too
3514 unwieldy, quite apart from causing all the code to be include in a
3515 statically linked application, when only some functions are used. This is
3516 relevant even without the DFA addition now that patterns can be compiled in
3517 one application and matched in another.
3518
3519 The downside of splitting up is that there have to be some external
3520 functions and data tables that are used internally in different modules of
3521 the library but which are not part of the API. These have all had their
3522 names changed to start with "_pcre_" so that they are unlikely to clash
3523 with other external names.
3524
3525 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3526 a different (DFA) algorithm. Although it is slower than the original
3527 function, it does have some advantages for certain types of matching
3528 problem.
3529
3530 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3531 including restarting after a partial match.
3532
3533 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3534 defined when compiling for Windows was sent to me. I have put it into the
3535 code, though I have no means of testing or verifying it.
3536
3537 7. Added the pcre_refcount() auxiliary function.
3538
3539 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3540 match before or at the first newline in the subject string. In pcretest,
3541 the /f option on a pattern can be used to set this.
3542
3543 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3544 would behave wrongly. This has been present in PCRE since release 4.0.
3545
3546 10. A number of changes to the pcregrep command:
3547
3548 (a) Refactored how -x works; insert ^(...)$ instead of setting
3549 PCRE_ANCHORED and checking the length, in preparation for adding
3550 something similar for -w.
3551
3552 (b) Added the -w (match as a word) option.
3553
3554 (c) Refactored the way lines are read and buffered so as to have more
3555 than one at a time available.
3556
3557 (d) Implemented a pcregrep test script.
3558
3559 (e) Added the -M (multiline match) option. This allows patterns to match
3560 over several lines of the subject. The buffering ensures that at least
3561 8K, or the rest of the document (whichever is the shorter) is available
3562 for matching (and similarly the previous 8K for lookbehind assertions).
3563
3564 (f) Changed the --help output so that it now says
3565
3566 -w, --word-regex(p)
3567
3568 instead of two lines, one with "regex" and the other with "regexp"
3569 because that confused at least one person since the short forms are the
3570 same. (This required a bit of code, as the output is generated
3571 automatically from a table. It wasn't just a text change.)
3572
3573 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3574 option but starts with a hyphen. Could be a pattern or a path name
3575 starting with a hyphen, for instance.
3576
3577 (h) "-" can be given as a file name to represent stdin.
3578
3579 (i) When file names are being printed, "(standard input)" is used for
3580 the standard input, for compatibility with GNU grep. Previously
3581 "<stdin>" was used.
3582
3583 (j) The option --label=xxx can be used to supply a name to be used for
3584 stdin when file names are being printed. There is no short form.
3585
3586 (k) Re-factored the options decoding logic because we are going to add
3587 two more options that take data. Such options can now be given in four
3588 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3589
3590 (l) Added the -A, -B, and -C options for requesting that lines of context
3591 around matches be printed.
3592
3593 (m) Added the -L option to print the names of files that do not contain
3594 any matching lines, that is, the complement of -l.
3595
3596 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3597 continue to scan other files.
3598
3599 (o) The -s option was incorrectly implemented. For compatibility with other
3600 greps, it now suppresses the error message for a non-existent or non-
3601 accessible file (but not the return code). There is a new option called
3602 -q that suppresses the output of matching lines, which was what -s was
3603 previously doing.
3604
3605 (p) Added --include and --exclude options to specify files for inclusion
3606 and exclusion when recursing.
3607
3608 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3609 Hopefully, it now does.
3610
3611 12. Missing cast in pcre_study().
3612
3613 13. Added an "uninstall" target to the makefile.
3614
3615 14. Replaced "extern" in the function prototypes in Makefile.in with
3616 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3617 world, but is set differently for Windows.
3618
3619 15. Added a second compiling function called pcre_compile2(). The only
3620 difference is that it has an extra argument, which is a pointer to an
3621 integer error code. When there is a compile-time failure, this is set
3622 non-zero, in addition to the error test pointer being set to point to an
3623 error message. The new argument may be NULL if no error number is required
3624 (but then you may as well call pcre_compile(), which is now just a
3625 wrapper). This facility is provided because some applications need a
3626 numeric error indication, but it has also enabled me to tidy up the way
3627 compile-time errors are handled in the POSIX wrapper.
3628
3629 16. Added VPATH=.libs to the makefile; this should help when building with one
3630 prefix path and installing with another. (Or so I'm told by someone who
3631 knows more about this stuff than I do.)
3632
3633 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3634 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3635 match everything, including newlines. This is not POSIX-compatible, but
3636 somebody wanted the feature. From pcretest it can be activated by using
3637 both the P and the s flags.
3638
3639 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3640
3641 19. libpcre.pc was being incorrectly installed as executable.
3642
3643 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3644 it now also looks for '\r' so that it will work unmodified on Windows.
3645
3646 21. Added Google's contributed C++ wrapper to the distribution.
3647
3648 22. Added some untidy missing memory free() calls in pcretest, to keep
3649 Electric Fence happy when testing.
3650
3651
3652
3653 Version 5.0 13-Sep-04
3654 ---------------------
3655
3656 1. Internal change: literal characters are no longer packed up into items
3657 containing multiple characters in a single byte-string. Each character
3658 is now matched using a separate opcode. However, there may be more than one
3659 byte in the character in UTF-8 mode.
3660
3661 2. The pcre_callout_block structure has two new fields: pattern_position and
3662 next_item_length. These contain the offset in the pattern to the next match
3663 item, and its length, respectively.
3664
3665 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3666 insertion of callouts before each pattern item. Added the /C option to
3667 pcretest to make use of this.
3668
3669 4. On the advice of a Windows user, the lines
3670
3671 #if defined(_WIN32) || defined(WIN32)
3672 _setmode( _fileno( stdout ), 0x8000 );
3673 #endif /* defined(_WIN32) || defined(WIN32) */
3674
3675 have been added to the source of pcretest. This apparently does useful
3676 magic in relation to line terminators.
3677
3678 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3679 for the benefit of those environments where the "b" makes a difference.
3680
3681 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3682 to know about it. I have put a hack into configure.in that adds in code
3683 to set GCC=yes if CC=icc. This seems to end up at a point in the
3684 generated configure script that is early enough to affect the setting of
3685 compiler options, which is what is needed, but I have no means of testing
3686 whether it really works. (The user who reported this had patched the
3687 generated configure script, which of course I cannot do.)
3688
3689 LATER: After change 22 below (new libtool files), the configure script
3690 seems to know about icc (and also ecc). Therefore, I have commented out
3691 this hack in configure.in.
3692
3693 7. Added support for pkg-config (2 patches were sent in).
3694
3695 8. Negated POSIX character classes that used a combination of internal tables
3696 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3697 [[:^ascii]]. Typically, they would match almost any characters. The other
3698 POSIX classes were not broken in this way.
3699
3700 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3701 to find the match, as PCRE was deluded into thinking that the match had to
3702 start at the start point or following a newline. The same bug applied to
3703 patterns with negative forward assertions or any backward assertions
3704 preceding ".*" at the start, unless the pattern required a fixed first
3705 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3706
3707 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3708 starting at the last subject character, bytes beyond the end of the subject
3709 string were read.
3710
3711 11. Renamed the variable "class" as "classbits" to make life easier for C++
3712 users. (Previously there was a macro definition, but it apparently wasn't
3713 enough.)
3714
3715 12. Added the new field "tables" to the extra data so that tables can be passed
3716 in at exec time, or the internal tables can be re-selected. This allows
3717 a compiled regex to be saved and re-used at a later time by a different
3718 program that might have everything at different addresses.
3719
3720 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3721 -R library as well as a -L library.
3722
3723 14. The debugging options of pcretest (-d on the command line or D on a
3724 pattern) showed incorrect output for anything following an extended class
3725 that contained multibyte characters and which was followed by a quantifier.
3726
3727 15. Added optional support for general category Unicode character properties
3728 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3729 support. It adds about 90K to the size of the library. The meanings of the
3730 inbuilt class escapes such as \d and \s have NOT been changed.
3731
3732 16. Updated pcredemo.c to include calls to free() to release the memory for the
3733 compiled pattern.
3734
3735 17. The generated file chartables.c was being created in the source directory
3736 instead of in the building directory. This caused the build to fail if the
3737 source directory was different from the building directory, and was
3738 read-only.
3739
3740 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3741 file. No doubt somebody will tell me if they don't make sense... Also added
3742 Dan Mooney's comments about building on OpenVMS.
3743
3744 19. Added support for partial matching via the PCRE_PARTIAL option for
3745 pcre_exec() and the \P data escape in pcretest.
3746
3747 20. Extended pcretest with 3 new pattern features:
3748
3749 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3750 write the compiled pattern to the file whose name is "rest-of-line".
3751 This is a straight binary dump of the data, with the saved pointer to
3752 the character tables forced to be NULL. The study data, if any, is
3753 written too. After writing, pcretest reads a new pattern.
3754
3755 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3756 compiled pattern from the given file. There must not be any
3757 occurrences of "<" in the file name (pretty unlikely); if there are,
3758 pcretest will instead treat the initial "<" as a pattern delimiter.
3759 After reading in the pattern, pcretest goes on to read data lines as
3760 usual.
3761
3762 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3763 and 16-bit fields in a compiled pattern, to simulate a pattern that
3764 was compiled on a host of opposite endianness.
3765
3766 21. The pcre-exec() function can now cope with patterns that were compiled on
3767 hosts of opposite endianness, with this restriction:
3768
3769 As for any compiled expression that is saved and used later, the tables
3770 pointer field cannot be preserved; the extra_data field in the arguments
3771 to pcre_exec() should be used to pass in a tables address if a value
3772 other than the default internal tables were used at compile time.
3773
3774 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3775 now diagnosed as an error. Previously, most of the time, a negative number
3776 would have been treated as zero, but if in addition "ovector" was passed as
3777 NULL, a crash could occur.
3778
3779 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3780 new versions from the libtool 1.5 distribution (the last one is a copy of
3781 a file called libtool.m4). This seems to have fixed the need to patch
3782 "configure" to support Darwin 1.3 (which I used to do). However, I still
3783 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3784 workstation).
3785
3786 24. Changed the PCRE licence to be the more standard "BSD" licence.
3787
3788
3789 Version 4.5 01-Dec-03
3790 ---------------------
3791
3792 1. There has been some re-arrangement of the code for the match() function so
3793 that it can be compiled in a version that does not call itself recursively.
3794 Instead, it keeps those local variables that need separate instances for
3795 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3796 needs to "recurse". Keeping track of where control must go is done by means
3797 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3798 hide most of the details from the main code, and operates only if
3799 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3800 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3801 operating.
3802
3803 To make it easier for callers to provide specially tailored get/free
3804 functions for this usage, two new functions, pcre_stack_malloc, and
3805 pcre_stack_free, are used. They are always called in strict stacking order,
3806 and the size of block requested is always the same.
3807
3808 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3809 PCRE has been compiled to use the stack or the heap for recursion. The
3810 -C option of pcretest uses this to show which version is compiled.
3811
3812 A new data escape \S, is added to pcretest; it causes the amounts of store
3813 obtained and freed by both kinds of malloc/free at match time to be added
3814 to the output.
3815
3816 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3817 what's available on my current Linux desktop machine.
3818
3819 3. When matching a UTF-8 string, the test for a valid string at the start has
3820 been extended. If start_offset is not zero, PCRE now checks that it points
3821 to a byte that is the start of a UTF-8 character. If not, it returns
3822 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3823 this is necessary because there may be backward assertions in the pattern.
3824 When matching the same subject several times, it may save resources to use
3825 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3826
3827 4. The code for checking the validity of UTF-8 strings has been tightened so
3828 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3829 containing "overlong sequences".
3830
3831 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3832 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3833 should have been "&", but it just so happened that all the cases this let
3834 through by mistake were picked up later in the function.
3835
3836 6. I had used a variable called "isblank" - this is a C99 function, causing
3837 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3838
3839 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3840 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3841 so the version is shown; (c) stop "make test" echoing "./RunTest".
3842
3843 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3844
3845 9. The prototype for memmove() for systems that don't have it was using
3846 size_t, but the inclusion of the header that defines size_t was later. I've
3847 moved the #includes for the C headers earlier to avoid this.
3848
3849 10. Added some adjustments to the code to make it easier to compiler on certain
3850 special systems:
3851
3852 (a) Some "const" qualifiers were missing.
3853 (b) Added the macro EXPORT before all exported functions; by default this
3854 is defined to be empty.
3855 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3856 that it reads its output file name as an argument instead of writing
3857 to the standard output and assuming this can be redirected.
3858
3859 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3860 class containing characters with values greater than 255, PCRE compilation
3861 went into a loop.
3862
3863 12. A recursive reference to a subpattern that was within another subpattern
3864 that had a minimum quantifier of zero caused PCRE to crash. For example,
3865 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3866 recursion. If the recursively-called subpattern itself had a zero repeat,
3867 that was OK.
3868
3869 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3870 buffer into which it was copied (for escape processing) was still set at
3871 1024, so long lines caused crashes.
3872
3873 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3874 "internal error: code overflow...". This applied to any character class
3875 that was followed by a possessive quantifier.
3876
3877 15. Modified the Makefile to add libpcre.la as a prerequisite for
3878 libpcreposix.la because I was told this is needed for a parallel build to
3879 work.
3880
3881 16. If a pattern that contained .* following optional items at the start was
3882 studied, the wrong optimizing data was generated, leading to matching
3883 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3884 matching string must start with a or b or c. The correct conclusion for
3885 this pattern is that a match can start with any character.
3886
3887
3888 Version 4.4 13-Aug-03
3889 ---------------------
3890
3891 1. In UTF-8 mode, a character class containing characters with values between
3892 127 and 255 was not handled correctly if the compiled pattern was studied.
3893 In fixing this, I have also improved the studying algorithm for such
3894 classes (slightly).
3895
3896 2. Three internal functions had redundant arguments passed to them. Removal
3897 might give a very teeny performance improvement.
3898
3899 3. Documentation bug: the value of the capture_top field in a callout is *one
3900 more than* the number of the hightest numbered captured substring.
3901
3902 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3903 in incorrectly linking with a previously installed version. They now link
3904 explicitly with libpcre.la.
3905
3906 5. configure.in no longer needs to recognize Cygwin specially.
3907
3908 6. A problem in pcre.in for Windows platforms is fixed.
3909
3910 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3911 pcretest, it used to include the size of the study block as part of its
3912 output. Unfortunately, the structure contains a field that has a different
3913 size on different hardware architectures. This meant that the tests that
3914 showed this size failed. As the block is currently always of a fixed size,
3915 this information isn't actually particularly useful in pcretest output, so
3916 I have just removed it.
3917
3918 8. Three pre-processor statements accidentally did not start in column 1.
3919 Sadly, there are *still* compilers around that complain, even though
3920 standard C has not required this for well over a decade. Sigh.
3921
3922 9. In pcretest, the code for checking callouts passed small integers in the
3923 callout_data field, which is a void * field. However, some picky compilers
3924 complained about the casts involved for this on 64-bit systems. Now
3925 pcretest passes the address of the small integer instead, which should get
3926 rid of the warnings.
3927
3928 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3929 both compile and run time, and gives an error if an invalid UTF-8 sequence
3930 is found. There is a option for disabling this check in cases where the
3931 string is known to be correct and/or the maximum performance is wanted.
3932
3933 11. In response to a bug report, I changed one line in Makefile.in from
3934
3935 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3936 to
3937 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3938
3939 to look similar to other lines, but I have no way of telling whether this
3940 is the right thing to do, as I do not use Windows. No doubt I'll get told
3941 if it's wrong...
3942
3943
3944 Version 4.3 21-May-03
3945 ---------------------
3946
3947 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3948 Makefile.
3949
3950 2. Some refactoring to improve the quality of the code:
3951
3952 (i) The utf8_table... variables are now declared "const".
3953
3954 (ii) The code for \cx, which used the "case flipping" table to upper case
3955 lower case letters, now just substracts 32. This is ASCII-specific,
3956 but the whole concept of \cx is ASCII-specific, so it seems
3957 reasonable.
3958
3959 (iii) PCRE was using its character types table to recognize decimal and
3960 hexadecimal digits in the pattern. This is silly, because it handles
3961 only 0-9, a-f, and A-F, but the character types table is locale-
3962 specific, which means strange things might happen. A private
3963 table is now used for this - though it costs 256 bytes, a table is
3964 much faster than multiple explicit tests. Of course, the standard
3965 character types table is still used for matching digits in subject
3966 strings against \d.
3967
3968 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3969 ending in _t are). So I've renamed it as ESC_tee.
3970
3971 3. The first argument for regexec() in the POSIX wrapper should have been
3972 defined as "const".
3973
3974 4. Changed pcretest to use malloc() for its buffers so that they can be
3975 Electric Fenced for debugging.
3976
3977 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3978 to read one or more bytes before the start of the subject string. Often this
3979 had no effect on PCRE's behaviour, but in some circumstances it could
3980 provoke a segmentation fault.
3981
3982 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3983 to try to read one or more bytes before the start of the subject string.
3984
3985 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3986 UTF-8 support could misbehave in various ways if the subject string
3987 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3988 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3989 back over UTF-8 characters.)
3990
3991
3992 Version 4.2 14-Apr-03
3993 ---------------------
3994
3995 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3996
3997 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3998 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3999 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4000 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4001 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4002 and BUILD_EXEEXT
4003 Note: automatic setting of the BUILD variables is not yet working
4004 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4005 compile-time but not at link-time
4006 [LINK]: use for linking executables only
4007 make different versions for Windows and non-Windows
4008 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4009 libraries
4010 [LINK_FOR_BUILD]: new variable
4011 [OBJEXT]: use throughout
4012 [EXEEXT]: use throughout
4013 <winshared>: new target
4014 <wininstall>: new target
4015 <dftables.o>: use native compiler
4016 <dftables>: use native linker
4017 <install>: handle Windows platform correctly
4018 <clean>: ditto
4019 <check>: ditto
4020 copy DLL to top builddir before testing
4021
4022 As part of these changes, -no-undefined was removed again. This was reported
4023 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4024 in any case.
4025
4026 3. Some tidies to get rid of compiler warnings:
4027
4028 . In the match_data structure, match_limit was an unsigned long int, whereas
4029 match_call_count was an int. I've made them both unsigned long ints.
4030
4031 . In pcretest the fact that a const uschar * doesn't automatically cast to
4032 a void * provoked a warning.
4033
4034 . Turning on some more compiler warnings threw up some "shadow" variables
4035 and a few more missing casts.
4036
4037 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4038 option, a class that contained a single character with a value between 128
4039 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4040
4041 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4042 option, a class that contained several characters, but with at least one
4043 whose value was between 128 and 255 caused PCRE to crash.
4044
4045
4046 Version 4.1 12-Mar-03
4047 ---------------------
4048
4049 1. Compiling with gcc -pedantic found a couple of places where casts were
4050 needed, and a string in dftables.c that was longer than standard compilers are
4051 required to support.
4052
4053 2. Compiling with Sun's compiler found a few more places where the code could
4054 be tidied up in order to avoid warnings.
4055
4056 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4057 first of these names is deprecated in the latest Autoconf in favour of the name
4058 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4059 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4060 analogy I have changed it to CFLAGS_FOR_BUILD.
4061
4062 4. Added -no-undefined to the linking command in the Makefile, because this is
4063 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4064 linking step for the pcreposix library.
4065
4066 5. PCRE was failing to diagnose the case of two named groups with the same
4067 name.
4068
4069 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4070 literal character that is needed in the subject for a match, and scans along to
4071 ensure that it is present before embarking on the full matching process. This
4072 saves time in cases of nested unlimited repeats that are never going to match.
4073 Problem: the scan can take a lot of time if the subject is very long (e.g.
4074 megabytes), thus penalizing straightforward matches. It is now done only if the
4075 amount of subject to be scanned is less than 1000 bytes.
4076
4077 7. A lesser problem with the same optimization is that it was recording the
4078 first character of an anchored pattern as "needed", thus provoking a search
4079 right along the subject, even when the first match of the pattern was going to
4080 fail. The "needed" character is now not set for anchored patterns, unless it
4081 follows something in the pattern that is of non-fixed length. Thus, it still
4082 fulfils its original purpose of finding quick non-matches in cases of nested
4083 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4084
4085
4086 Version 4.0 17-Feb-03
4087 ---------------------
4088
4089 1. If a comment in an extended regex that started immediately after a meta-item
4090 extended to the end of string, PCRE compiled incorrect data. This could lead to
4091 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4092
4093 2. Moved to autoconf 2.53 and libtool 1.4.2.
4094
4095 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4096 the special perltest8 script is no longer needed - all the tests can be run
4097 from a single perltest script.
4098
4099 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4100 by \s. It has now been removed in PCRE. This means it isn't recognized as
4101 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4102 class [:space:] *does* include VT, thereby creating a mess.
4103
4104 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4105 space and tab.
4106
4107 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4108 its new features into the main test script, reducing the number of scripts.
4109
4110 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4111 were backward compatible, and made the (?i) apply to the whole pattern, as if
4112 /i were given. Now it behaves more logically, and applies the option setting
4113 only to what follows. PCRE has been changed to follow suit. However, if it
4114 finds options settings right at the start of the pattern, it extracts them into
4115 the global options, as before. Thus, they show up in the info data.
4116
4117 8. Added support for the \Q...\E escape sequence. Characters in between are
4118 treated as literals. This is slightly different from Perl in that $ and @ are
4119 also handled as literals inside the quotes. In Perl, they will cause variable
4120 interpolation. Note the following examples:
4121
4122 Pattern PCRE matches Perl matches
4123
4124 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4125 \Qabc\$xyz\E abc\$xyz abc\$xyz
4126 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4127
4128 For compatibility with Perl, \Q...\E sequences are recognized inside character
4129 classes as well as outside them.
4130
4131 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4132 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4133 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4134 signed/unsigned warnings.
4135
4136 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4137 option for pcretest, so I've replaced it by a simple function that does just
4138 that job.
4139
4140 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4141 "pcregrep -".
4142
4143 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4144 Java package. This provides some syntactic sugar for simple cases of what my
4145 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4146 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4147 item, you can use this simplified notation. Note that only makes sense with
4148 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4149 greediness, whatever the setting of the PCRE_UNGREEDY option.
4150
4151 13. A change of greediness default within a pattern was not taking effect at
4152 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4153 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4154 was abstracted outside.
4155
4156 14. PCRE now supports the \G assertion. It is true when the current matching
4157 position is at the start point of the match. This differs from \A when the
4158 starting offset is non-zero. Used with the /g option of pcretest (or similar
4159 code), it works in the same way as it does for Perl's /g option. If all
4160 alternatives of a regex begin with \G, the expression is anchored to the start
4161 match position, and the "anchored" flag is set in the compiled expression.
4162
4163 15. Some bugs concerning the handling of certain option changes within patterns
4164 have been fixed. These applied to options other than (?ims). For example,
4165 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4166 been the other way round. Some of this was related to change 7 above.
4167
4168 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4169 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4170 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4171 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4172
4173 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4174 mode. Unlike ".", it always matches newline, whatever the setting of
4175 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4176 assertions. Perl allows it, but it doesn't (in general) work because it can't
4177 calculate the length of the lookbehind. At least, that's the case for Perl
4178 5.8.0 - I've been told they are going to document that it doesn't work in
4179 future.
4180
4181 18. Added an error diagnosis for escapes that PCRE does not support: these are
4182 \L, \l, \N, \P, \p, \U, \u, and \X.
4183
4184 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4185 reading past the end of the pattern in cases such as /[abcd/.
4186
4187 20. PCRE was getting more memory than necessary for patterns with classes that
4188 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4189
4190 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4191 compiling PCRE for use with Virtual Pascal.
4192
4193 22. Small fix to the Makefile to make it work properly if the build is done
4194 outside the source tree.
4195
4196 23. Added a new extension: a condition to go with recursion. If a conditional
4197 subpattern starts with (?(R) the "true" branch is used if recursion has
4198 happened, whereas the "false" branch is used only at the top level.
4199
4200 24. When there was a very long string of literal characters (over 255 bytes
4201 without UTF support, over 250 bytes with UTF support), the computation of how
4202 much memory was required could be incorrect, leading to segfaults or other
4203 strange effects.
4204
4205 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4206 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4207 there was a subsequent back reference to those brackets. This meant that, for
4208 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4209 possible to check for precisely this case. All we can do is abandon the
4210 optimization if .* occurs inside capturing brackets when there are any back
4211 references whatsoever. (See below for a better fix that came later.)
4212
4213 26. The handling of the optimization for finding the first character of a
4214 non-anchored pattern, and for finding a character that is required later in the
4215 match were failing in some cases. This didn't break the matching; it just
4216 failed to optimize when it could. The way this is done has been re-implemented.
4217
4218 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4219
4220 28. Added a new feature that provides some of the functionality that Perl
4221 provides with (?{...}). The facility is termed a "callout". The way it is done
4222 in PCRE is for the caller to provide an optional function, by setting
4223 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4224 global variable. By default it is unset, which disables all calling out. To get
4225 the function called, the regex must include (?C) at appropriate points. This
4226 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4227 This provides a means of identifying different callout points. When PCRE
4228 reaches such a point in the regex, if pcre_callout has been set, the external
4229 function is called. It is provided with data in a structure called
4230 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4231 matching continues; if it returns a non-zero value, the match at the current
4232 point fails. However, backtracking will occur if possible. [This was changed
4233 later and other features added - see item 49 below.]
4234
4235 29. pcretest is upgraded to test the callout functionality. It provides a
4236 callout function that displays information. By default, it shows the start of
4237 the match and the current position in the text. There are some new data escapes
4238 to vary what happens:
4239
4240 \C+ in addition, show current contents of captured substrings
4241 \C- do not supply a callout function
4242 \C!n return 1 when callout number n is reached
4243 \C!n!m return 1 when callout number n is reached for the mth time
4244
4245 30. If pcregrep was called with the -l option and just a single file name, it
4246 output "<stdin>" if a match was found, instead of the file name.
4247
4248 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4249 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4250 pcre_exec(). This saves a malloc/free per call. The default value of
4251 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4252 when configuring.
4253
4254 32. The default maximum size of a compiled pattern is 64K. There have been a
4255 few cases of people hitting this limit. The code now uses macros to handle the
4256 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4257 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4258 configuring. Tests 2 and 5 work only with 2-byte links because they output
4259 debugging information about compiled patterns.
4260
4261 33. Internal code re-arrangements:
4262
4263 (a) Moved the debugging function for printing out a compiled regex into
4264 its own source file (printint.c) and used #include to pull it into
4265 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4266 separate copies.
4267
4268 (b) Defined the list of op-code names for debugging as a macro in
4269 internal.h so that it is next to the definition of the opcodes.
4270
4271 (c) Defined a table of op-code lengths for simpler skipping along compiled
4272 code. This is again a macro in internal.h so that it is next to the
4273 definition of the opcodes.
4274
4275 34. Added support for recursive calls to individual subpatterns, along the
4276 lines of Robin Houston's patch (but implemented somewhat differently).
4277
4278 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4279 allow it to read and process whole directories in Win32. This code was
4280 contributed by Lionel Fourquaux; it has not been tested by me.
4281
4282 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4283 used to name a group. Names consist of alphanumerics and underscores, and must
4284 be unique. Back references use the syntax (?P=name) and recursive calls use
4285 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4286 numbers. The function pcre_fullinfo() can be used after compilation to extract
4287 a name/number map. There are three relevant calls:
4288
4289 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4290 PCRE_INFO_NAMECOUNT yields the number of entries
4291 PCRE_INFO_NAMETABLE yields a pointer to the map.
4292
4293 The map is a vector of fixed-size entries. The size of each entry depends on
4294 the length of the longest name used. The first two bytes of each entry are the
4295 group number, most significant byte first. There follows the corresponding
4296 name, zero terminated. The names are in alphabetical order.
4297
4298 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4299 case instead of 255. Making it the same both with and without UTF-8 support
4300 means that the same test output works with both.
4301
4302 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4303 calling malloc() with a zero argument.
4304
4305 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4306 optimization. I've improved things by keeping a bitmap of backreferences with
4307 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4308 fact referenced, the optimization can be applied. It is unlikely that a
4309 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4310 the match to follow \n) will appear inside brackets with a number greater than
4311 31, but if it does, any back reference > 31 suppresses the optimization.
4312
4313 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4314 of disabling numbered capturing parentheses. Any opening parenthesis that is
4315 not followed by ? behaves as if it were followed by ?: but named parentheses
4316 can still be used for capturing (and they will acquire numbers in the usual
4317 way).
4318
4319 41. Redesigned the return codes from the match() function into yes/no/error so
4320 that errors can be passed back from deep inside the nested calls. A malloc
4321 failure while inside a recursive subpattern call now causes the
4322 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4323
4324 42. It is now possible to set a limit on the number of times the match()
4325 function is called in a call to pcre_exec(). This facility makes it possible to
4326 limit the amount of recursion and backtracking, though not in a directly
4327 obvious way, because the match() function is used in a number of different
4328 circumstances. The count starts from zero for each position in the subject
4329 string (for non-anchored patterns). The default limit is, for compatibility, a
4330 large number, namely 10 000 000. You can change this in two ways:
4331
4332 (a) When configuring PCRE before making, you can use --with-match-limit=n
4333 to set a default value for the compiled library.
4334
4335 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4336 a different value is set. See 45 below.
4337
4338 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4339
4340 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4341 of things that can be changed at compile time. The first argument specifies
4342 what is wanted and the second points to where the information is to be placed.
4343 The current list of available information is:
4344
4345 PCRE_CONFIG_UTF8
4346
4347 The output is an integer that is set to one if UTF-8 support is available;
4348 otherwise it is set to zero.
4349
4350 PCRE_CONFIG_NEWLINE
4351
4352 The output is an integer that it set to the value of the code that is used for
4353 newline. It is either LF (10) or CR (13).
4354
4355 PCRE_CONFIG_LINK_SIZE
4356
4357 The output is an integer that contains the number of bytes used for internal
4358 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4359
4360 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4361
4362 The output is an integer that contains the threshold above which the POSIX
4363 interface uses malloc() for output vectors. See item 31 above.
4364
4365 PCRE_CONFIG_MATCH_LIMIT
4366
4367 The output is an unsigned integer that contains the default limit of the number
4368 of match() calls in a pcre_exec() execution. See 42 above.
4369
4370 44. pcretest has been upgraded by the addition of the -C option. This causes it
4371 to extract all the available output from the new pcre_config() function, and to
4372 output it. The program then exits immediately.
4373
4374 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4375 order to support additional features. One way would have been to define
4376 pcre_exec2() (for example) with extra arguments, but this would not have been
4377 extensible, and would also have required all calls to the original function to
4378 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4379 is used for passing in "extra" data from pcre_study().
4380
4381 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4382 contains the following fields:
4383
4384 flags a bitmap indicating which of the following fields are set
4385 study_data opaque data from pcre_study()
4386 match_limit a way of specifying a limit on match() calls for a specific
4387 call to pcre_exec()
4388 callout_data data for callouts (see 49 below)
4389
4390 The flag bits are also defined in pcre.h, and are
4391
4392 PCRE_EXTRA_STUDY_DATA
4393 PCRE_EXTRA_MATCH_LIMIT
4394 PCRE_EXTRA_CALLOUT_DATA
4395
4396 The pcre_study() function now returns one of these new pcre_extra blocks, with
4397 the actual study data pointed to by the study_data field, and the
4398 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4399 before. That is, this change is entirely upwards-compatible and requires no
4400 change to existing code.
4401
4402 If you want to pass in additional data to pcre_exec(), you can either place it
4403 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4404 block.
4405
4406 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4407 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4408 times with different match limits, until it finds the minimum value needed for
4409 pcre_exec() to complete. The value is then output. This can be instructive; for
4410 most simple matches the number is quite small, but for pathological cases it
4411 gets very large very quickly.
4412
4413 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4414 returns the size of the data block pointed to by the study_data field in a
4415 pcre_extra block, that is, the value that was passed as the argument to
4416 pcre_malloc() when PCRE was getting memory in which to place the information
4417 created by pcre_study(). The fourth argument should point to a size_t variable.
4418 pcretest has been extended so that this information is shown after a successful
4419 pcre_study() call when information about the compiled regex is being displayed.
4420
4421 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4422 because what follows is always an absolute path. (Later: it turns out that this
4423 is more than cosmetic for MinGW, because it doesn't like empty path
4424 components.)
4425
4426 49. Some changes have been made to the callout feature (see 28 above):
4427
4428 (i) A callout function now has three choices for what it returns:
4429
4430 0 => success, carry on matching
4431 > 0 => failure at this point, but backtrack if possible
4432 < 0 => serious error, return this value from pcre_exec()
4433
4434 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4435 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4436 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4437 use by callout functions. It will never be used by PCRE itself.
4438
4439 (ii) The pcre_extra structure (see 45 above) has a void * field called
4440 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4441 pcre_callout_block structure has a field of the same name. The contents of
4442 the field passed in the pcre_extra structure are passed to the callout
4443 function in the corresponding field in the callout block. This makes it
4444 easier to use the same callout-containing regex from multiple threads. For
4445 testing, the pcretest program has a new data escape
4446
4447 \C*n pass the number n (may be negative) as callout_data
4448
4449 If the callout function in pcretest receives a non-zero value as
4450 callout_data, it returns that value.
4451
4452 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4453 there were some redundant $(CFLAGS) in commands that are now specified as
4454 $(LINK), which already includes $(CFLAGS).
4455
4456 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4457 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4458 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4459 one-byte characters throughout. Note that case-insensitive matching applies
4460 only to characters whose values are less than 256. PCRE doesn't support the
4461 notion of cases for higher-valued characters.
4462
4463 (i) A character class whose characters are all within 0-255 is handled as
4464 a bit map, and the map is inverted for negative classes. Previously, a
4465 character > 255 always failed to match such a class; however it should
4466 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4467
4468 (ii) A negated character class with a single character < 255 is coded as
4469 "not this character" (OP_NOT). This wasn't working properly when the test
4470 character was multibyte, either singly or repeated.
4471
4472 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4473 mode, for example: \x{100}{2,3}.
4474
4475 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4476 singly or repeated) now correctly test multibyte characters. However,
4477 PCRE doesn't recognize any characters with values greater than 255 as
4478 digits, spaces, or word characters. Such characters always match \D, \S,
4479 and \W, and never match \d, \s, or \w.
4480
4481 (v) Classes may now contain characters and character ranges with values
4482 greater than 255. For example: [ab\x{100}-\x{400}].
4483
4484 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4485 PCRE in UTF-8 mode.
4486
4487 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4488 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4489 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4490 value.)
4491
4492 53. The single man page has become too large. I have therefore split it up into
4493 a number of separate man pages. These also give rise to individual HTML pages;
4494 these are now put in a separate directory, and there is an index.html page that
4495 lists them all. Some hyperlinking between the pages has been installed.
4496
4497 54. Added convenience functions for handling named capturing parentheses.
4498
4499 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4500 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4501 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4502 are faulted.
4503
4504 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4505 calling configure. These values are used when compiling the dftables.c program
4506 which is run to generate the source of the default character tables. They
4507 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4508 you will need to set these values.
4509
4510 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4511
4512
4513 Version 3.9 02-Jan-02
4514 ---------------------
4515
4516 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4517
4518 2. If --disable-static was given, the building process failed when trying to
4519 build pcretest and pcregrep. (For some reason it was using libtool to compile
4520 them, which is not right, as they aren't part of the library.)
4521
4522
4523 Version 3.8 18-Dec-01
4524 ---------------------
4525
4526 1. The experimental UTF-8 code was completely screwed up. It was packing the
4527 bytes in the wrong order. How dumb can you get?
4528
4529
4530 Version 3.7 29-Oct-01
4531 ---------------------
4532
4533 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4534 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4535 this didn't happen under Solaris 8, where I normally test things.
4536
4537 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4538 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4539 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4540 this fix an hour or so after the initial 3.7 release.)
4541
4542
4543 Version 3.6 23-Oct-01
4544 ---------------------
4545
4546 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4547 offsets passed as NULL with zero offset count.
4548
4549 2. The config.guess and config.sub files had not been updated when I moved to
4550 the latest autoconf.
4551
4552
4553 Version 3.5 15-Aug-01
4554 ---------------------
4555
4556 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4557 had been forgotten.
4558
4559 2. By using declared but undefined structures, we can avoid using "void"
4560 definitions in pcre.h while keeping the internal definitions of the structures
4561 private.
4562
4563 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4564 user point of view, this means that both static and shared libraries are built
4565 by default, but this can be individually controlled. More of the work of
4566 handling this static/shared cases is now inside libtool instead of PCRE's make
4567 file.
4568
4569 4. The pcretest utility is now installed along with pcregrep because it is
4570 useful for users (to test regexs) and by doing this, it automatically gets
4571 relinked by libtool. The documentation has been turned into a man page, so
4572 there are now .1, .txt, and .html versions in /doc.
4573
4574 5. Upgrades to pcregrep:
4575 (i) Added long-form option names like gnu grep.
4576 (ii) Added --help to list all options with an explanatory phrase.
4577 (iii) Added -r, --recursive to recurse into sub-directories.
4578 (iv) Added -f, --file to read patterns from a file.
4579
4580 6. pcre_exec() was referring to its "code" argument before testing that
4581 argument for NULL (and giving an error if it was NULL).
4582
4583 7. Upgraded Makefile.in to allow for compiling in a different directory from
4584 the source directory.
4585
4586 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4587 options bits, the pointer it was passed was to an int instead of to an unsigned
4588 long int. This mattered only on 64-bit systems.
4589
4590 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4591 generated) instead of pcre.in, which it its source. Also made the same change
4592 in several of the .c files.
4593
4594 10. A new release of gcc defines printf() as a macro, which broke pcretest
4595 because it had an ifdef in the middle of a string argument for printf(). Fixed
4596 by using separate calls to printf().
4597
4598 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4599 script, to force use of CR or LF instead of \n in the source. On non-Unix
4600 systems, the value can be set in config.h.
4601
4602 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4603 absolute limit. Changed the text of the error message to make this clear, and
4604 likewise updated the man page.
4605
4606 13. The limit of 99 on the number of capturing subpatterns has been removed.
4607 The new limit is 65535, which I hope will not be a "real" limit.
4608
4609
4610 Version 3.4 22-Aug-00
4611 ---------------------
4612
4613 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4614
4615 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4616
4617
4618 Version 3.3 01-Aug-00
4619 ---------------------
4620
4621 1. If an octal character was given, but the value was greater than \377, it
4622 was not getting masked to the least significant bits, as documented. This could
4623 lead to crashes in some systems.
4624
4625 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4626 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4627
4628 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4629 These just pass their arguments on to (pcre_free)(), but they are provided
4630 because some uses of PCRE bind it to non-C systems that can call its functions,
4631 but cannot call free() or pcre_free() directly.
4632
4633 4. Add "make test" as a synonym for "make check". Corrected some comments in
4634 the Makefile.
4635
4636 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4637 Makefile.
4638
4639 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4640 command called pgrep for grepping around the active processes.
4641
4642 7. Added the beginnings of support for UTF-8 character strings.
4643
4644 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4645 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4646 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4647 out for the ar command.)
4648
4649
4650 Version 3.2 12-May-00
4651 ---------------------
4652
4653 This is purely a bug fixing release.
4654
4655 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4656 of ZA. This was just one example of several cases that could provoke this bug,
4657 which was introduced by change 9 of version 2.00. The code for breaking
4658 infinite loops after an iteration that matches an empty string was't working
4659 correctly.
4660
4661 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4662 when matched against abbab (for example). After matching an empty string, it
4663 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4664 caused it to match further down the string than it should.
4665
4666 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4667 was there because it doesn't seem to be needed, and it causes trouble on some
4668 systems, as it is not a Standard C header. It has been removed.
4669
4670 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4671 were reported on the Macintosh. The changes were from
4672
4673 while ((c = *(++ptr)) != 0 && c != '\n');
4674 to
4675 while ((c = *(++ptr)) != 0 && c != '\n') ;
4676
4677 Totally extraordinary, but if that's what it takes...
4678
4679 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4680 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4681 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4682 assumes the way PCRE uses memmove() (always moving upwards).
4683
4684 6. PCRE is being used in one environment where strchr() is not available. There
4685 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4686 faster code anyway.
4687
4688
4689 Version 3.1 09-Feb-00
4690 ---------------------
4691
4692 The only change in this release is the fixing of some bugs in Makefile.in for
4693 the "install" target:
4694
4695 (1) It was failing to install pcreposix.h.
4696
4697 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4698
4699
4700 Version 3.0 01-Feb-00
4701 ---------------------
4702
4703 1. Add support for the /+ modifier to perltest (to output $` like it does in
4704 pcretest).
4705
4706 2. Add support for the /g modifier to perltest.
4707
4708 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4709 matches null strings.
4710
4711 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4712 pattern. Perl treats empty patterns specially - it reuses the most recent
4713 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4714 effect.
4715
4716 5. The POSIX interface was broken in that it was just handing over the POSIX
4717 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4718 required a bigger vector, with some working space on the end. This means that
4719 the POSIX wrapper now has to get and free some memory, and copy the results.
4720
4721 6. Added some simple autoconf support, placing the test data and the
4722 documentation in separate directories, re-organizing some of the
4723 information files, and making it build pcre-config (a GNU standard). Also added
4724 libtool support for building PCRE as a shared library, which is now the
4725 default.
4726
4727 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4728 09 are not valid octal constants. Single digits will be used for minor values
4729 less than 10.
4730
4731 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4732 existing programs that set these in the POSIX interface can use PCRE without
4733 modification.
4734
4735 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4736 return all that pcre_info() returns, plus additional data. The pcre_info()
4737 function is retained for compatibility, but is considered to be obsolete.
4738
4739 10. Added experimental recursion feature (?R) to handle one common case that
4740 Perl 5.6 will be able to do with (?p{...}).
4741
4742 11. Added support for POSIX character classes like [:alpha:], which Perl is
4743 adopting.
4744
4745
4746 Version 2.08 31-Aug-99
4747 ----------------------
4748
4749 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4750 trying to match at the startoffset position, but instead was moving forward to
4751 the next newline as if a previous match had failed.
4752
4753 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4754 and could get into a loop if a null string was matched other than at the start
4755 of the subject.
4756
4757 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4758 be distinguished at compile time, and for completeness also added PCRE_DATE.
4759
4760 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4761 in GnuWin32 environments.
4762
4763
4764 Version 2.07 29-Jul-99
4765 ----------------------
4766
4767 1. The documentation is now supplied in plain text form and HTML as well as in
4768 the form of man page sources.
4769
4770 2. C++ compilers don't like assigning (void *) values to other pointer types.
4771 In particular this affects malloc(). Although there is no problem in Standard
4772 C, I've put in casts to keep C++ compilers happy.
4773
4774 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4775 should be (const char *).
4776
4777 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4778 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4779 However, I haven't made this a standard facility. The documentation doesn't
4780 mention it, and the Makefile doesn't support it.
4781
4782 5. The Makefile now contains an "install" target, with editable destinations at
4783 the top of the file. The pcretest program is not installed.
4784
4785 6. pgrep -V now gives the PCRE version number and date.
4786
4787 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4788 causing the entire string to be ignored, instead of just the last character.
4789
4790 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4791 non-matching string, it can take a very, very long time, even for strings of
4792 quite modest length, because of the nested recursion. PCRE now does better in
4793 some of these cases. It does this by remembering the last required literal
4794 character in the pattern, and pre-searching the subject to ensure it is present
4795 before running the real match. In other words, it applies a heuristic to detect
4796 some types of certain failure quickly, and in the above example, if presented
4797 with a string that has no trailing " it gives "no match" very quickly.
4798
4799 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4800 other alternatives are tried instead.
4801
4802
4803 Version 2.06 09-Jun-99
4804 ----------------------
4805
4806 1. Change pcretest's output for amount of store used to show just the code
4807 space, because the remainder (the data block) varies in size between 32-bit and
4808 64-bit systems.
4809
4810 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4811 start matching at. This allows lookbehinds to work when searching for multiple
4812 occurrences in a string.
4813
4814 3. Added additional options to pcretest for testing multiple occurrences:
4815
4816 /+ outputs the rest of the string that follows a match
4817 /g loops for multiple occurrences, using the new startoffset argument
4818 /G loops for multiple occurrences by passing an incremented pointer
4819
4820 4. PCRE wasn't doing the "first character" optimization for patterns starting
4821 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4822 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4823 the letter 'x'. On long subject strings, this gives a significant speed-up.
4824
4825
4826 Version 2.05 21-Apr-99
4827 ----------------------
4828
4829 1. Changed the type of magic_number from int to long int so that it works
4830 properly on 16-bit systems.
4831
4832 2. Fixed a bug which caused patterns starting with .* not to work correctly
4833 when the subject string contained newline characters. PCRE was assuming
4834 anchoring for such patterns in all cases, which is not correct because .* will
4835 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4836 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4837 must be retried after every newline in the subject.
4838
4839
4840 Version 2.04 18-Feb-99
4841 ----------------------
4842
4843 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4844 computation of the store needed to hold the pattern was incorrect (too large).
4845 If such patterns were nested a few deep, this could multiply and become a real
4846 problem.
4847
4848 2. Added /M option to pcretest to show the memory requirement of a specific
4849 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4850
4851 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4852 compiled in such a way that the backtracking after subsequent failure was
4853 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4854 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4855
4856
4857 Version 2.03 02-Feb-99
4858 ----------------------
4859
4860 1. Fixed typo and small mistake in man page.
4861
4862 2. Added 4th condition (GPL supersedes if conflict) and created separate
4863 LICENCE file containing the conditions.
4864
4865 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4866 Perl, that is the internal \ allows the delimiter to be included in the
4867 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4868 the final delimiter, add \ to the end of the pattern (to test the error).
4869
4870 4. Added the convenience functions for extracting substrings after a successful
4871 match. Updated pcretest to make it able to test these functions.
4872
4873
4874 Version 2.02 14-Jan-99
4875 ----------------------
4876
4877 1. Initialized the working variables associated with each extraction so that
4878 their saving and restoring doesn't refer to uninitialized store.
4879
4880 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4881 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4882 fix the problem.
4883
4884 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4885 calls, and wasn't printing the correct value for compiling calls. Increased the
4886 default value of LOOPREPEAT, and the number of significant figures in the
4887 times.
4888
4889 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4890
4891 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4892 a building problem on Windows NT with a FAT file system.
4893
4894
4895 Version 2.01 21-Oct-98
4896 ----------------------
4897
4898 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4899 to character tables built by pcre_maketables() in the current locale. If NULL
4900 is passed, the default tables are used.
4901
4902
4903 Version 2.00 24-Sep-98
4904 ----------------------
4905
4906 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4907 it any more.
4908
4909 2. Allow quantification of (?>) groups, and make it work correctly.
4910
4911 3. The first character computation wasn't working for (?>) groups.
4912
4913 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4914 end of the subject) and add 5.005's \z, which really does match only at the
4915 very end of the subject.
4916
4917 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4918
4919 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4920 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4921 localized options. All options to pcre_study() were also removed.
4922
4923 7. Add other new features from 5.005:
4924
4925 $(?<= positive lookbehind
4926 $(?<! negative lookbehind
4927 (?imsx-imsx) added the unsetting capability
4928 such a setting is global if at outer level; local otherwise
4929 (?imsx-imsx:) non-capturing groups with option setting
4930 (?(cond)re|re) conditional pattern matching
4931
4932 A backreference to itself in a repeated group matches the previous
4933 captured string.
4934
4935 8. General tidying up of studying (both automatic and via "study")
4936 consequential on the addition of new assertions.
4937
4938 9. As in 5.005, unlimited repeated groups that could match an empty substring
4939 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4940 runtime if any iteration does actually match an empty substring.
4941
4942 10. Include the RunTest script in the distribution.
4943
4944 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4945 discrepancies, some of which were old and were also with respect to 5.004. They
4946 have now been fixed.
4947
4948
4949 Version 1.09 28-Apr-98
4950 ----------------------
4951
4952 1. A negated single character class followed by a quantifier with a minimum
4953 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4954 program crashes, or just wrong answers. This did not apply to negated classes
4955 containing more than one character, or to minima other than one.
4956
4957
4958 Version 1.08 27-Mar-98
4959 ----------------------
4960
4961 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4962
4963 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4964 latter must appear before anything that relies on it in the pattern.
4965
4966
4967 Version 1.07 16-Feb-98
4968 ----------------------
4969
4970 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4971 repeat of a potentially empty string).
4972
4973
4974 Version 1.06 23-Jan-98
4975 ----------------------
4976
4977 1. Added Markus Oberhumer's little patches for C++.
4978
4979 2. Literal strings longer than 255 characters were broken.
4980
4981
4982 Version 1.05 23-Dec-97
4983 ----------------------
4984
4985 1. Negated character classes containing more than one character were failing if
4986 PCRE_CASELESS was set at run time.
4987
4988
4989 Version 1.04 19-Dec-97
4990 ----------------------
4991
4992 1. Corrected the man page, where some "const" qualifiers had been omitted.
4993
4994 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4995 input syntax.
4996
4997 3. Fixed memory leak which occurred when a regex with back references was
4998 matched with an offsets vector that wasn't big enough. The temporary memory
4999 that is used in this case wasn't being freed if the match failed.
5000
5001 4. Tidied pcretest to ensure it frees memory that it gets.
5002
5003 5. Temporary memory was being obtained in the case where the passed offsets
5004 vector was exactly big enough.
5005
5006 6. Corrected definition of offsetof() from change 5 below.
5007
5008 7. I had screwed up change 6 below and broken the rules for the use of
5009 setjmp(). Now fixed.
5010
5011
5012 Version 1.03 18-Dec-97
5013 ----------------------
5014
5015 1. A erroneous regex with a missing opening parenthesis was correctly
5016 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5017 on some systems.
5018
5019 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5020 it was reported that one broken compiler failed on the former because "code" is
5021 also an independent variable.
5022
5023 3. The erroneous regex a[]b caused an array overrun reference.
5024
5025 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5026 fail on data ending with that character. (It was going on too far, and checking
5027 the next character, typically a binary zero.) This was specific to the
5028 optimized code for single-character negative classes.
5029
5030 5. Added a contributed patch from the TIN world which does the following:
5031
5032 + Add an undef for memmove, in case the the system defines a macro for it.
5033
5034 + Add a definition of offsetof(), in case there isn't one. (I don't know
5035 the reason behind this - offsetof() is part of the ANSI standard - but
5036 it does no harm).
5037
5038 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5039 most of the places where whitespace preceded '#'. I have given up and
5040 allowed the remaining 2 cases to be at the margin.
5041
5042 + Rename some variables in pcre to eliminate shadowing. This seems very
5043 pedantic, but does no harm, of course.
5044
5045 6. Moved the call to setjmp() into its own function, to get rid of warnings
5046 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5047
5048 7. Constructs such as \d{8,} were compiling into the equivalent of
5049 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5050 outcome, but in this particular case used more store than had been allocated,
5051 which caused the bug to be discovered because it threw up an internal error.
5052
5053 8. The debugging code in both pcre and pcretest for outputting the compiled
5054 form of a regex was going wrong in the case of back references followed by
5055 curly-bracketed repeats.
5056
5057
5058 Version 1.02 12-Dec-97
5059 ----------------------
5060
5061 1. Typos in pcre.3 and comments in the source fixed.
5062
5063 2. Applied a contributed patch to get rid of places where it used to remove
5064 'const' from variables, and fixed some signed/unsigned and uninitialized
5065 variable warnings.
5066
5067 3. Added the "runtest" target to Makefile.
5068
5069 4. Set default compiler flag to -O2 rather than just -O.
5070
5071
5072 Version 1.01 19-Nov-97
5073 ----------------------
5074
5075 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5076 like /([ab]*)*/, that is, for classes with more than one character in them.
5077
5078 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5079 as /((?>a*))*/ (a PCRE_EXTRA facility).
5080
5081
5082 Version 1.00 18-Nov-97
5083 ----------------------
5084
5085 1. Added compile-time macros to support systems such as SunOS4 which don't have
5086 memmove() or strerror() but have other things that can be used instead.
5087
5088 2. Arranged that "make clean" removes the executables.
5089
5090
5091 Version 0.99 27-Oct-97
5092 ----------------------
5093
5094 1. Fixed bug in code for optimizing classes with only one character. It was
5095 initializing a 32-byte map regardless, which could cause it to run off the end
5096 of the memory it had got.
5097
5098 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5099
5100
5101 Version 0.98 22-Oct-97
5102 ----------------------
5103
5104 1. Fixed bug in code for handling temporary memory usage when there are more
5105 back references than supplied space in the ovector. This could cause segfaults.
5106
5107
5108 Version 0.97 21-Oct-97
5109 ----------------------
5110
5111 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5112
5113 2. Optimized negated single characters not to use a bit map.
5114
5115 3. Brought error texts together as macro definitions; clarified some of them;
5116 fixed one that was wrong - it said "range out of order" when it meant "invalid
5117 escape sequence".
5118
5119 4. Changed some char * arguments to const char *.
5120
5121 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5122
5123 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5124 pcretest.
5125
5126
5127 Version 0.96 16-Oct-97
5128 ----------------------
5129
5130 1. Added a simple "pgrep" utility to the distribution.
5131
5132 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5133 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5134 where "ddd" means "one or more decimal digits".
5135
5136 3. Fixed serious bug. If a pattern had a back reference, but the call to
5137 pcre_exec() didn't supply a large enough ovector to record the related
5138 identifying subpattern, the match always failed. PCRE now remembers the number
5139 of the largest back reference, and gets some temporary memory in which to save
5140 the offsets during matching if necessary, in order to ensure that
5141 backreferences always work.
5142
5143 4. Increased the compatibility with Perl in a number of ways:
5144
5145 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5146 to request this handling. The option can be set at compile or exec time.
5147
5148 (b) $ matches before a terminating newline by default; an option
5149 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5150 mode). The option can be set at compile or exec time.
5151
5152 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5153 the same as Perl's. If the decimal number it represents is less than 10
5154 or there aren't that many previous left capturing parentheses, an octal
5155 escape is read. Inside a character class, it's always an octal escape,
5156 even if it is a single digit.
5157
5158 (d) An escaped but undefined alphabetic character is taken as a literal,
5159 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5160 escapes.
5161
5162 (e) {0} is now permitted. (The previous item is removed from the compiled
5163 pattern).
5164
5165 5. Changed all the names of code files so that the basic parts are no longer
5166 than 10 characters, and abolished the teeny "globals.c" file.
5167
5168 6. Changed the handling of character classes; they are now done with a 32-byte
5169 bit map always.
5170
5171 7. Added the -d and /D options to pcretest to make it possible to look at the
5172 internals of compilation without having to recompile pcre.
5173
5174
5175 Version 0.95 23-Sep-97
5176 ----------------------
5177
5178 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5179 \x20 at the start of a run of normal characters. These were being treated as
5180 real characters, instead of the source characters being re-checked.
5181
5182
5183 Version 0.94 18-Sep-97
5184 ----------------------
5185
5186 1. The functions are now thread-safe, with the caveat that the global variables
5187 containing pointers to malloc() and free() or alternative functions are the
5188 same for all threads.
5189
5190 2. Get pcre_study() to generate a bitmap of initial characters for non-
5191 anchored patterns when this is possible, and use it if passed to pcre_exec().
5192
5193
5194 Version 0.93 15-Sep-97
5195 ----------------------
5196
5197 1. /(b)|(:+)/ was computing an incorrect first character.
5198
5199 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5200 but not actually doing anything yet.
5201
5202 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5203 as Perl does (e.g. [-az] or [az-]).
5204
5205 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5206 all possible positions.
5207
5208 5. Split up into different modules to avoid including unneeded functions in a
5209 compiled binary. However, compile and exec are still in one module. The "study"
5210 function is split off.
5211
5212 6. The character tables are now in a separate module whose source is generated
5213 by an auxiliary program - but can then be edited by hand if required. There are
5214 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5215 toupper() in the code.
5216
5217 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5218 make them global. Abolish the function for setting them, as the caller can now
5219 set them directly.
5220
5221
5222 Version 0.92 11-Sep-97
5223 ----------------------
5224
5225 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5226 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5227
5228 2. Caseless matching was not working in character classes if the characters in
5229 the pattern were in upper case.
5230
5231 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5232
5233 4. Make PCRE_ANCHORED public and accept as a compile option.
5234
5235 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5236 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5237 pass them.
5238
5239 6. Give an error if bad option bits passed at compile or run time.
5240
5241 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5242 pcretest to cause it to pass that flag.
5243
5244 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5245 options, and the first character, if set.
5246
5247 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5248
5249
5250 Version 0.91 10-Sep-97
5251 ----------------------
5252
5253 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5254 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5255
5256 2. PCRE was looping on encountering an indefinitely repeated back reference to
5257 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5258 Perl does - treats the match as successful.
5259
5260 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5