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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1345 - (show annotations)
Tue Jul 2 17:52:28 2013 UTC (6 years, 1 month ago) by ph10
File size: 248191 byte(s)
Small performance improvement in strlen16 and strlen32 in pcretest.
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
22 Version 8.33 28-May-2013
23 --------------------------
24
25 1. Added 'U' to some constants that are compared to unsigned integers, to
26 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
27 variables that are added to signed variables, to ensure the result is
28 signed and can be negated.
29
30 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
31 CMake config files.
32
33 3. Revise the creation of config.h.generic so that all boolean macros are
34 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
35 overriding via -D on the command line possible.
36
37 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
38 to unsigned int is reported to make a quite noticeable speed difference in
39 a specific Windows environment. Testing on Linux did also appear to show
40 some benefit (and it is clearly not harmful). Also fixed the definition of
41 Xop which should be unsigned.
42
43 5. Related to (4), changing the definition of the intermediate variable cc
44 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
45 improvements.
46
47 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
48 unnecessary spaces.
49
50 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
51 and later.
52
53 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
54
55 9. Optimizing fast_forward_start_bits in JIT.
56
57 10. Adding support for callouts in JIT, and fixing some issues revealed
58 during this work. Namely:
59
60 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
61
62 (b) Minimum length was not checked before the matching is started.
63
64 11. The value of capture_last that is passed to callouts was incorrect in some
65 cases when there was a capture on one path that was subsequently abandoned
66 after a backtrack. Also, the capture_last value is now reset after a
67 recursion, since all captures are also reset in this case.
68
69 12. The interpreter no longer returns the "too many substrings" error in the
70 case when an overflowing capture is in a branch that is subsequently
71 abandoned after a backtrack.
72
73 13. In the pathological case when an offset vector of size 2 is used, pcretest
74 now prints out the matched string after a yield of 0 or 1.
75
76 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
77 Only supported by the JIT compiler at the moment.
78
79 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
80
81 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
82 offset of the starting point of the matching process, provided the offsets
83 vector is large enough.
84
85 17. The \A escape now records a lookbehind value of 1, though its execution
86 does not actually inspect the previous character. This is to ensure that,
87 in partial multi-segment matching, at least one character from the old
88 segment is retained when a new segment is processed. Otherwise, if there
89 are no lookbehinds in the pattern, \A might match incorrectly at the start
90 of a new segment.
91
92 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
93
94 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
95 gives some modest performance improvement in 8-bit mode.
96
97 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
98 be expressed in certain programming languages using Universal Character
99 Names.
100
101 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
102 which points out that "non characters" are not "characters that may not
103 appear in Unicode strings" but rather "characters that are reserved for
104 internal use and have only local meaning".
105
106 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
107 there was a conditional group that depended on an assertion, if the
108 assertion was false, the callout that immediately followed the alternation
109 in the condition was skipped when pcre_exec() was used for matching.
110
111 23. Allow an explicit callout to be inserted before an assertion that is the
112 condition for a conditional group, for compatibility with automatic
113 callouts, which always insert a callout at this point.
114
115 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
116 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
117
118 25. (*PRUNE) is now supported by the JIT compiler.
119
120 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
121
122 27. Fix the case where there are two or more SKIPs with arguments that may be
123 ignored.
124
125 28. (*SKIP) is now supported by the JIT compiler.
126
127 29. (*THEN) is now supported by the JIT compiler.
128
129 30. Update RunTest with additional test selector options.
130
131 31. The way PCRE handles backtracking verbs has been changed in two ways.
132
133 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
134 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
135 backtracking. In some cases this makes it more Perl-compatible, but Perl's
136 rather obscure rules do not always do the same thing.
137
138 (2) Previously, backtracking verbs were confined within assertions. This is
139 no longer the case for positive assertions, except for (*ACCEPT). Again,
140 this sometimes improves Perl compatibility, and sometimes does not.
141
142 32. A number of tests that were in test 2 because Perl did things differently
143 have been moved to test 1, because either Perl or PCRE has changed, and
144 these tests are now compatible.
145
146 32. Backtracking control verbs are now handled in the same way in JIT and
147 interpreter.
148
149 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
150 contained a forward subroutine reference caused a compile error.
151
152 34. Auto-detect and optimize limited repetitions in JIT.
153
154 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
155 blocking (*UTF) etc.
156
157 36. In the interpreter, maximizing pattern repetitions for characters and
158 character types now use tail recursion, which reduces stack usage.
159
160 37. The value of the max lookbehind was not correctly preserved if a compiled
161 and saved regex was reloaded on a host of different endianness.
162
163 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
164 of the compiled pattern block, expand the flags field from 16 to 32 bits
165 because it was almost full.
166
167 39. Try madvise first before posix_madvise.
168
169 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
170 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
171 original change locked out all matches of empty strings. This has been
172 changed so that one match of an empty string per line is recognized.
173 Subsequent searches on the same line (for colouring or for --only-matching,
174 for example) do not recognize empty strings.
175
176 41. Applied a user patch to fix a number of spelling mistakes in comments.
177
178 42. Data lines longer than 65536 caused pcretest to crash.
179
180 43. Clarified the data type for length and startoffset arguments for pcre_exec
181 and pcre_dfa_exec in the function-specific man pages, where they were
182 explicitly stated to be in bytes, never having been updated. I also added
183 some clarification to the pcreapi man page.
184
185 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
186 a segmentation fault.
187
188
189 Version 8.32 30-November-2012
190 -----------------------------
191
192 1. Improved JIT compiler optimizations for first character search and single
193 character iterators.
194
195 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
196 Patch by Daniel Richard G.
197
198 3. Single character iterator optimizations in the JIT compiler.
199
200 4. Improved JIT compiler optimizations for character ranges.
201
202 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
203 Reported by Giuseppe D'Angelo.
204
205 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
206 of a line, was being set incorrectly in cases where .* appeared inside
207 atomic brackets at the start of a pattern, or where there was a subsequent
208 *PRUNE or *SKIP.
209
210 7. Improved instruction cache flush for POWER/PowerPC.
211 Patch by Daniel Richard G.
212
213 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
214 grep:
215
216 (a) There is now no limit to the number of patterns to be matched.
217
218 (b) An error is given if a pattern is too long.
219
220 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
221 are now supported.
222
223 (d) --exclude-from and --include-from (multiple use) have been added.
224
225 (e) Exclusions and inclusions now apply to all files and directories, not
226 just to those obtained from scanning a directory recursively.
227
228 (f) Multiple uses of -f and --file-list are now supported.
229
230 (g) In a Windows environment, the default for -d has been changed from
231 "read" (the GNU grep default) to "skip", because otherwise the presence
232 of a directory in the file list provokes an error.
233
234 (h) The documentation has been revised and clarified in places.
235
236 9. Improve the matching speed of capturing brackets.
237
238 10. Changed the meaning of \X so that it now matches a Unicode extended
239 grapheme cluster.
240
241 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
242 out POSIX threads when JIT support is configured.
243
244 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
245
246 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
247 structure could go wrong in environments where size_t is not the same size
248 as int.
249
250 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
251 set.
252
253 15. The EBCDIC support had decayed; later updates to the code had included
254 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
255 general tidy up of EBCDIC-related issues, and the documentation was also
256 not quite right. There is now a test that can be run on ASCII systems to
257 check some of the EBCDIC-related things (but is it not a full test).
258
259 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
260 in a small tidy to the code.
261
262 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
263
264 18. If the --only-matching (-o) option in pcregrep is specified multiple
265 times, each one causes appropriate output. For example, -o1 -o2 outputs the
266 substrings matched by the 1st and 2nd capturing parentheses. A separating
267 string can be specified by --om-separator (default empty).
268
269 19. Improving the first n character searches.
270
271 20. Turn case lists for horizontal and vertical white space into macros so that
272 they are defined only once.
273
274 21. This set of changes together give more compatible Unicode case-folding
275 behaviour for characters that have more than one other case when UCP
276 support is available.
277
278 (a) The Unicode property table now has offsets into a new table of sets of
279 three or more characters that are case-equivalent. The MultiStage2.py
280 script that generates these tables (the pcre_ucd.c file) now scans
281 CaseFolding.txt instead of UnicodeData.txt for character case
282 information.
283
284 (b) The code for adding characters or ranges of characters to a character
285 class has been abstracted into a generalized function that also handles
286 case-independence. In UTF-mode with UCP support, this uses the new data
287 to handle characters with more than one other case.
288
289 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
290 whose other case is greater than 256 are now correctly matched
291 caselessly. Previously, the high codepoint matched the low one, but not
292 vice versa.
293
294 (d) The processing of \h, \H, \v, and \ in character classes now makes use
295 of the new class addition function, using character lists defined as
296 macros alongside the case definitions of 20 above.
297
298 (e) Caseless back references now work with characters that have more than
299 one other case.
300
301 (f) General caseless matching of characters with more than one other case
302 is supported.
303
304 22. Unicode character properties were updated from Unicode 6.2.0
305
306 23. Improved CMake support under Windows. Patch by Daniel Richard G.
307
308 24. Add support for 32-bit character strings, and UTF-32
309
310 25. Major JIT compiler update (code refactoring and bugfixing).
311 Experimental Sparc 32 support is added.
312
313 26. Applied a modified version of Daniel Richard G's patch to create
314 pcre.h.generic and config.h.generic by "make" instead of in the
315 PrepareRelease script.
316
317 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
318 pcre_compile.c when checking for a zero character.
319
320 28. Introducing a native interface for JIT. Through this interface, the compiled
321 machine code can be directly executed. The purpose of this interface is to
322 provide fast pattern matching, so several sanity checks are not performed.
323 However, feature tests are still performed. The new interface provides
324 1.4x speedup compared to the old one.
325
326 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
327 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
328 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
329
330 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
331 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
332 the "old" RFC 2279). Instead, it ended up passing a negative length to
333 pcre_exec().
334
335 31. Add support for GCC's visibility feature to hide internal functions.
336
337 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
338 "unknown -C option" after outputting 0 or 1.
339
340 33. There is now support for generating a code coverage report for the test
341 suite in environments where gcc is the compiler and lcov is installed. This
342 is mainly for the benefit of the developers.
343
344 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
345 unaddressable using valgrind annotations, allowing valgrind to detect
346 invalid memory accesses. This is mainly for the benefit of the developers.
347
348 25. (*UTF) can now be used to start a pattern in any of the three libraries.
349
350 26. Give configure error if --enable-cpp but no C++ compiler found.
351
352
353 Version 8.31 06-July-2012
354 -------------------------
355
356 1. Fixing a wrong JIT test case and some compiler warnings.
357
358 2. Removed a bashism from the RunTest script.
359
360 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
361 to unsigned type, result still unsigned" that was given by an MS compiler
362 on encountering the code "-sizeof(xxx)".
363
364 4. Partial matching support is added to the JIT compiler.
365
366 5. Fixed several bugs concerned with partial matching of items that consist
367 of more than one character:
368
369 (a) /^(..)\1/ did not partially match "aba" because checking references was
370 done on an "all or nothing" basis. This also applied to repeated
371 references.
372
373 (b) \R did not give a hard partial match if \r was found at the end of the
374 subject.
375
376 (c) \X did not give a hard partial match after matching one or more
377 characters at the end of the subject.
378
379 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
380 a partial match for the string "\r".
381
382 (e) When newline was set to CRLF, the metacharacter "." did not recognize
383 a partial match for a CR character at the end of the subject string.
384
385 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
386 running pcretest, the text "(JIT)" added to the output whenever JIT is
387 actually used to run the match.
388
389 7. Individual JIT compile options can be set in pcretest by following -s+[+]
390 or /S+[+] with a digit between 1 and 7.
391
392 8. OP_NOT now supports any UTF character not just single-byte ones.
393
394 9. (*MARK) control verb is now supported by the JIT compiler.
395
396 10. The command "./RunTest list" lists the available tests without actually
397 running any of them. (Because I keep forgetting what they all are.)
398
399 11. Add PCRE_INFO_MAXLOOKBEHIND.
400
401 12. Applied a (slightly modified) user-supplied patch that improves performance
402 when the heap is used for recursion (compiled with --disable-stack-for-
403 recursion). Instead of malloc and free for each heap frame each time a
404 logical recursion happens, frames are retained on a chain and re-used where
405 possible. This sometimes gives as much as 30% improvement.
406
407 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
408 call.
409
410 14. As documented, (*COMMIT) is now confined to within a positive assertion.
411
412 15. It is now possible to link pcretest with libedit as an alternative to
413 libreadline.
414
415 16. (*COMMIT) control verb is now supported by the JIT compiler.
416
417 17. The Unicode data tables have been updated to Unicode 6.1.0.
418
419 18. Added --file-list option to pcregrep.
420
421 19. Added binary file support to pcregrep, including the -a, --binary-files,
422 -I, and --text options.
423
424 20. The madvise function is renamed for posix_madvise for QNX compatibility
425 reasons. Fixed by Giuseppe D'Angelo.
426
427 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
428
429 22. Changed the option for creating symbolic links for 16-bit man pages from
430 -s to -sf so that re-installing does not cause issues.
431
432 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
433
434 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
435 matches in certain environments (the workspace was not being correctly
436 retained). Also added to pcre_dfa_exec() a simple plausibility check on
437 some of the workspace data at the beginning of a restart.
438
439 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
440 was not doing so when it should - probably a typo introduced by SVN 528
441 (change 8.10/14).
442
443 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
444 \w+ when the character tables indicated that \x{c4} was a word character.
445 There were several related cases, all because the tests for doing a table
446 lookup were testing for characters less than 127 instead of 255.
447
448 27. If a pattern contains capturing parentheses that are not used in a match,
449 their slots in the ovector are set to -1. For those that are higher than
450 any matched groups, this happens at the end of processing. In the case when
451 there were back references that the ovector was too small to contain
452 (causing temporary malloc'd memory to be used during matching), and the
453 highest capturing number was not used, memory off the end of the ovector
454 was incorrectly being set to -1. (It was using the size of the temporary
455 memory instead of the true size.)
456
457 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
458 ovector size, it uses memory at the end of the block that it has got.
459
460 29. Check for an overlong MARK name and give an error at compile time. The
461 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
462
463 30. JIT compiler update.
464
465 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
466 Rill for the patch.
467
468 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
469
470 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
471
472 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
473 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
474 was enabled.
475
476 35. Improve JIT code generation for greedy plus quantifier.
477
478 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
479 1 to "aa" instead of to an empty string. The bug affected repeated groups
480 that could potentially match an empty string.
481
482 37. Optimizing single character iterators in JIT.
483
484 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
485 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
486 codepoints that are too big for the mode are faulted, and in a UTF mode,
487 disallowed codepoints are also faulted.
488
489 39. If PCRE was compiled with UTF support, in three places in the DFA
490 matcher there was code that should only have been obeyed in UTF mode, but
491 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
492 processing when bytes with values greater than 127 were present. In 16-bit
493 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
494 both cases the values are those that cannot be the first data item in a UTF
495 character. The three items that might have provoked this were recursions,
496 possessively repeated groups, and atomic groups.
497
498 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
499 and pcregrep, because some OS require shared objects to be explicitly
500 passed to ld, causing the link step to fail if they are not.
501
502 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
503 mode, patterns that started with \h* or \R* might be incorrectly matched.
504
505
506 Version 8.30 04-February-2012
507 -----------------------------
508
509 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
510 name is defined in ctype.h.
511
512 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
513 only in quite long subpatterns.
514
515 3. Removed the function pcre_info(), which has been obsolete and deprecated
516 since it was replaced by pcre_fullinfo() in February 2000.
517
518 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
519 match a (*MARK), and the match failed at the start of the subject, a
520 reference to memory before the start of the subject could occur. This bug
521 was introduced by fix 17 of release 8.21.
522
523 5. A reference to an unset group with zero minimum repetition was giving
524 totally wrong answers (in non-JavaScript-compatibility mode). For example,
525 /(another)?(\1?)test/ matched against "hello world test". This bug was
526 introduced in release 8.13.
527
528 6. Add support for 16-bit character strings (a large amount of work involving
529 many changes and refactorings).
530
531 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
532 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
533 from a file.
534
535 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
536 rounding is not applied in this particular case).
537
538 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
539 if they appear, or are escaped, in patterns.
540
541 10. Get rid of a number of -Wunused-but-set-variable warnings.
542
543 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
544 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
545 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
546 also returns the mark "x". This bug applied to capturing parentheses,
547 non-capturing parentheses, and atomic parentheses. It also applied to some
548 assertions.
549
550 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
551 information out of configure.ac instead of relying on pcre.h.generic, which
552 is not stored in the repository.
553
554 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
555 -lreadline.
556
557 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
558
559 15. Applied Graycode's patch to put the top-level frame on the stack rather
560 than the heap when not using the stack for recursion. This gives a
561 performance improvement in many cases when recursion is not deep.
562
563 16. Experimental code added to "pcretest -C" to output the stack frame size.
564
565
566 Version 8.21 12-Dec-2011
567 ------------------------
568
569 1. Updating the JIT compiler.
570
571 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
572 are added as well.
573
574 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
575 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
576 calling _pcre_jit_exec. Some extra comments are added.
577
578 4. (*MARK) settings inside atomic groups that do not contain any capturing
579 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
580 was introduced by change 18 for 8.20.
581
582 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
583 ECMA-262 standard.
584
585 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
586 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
587 This bug was probably introduced by change 9 of 8.13.
588
589 7. While fixing 6 above, I noticed that a number of other items were being
590 incorrectly rejected as "not fixed length". This arose partly because newer
591 opcodes had not been added to the fixed-length checking code. I have (a)
592 corrected the bug and added tests for these items, and (b) arranged for an
593 error to occur if an unknown opcode is encountered while checking for fixed
594 length instead of just assuming "not fixed length". The items that were
595 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
596 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
597 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
598
599 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
600 being incorrectly compiled and would have given unpredicatble results.
601
602 9. A possessively repeated subpattern with minimum repeat count greater than
603 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
604 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
605 the first (A) could occur when it should not.
606
607 10. Add a cast and remove a redundant test from the code.
608
609 11. JIT should use pcre_malloc/pcre_free for allocation.
610
611 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
612 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
613 is anything other than /usr, -L is still shown).
614
615 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
616
617 14. Perl does not support \N without a following name in a [] class; PCRE now
618 also gives an error.
619
620 15. If a forward reference was repeated with an upper limit of around 2000,
621 it caused the error "internal error: overran compiling workspace". The
622 maximum number of forward references (including repeats) was limited by the
623 internal workspace, and dependent on the LINK_SIZE. The code has been
624 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
625 the default depends on LINK_SIZE. There is a new upper limit (for safety)
626 of around 200,000 forward references. While doing this, I also speeded up
627 the filling in of repeated forward references.
628
629 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
630 incorrectly expecting the subject to contain another "a" after the start.
631
632 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
633 in the match, the SKIP should be ignored. This was not happening; instead
634 the SKIP was being treated as NOMATCH. For patterns such as
635 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
636 tested.
637
638 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
639 now much more compatible with Perl, in particular in cases where the result
640 is a non-match for a non-anchored pattern. For example, if
641 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
642 "m", where previously it did not return a name. A side effect of this
643 change is that for partial matches, the last encountered mark name is
644 returned, as for non matches. A number of tests that were previously not
645 Perl-compatible have been moved into the Perl-compatible test files. The
646 refactoring has had the pleasing side effect of removing one argument from
647 the match() function, thus reducing its stack requirements.
648
649 19. If the /S+ option was used in pcretest to study a pattern using JIT,
650 subsequent uses of /S (without +) incorrectly behaved like /S+.
651
652 21. Retrieve executable code size support for the JIT compiler and fixing
653 some warnings.
654
655 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
656 not work when the shorter character appeared right at the end of the
657 subject string.
658
659 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
660 systems.
661
662 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
663 output it when the /M option is used in pcretest.
664
665 25. The CheckMan script was not being included in the distribution. Also, added
666 an explicit "perl" to run Perl scripts from the PrepareRelease script
667 because this is reportedly needed in Windows.
668
669 26. If study data was being save in a file and studying had not found a set of
670 "starts with" bytes for the pattern, the data written to the file (though
671 never used) was taken from uninitialized memory and so caused valgrind to
672 complain.
673
674 27. Updated RunTest.bat as provided by Sheri Pierce.
675
676 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
677
678 29. Computation of memory usage for the table of capturing group names was
679 giving an unnecessarily large value.
680
681
682 Version 8.20 21-Oct-2011
683 ------------------------
684
685 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
686 a POSIX class. After further experiments with Perl, which convinced me that
687 Perl has bugs and confusions, a closing square bracket is no longer allowed
688 in a POSIX name. This bug also affected patterns with classes that started
689 with full stops.
690
691 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
692 captured substring, but while checking the failing first alternative,
693 substring 1 is temporarily captured. If the output vector supplied to
694 pcre_exec() was not big enough for this capture, the yield of the function
695 was still zero ("insufficient space for captured substrings"). This cannot
696 be totally fixed without adding another stack variable, which seems a lot
697 of expense for a edge case. However, I have improved the situation in cases
698 such as /(a)(b)x|abc/ matched against "abc", where the return code
699 indicates that fewer than the maximum number of slots in the ovector have
700 been set.
701
702 3. Related to (2) above: when there are more back references in a pattern than
703 slots in the output vector, pcre_exec() uses temporary memory during
704 matching, and copies in the captures as far as possible afterwards. It was
705 using the entire output vector, but this conflicts with the specification
706 that only 2/3 is used for passing back captured substrings. Now it uses
707 only the first 2/3, for compatibility. This is, of course, another edge
708 case.
709
710 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
711 main code base, and can be used by building with --enable-jit. When this is
712 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
713 runtime --no-jit option is given.
714
715 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
716 ovector, the return from the function was zero, implying that there were
717 other matches that did not fit. The correct "exactly full" value is now
718 returned.
719
720 6. If a subpattern that was called recursively or as a subroutine contained
721 (*PRUNE) or any other control that caused it to give a non-standard return,
722 invalid errors such as "Error -26 (nested recursion at the same subject
723 position)" or even infinite loops could occur.
724
725 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
726 computing the minimum length on reaching *ACCEPT, and so ended up with the
727 wrong value of 1 rather than 0. Further investigation indicates that
728 computing a minimum subject length in the presence of *ACCEPT is difficult
729 (think back references, subroutine calls), and so I have changed the code
730 so that no minimum is registered for a pattern that contains *ACCEPT.
731
732 8. If (*THEN) was present in the first (true) branch of a conditional group,
733 it was not handled as intended. [But see 16 below.]
734
735 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
736 Sheri Pierce.
737
738 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
739 the first byte in a match must be "a".
740
741 11. Change 17 for 8.13 increased the recursion depth for patterns like
742 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
743 pattern contains any instances of (*THEN). If it does not, the old
744 optimizations are restored. It would be nice to do this on a per-group
745 basis, but at the moment that is not feasible.
746
747 12. In some environments, the output of pcretest -C is CRLF terminated. This
748 broke RunTest's code that checks for the link size. A single white space
749 character after the value is now allowed for.
750
751 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
752 For "fr", it uses the Windows-specific input and output files.
753
754 14. If (*THEN) appeared in a group that was called recursively or as a
755 subroutine, it did not work as intended. [But see next item.]
756
757 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
758 pattern fragments (but not containing any | characters). If A and B are
759 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
760 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
761 D. In other words, Perl considers parentheses that do not contain any |
762 characters to be part of a surrounding alternative, whereas PCRE was
763 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
764 differently. PCRE now behaves in the same way as Perl, except in the case
765 of subroutine/recursion calls such as (?1) which have in any case always
766 been different (but PCRE had them first :-).
767
768 16. Related to 15 above: Perl does not treat the | in a conditional group as
769 creating alternatives. Such a group is treated in the same way as an
770 ordinary group without any | characters when processing (*THEN). PCRE has
771 been changed to match Perl's behaviour.
772
773 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
774 RunGrepTest script failed.
775
776 18. Change 22 for version 13 caused atomic groups to use more stack. This is
777 inevitable for groups that contain captures, but it can lead to a lot of
778 stack use in large patterns. The old behaviour has been restored for atomic
779 groups that do not contain any capturing parentheses.
780
781 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
782 suppress the check for a minimum subject length at run time. (If it was
783 given to pcre_exec() or pcre_dfa_exec() it did work.)
784
785 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
786 fail to work when decoding hex characters in data strings in EBCDIC
787 environments.
788
789 21. It appears that in at least one Mac OS environment, the isxdigit() function
790 is implemented as a macro that evaluates to its argument more than once,
791 contravening the C 90 Standard (I haven't checked a later standard). There
792 was an instance in pcretest which caused it to go wrong when processing
793 \x{...} escapes in subject strings. The has been rewritten to avoid using
794 things like p++ in the argument of isxdigit().
795
796
797 Version 8.13 16-Aug-2011
798 ------------------------
799
800 1. The Unicode data tables have been updated to Unicode 6.0.0.
801
802 2. Two minor typos in pcre_internal.h have been fixed.
803
804 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
805 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
806 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
807
808 4. There were a number of related bugs in the code for matching backrefences
809 caselessly in UTF-8 mode when codes for the characters concerned were
810 different numbers of bytes. For example, U+023A and U+2C65 are an upper
811 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
812 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
813 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
814 2-byte code at the end of the subject (it thought there wasn't enough data
815 left).
816
817 5. Comprehensive information about what went wrong is now returned by
818 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
819 as the output vector has at least 2 elements. The offset of the start of
820 the failing character and a reason code are placed in the vector.
821
822 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
823 now returned is for the first byte of the failing character, instead of the
824 last byte inspected. This is an incompatible change, but I hope it is small
825 enough not to be a problem. It makes the returned offset consistent with
826 pcre_exec() and pcre_dfa_exec().
827
828 7. pcretest now gives a text phrase as well as the error number when
829 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
830 failure, the offset and reason code are output.
831
832 8. When \R was used with a maximizing quantifier it failed to skip backwards
833 over a \r\n pair if the subsequent match failed. Instead, it just skipped
834 back over a single character (\n). This seems wrong (because it treated the
835 two characters as a single entity when going forwards), conflicts with the
836 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
837 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
838 has been changed.
839
840 9. Some internal refactoring has changed the processing so that the handling
841 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
842 time (the PCRE_DOTALL option was changed this way some time ago: version
843 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
844 which was always a bit of a fudge. It also means that there is one less
845 argument for the match() function, which reduces its stack requirements
846 slightly. This change also fixes an incompatibility with Perl: the pattern
847 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
848
849 10. More internal refactoring has drastically reduced the number of recursive
850 calls to match() for possessively repeated groups such as (abc)++ when
851 using pcre_exec().
852
853 11. While implementing 10, a number of bugs in the handling of groups were
854 discovered and fixed:
855
856 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
857 (a|)*(?1) gave a compile-time internal error.
858 ((a|)+)+ did not notice that the outer group could match an empty string.
859 (^a|^)+ was not marked as anchored.
860 (.*a|.*)+ was not marked as matching at start or after a newline.
861
862 12. Yet more internal refactoring has removed another argument from the match()
863 function. Special calls to this function are now indicated by setting a
864 value in a variable in the "match data" data block.
865
866 13. Be more explicit in pcre_study() instead of relying on "default" for
867 opcodes that mean there is no starting character; this means that when new
868 ones are added and accidentally left out of pcre_study(), testing should
869 pick them up.
870
871 14. The -s option of pcretest has been documented for ages as being an old
872 synonym of -m (show memory usage). I have changed it to mean "force study
873 for every regex", that is, assume /S for every regex. This is similar to -i
874 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
875 using it. It makes it easier to run collections of tests with and without
876 study enabled, and thereby test pcre_study() more easily. All the standard
877 tests are now run with and without -s (but some patterns can be marked as
878 "never study" - see 20 below).
879
880 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
881 restoration of the capturing data to the outer values was not happening
882 correctly.
883
884 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
885 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
886 pattern had matched an empty string, and so incorrectly returned a no
887 match.
888
889 17. There was optimizing code for the last branch of non-capturing parentheses,
890 and also for the obeyed branch of a conditional subexpression, which used
891 tail recursion to cut down on stack usage. Unfortunately, now that there is
892 the possibility of (*THEN) occurring in these branches, tail recursion is
893 no longer possible because the return has to be checked for (*THEN). These
894 two optimizations have therefore been removed. [But see 8.20/11 above.]
895
896 18. If a pattern containing \R was studied, it was assumed that \R always
897 matched two bytes, thus causing the minimum subject length to be
898 incorrectly computed because \R can also match just one byte.
899
900 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
901 was incorrectly computed.
902
903 20. If /S is present twice on a test pattern in pcretest input, it now
904 *disables* studying, thereby overriding the use of -s on the command line
905 (see 14 above). This is necessary for one or two tests to keep the output
906 identical in both cases.
907
908 21. When (*ACCEPT) was used in an assertion that matched an empty string and
909 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
910
911 22. When an atomic group that contained a capturing parenthesis was
912 successfully matched, but the branch in which it appeared failed, the
913 capturing was not being forgotten if a higher numbered group was later
914 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
915 group 1 to "a", when in fact it should be unset. This applied to multi-
916 branched capturing and non-capturing groups, repeated or not, and also to
917 positive assertions (capturing in negative assertions does not happen
918 in PCRE) and also to nested atomic groups.
919
920 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
921 subject after a captured substring, to make it easier to tell which of a
922 number of identical substrings has been captured.
923
924 24. The way atomic groups are processed by pcre_exec() has been changed so that
925 if they are repeated, backtracking one repetition now resets captured
926 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
927 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
928 "aaa". Previously, it would have been "a". As part of this code
929 refactoring, the way recursive calls are handled has also been changed.
930
931 25. If an assertion condition captured any substrings, they were not passed
932 back unless some other capturing happened later. For example, if
933 (?(?=(a))a) was matched against "a", no capturing was returned.
934
935 26. When studying a pattern that contained subroutine calls or assertions,
936 the code for finding the minimum length of a possible match was handling
937 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
938 group 1 called group 2 while simultaneously a separate group 2 called group
939 1). A stack overflow occurred in this case. I have fixed this by limiting
940 the recursion depth to 10.
941
942 27. Updated RunTest.bat in the distribution to the version supplied by Tom
943 Fortmann. This supports explicit test numbers on the command line, and has
944 argument validation and error reporting.
945
946 28. An instance of \X with an unlimited repeat could fail if at any point the
947 first character it looked at was a mark character.
948
949 29. Some minor code refactoring concerning Unicode properties and scripts
950 should reduce the stack requirement of match() slightly.
951
952 30. Added the '=' option to pcretest to check the setting of unused capturing
953 slots at the end of the pattern, which are documented as being -1, but are
954 not included in the return count.
955
956 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
957 compiled something random. Now it gives a compile-time error (as does
958 Perl).
959
960 32. A *MARK encountered during the processing of a positive assertion is now
961 recorded and passed back (compatible with Perl).
962
963 33. If --only-matching or --colour was set on a pcregrep call whose pattern
964 had alternative anchored branches, the search for a second match in a line
965 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
966 matched the line "0102" twice. The same bug affected patterns that started
967 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
968 twice.
969
970 34. Previously, PCRE did not allow quantification of assertions. However, Perl
971 does, and because of capturing effects, quantifying parenthesized
972 assertions may at times be useful. Quantifiers are now allowed for
973 parenthesized assertions.
974
975 35. A minor code tidy in pcre_compile() when checking options for \R usage.
976
977 36. \g was being checked for fancy things in a character class, when it should
978 just be a literal "g".
979
980 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
981 appearance of a nested POSIX class supersedes an apparent external class.
982 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
983 unescaped square brackets may also appear as part of class names. For
984 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
985 more like Perl. (But see 8.20/1 above.)
986
987 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
988 was because it thought it was \N{name}, which is not supported).
989
990 39. Add minix to OS list not supporting the -S option in pcretest.
991
992 40. PCRE tries to detect cases of infinite recursion at compile time, but it
993 cannot analyze patterns in sufficient detail to catch mutual recursions
994 such as ((?1))((?2)). There is now a runtime test that gives an error if a
995 subgroup is called recursively as a subpattern for a second time at the
996 same position in the subject string. In previous releases this might have
997 been caught by the recursion limit, or it might have run out of stack.
998
999 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1000 happen only once. PCRE was, however incorrectly giving a compile time error
1001 "recursive call could loop indefinitely" because it cannot analyze the
1002 pattern in sufficient detail. The compile time test no longer happens when
1003 PCRE is compiling a conditional subpattern, but actual runaway loops are
1004 now caught at runtime (see 40 above).
1005
1006 42. It seems that Perl allows any characters other than a closing parenthesis
1007 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1008 has been changed to be the same.
1009
1010 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1011 as not to get warnings when autogen.sh is called. Also changed
1012 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1013
1014 44. To help people who use pcregrep to scan files containing exceedingly long
1015 lines, the following changes have been made:
1016
1017 (a) The default value of the buffer size parameter has been increased from
1018 8K to 20K. (The actual buffer used is three times this size.)
1019
1020 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1021 PCRE is built.
1022
1023 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1024 to be set at run time.
1025
1026 (d) Numerical values in pcregrep options can be followed by K or M, for
1027 example --buffer-size=50K.
1028
1029 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1030 given and the return code is set to 2.
1031
1032 45. Add a pointer to the latest mark to the callout data block.
1033
1034 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1035 partial match of an empty string instead of no match. This was specific to
1036 the use of ".".
1037
1038 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1039 complete match instead of a partial match. This bug was dependent on both
1040 the PCRE_UTF8 and PCRE_DOTALL options being set.
1041
1042 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1043 starting byte set, because \b was not being ignored.
1044
1045
1046 Version 8.12 15-Jan-2011
1047 ------------------------
1048
1049 1. Fixed some typos in the markup of the man pages, and wrote a script that
1050 checks for such things as part of the documentation building process.
1051
1052 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1053 --match-limit and --recursion-limit options (added for 8.11). In
1054 particular, this made one of the standard tests fail. (The integer value
1055 went into the wrong half of a long int.)
1056
1057 3. If the --colour option was given to pcregrep with -v (invert match), it
1058 did strange things, either producing crazy output, or crashing. It should,
1059 of course, ignore a request for colour when reporting lines that do not
1060 match.
1061
1062 4. Another pcregrep bug caused similar problems if --colour was specified with
1063 -M (multiline) and the pattern match finished with a line ending.
1064
1065 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1066 matched in multiline mode, the following line was shown as part of the
1067 match. This seems wrong, so I have changed it.
1068
1069 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1070 the check for further matches in the same line (so they could be coloured)
1071 to overrun the end of the current line. If another match was found, it was
1072 incorrectly shown (and then shown again when found in the next line).
1073
1074 7. If pcregrep was compiled under Windows, there was a reference to the
1075 function pcregrep_exit() before it was defined. I am assuming this was
1076 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1077 reported by a user. I've moved the definition above the reference.
1078
1079
1080 Version 8.11 10-Dec-2010
1081 ------------------------
1082
1083 1. (*THEN) was not working properly if there were untried alternatives prior
1084 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1085 backtracked to try for "b" instead of moving to the next alternative branch
1086 at the same level (in this case, to look for "c"). The Perl documentation
1087 is clear that when (*THEN) is backtracked onto, it goes to the "next
1088 alternative in the innermost enclosing group".
1089
1090 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1091 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1092 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1093 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1094 (*THEN).
1095
1096 3. If \s appeared in a character class, it removed the VT character from
1097 the class, even if it had been included by some previous item, for example
1098 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1099 of \s, but is part of the POSIX "space" class.)
1100
1101 4. A partial match never returns an empty string (because you can always
1102 match an empty string at the end of the subject); however the checking for
1103 an empty string was starting at the "start of match" point. This has been
1104 changed to the "earliest inspected character" point, because the returned
1105 data for a partial match starts at this character. This means that, for
1106 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1107 (previously it gave "no match").
1108
1109 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1110 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1111 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1112 has an implication that the given string is incomplete (because a partial
1113 match is preferred over a full match). For this reason, these items now
1114 give a partial match in this situation. [Aside: previously, the one case
1115 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1116 match rather than a full match, which was wrong by the old rules, but is
1117 now correct.]
1118
1119 6. There was a bug in the handling of #-introduced comments, recognized when
1120 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1121 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1122 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1123 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1124 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1125 places in pcre_compile().
1126
1127 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1128 comments when looking ahead for named forward references to subpatterns,
1129 the only newline sequence it recognized was NL. It now handles newlines
1130 according to the set newline convention.
1131
1132 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1133 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1134 cater for a lack of strerror(). These oversights have been fixed.
1135
1136 9. Added --match-limit and --recursion-limit to pcregrep.
1137
1138 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1139
1140 11. When the -o option was used, pcregrep was setting a return code of 1, even
1141 when matches were found, and --line-buffered was not being honoured.
1142
1143 12. Added an optional parentheses number to the -o and --only-matching options
1144 of pcregrep.
1145
1146 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1147 can match an empty string. The code to do it in pcretest and pcredemo
1148 needed fixing:
1149
1150 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1151 only one byte after an empty string match just before CRLF (this case
1152 just got forgotten; "any" and "anycrlf" were OK).
1153
1154 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1155 mode when an empty string match preceded an ASCII character followed by
1156 a non-ASCII character. (The code for advancing by one character rather
1157 than one byte was nonsense.)
1158
1159 (c) The pcredemo.c sample program did not have any code at all to handle
1160 the cases when CRLF is a valid newline sequence.
1161
1162 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1163 as a starting offset was within the subject string. There is now a new
1164 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1165 negative or greater than the length of the string. In order to test this,
1166 pcretest is extended to allow the setting of negative starting offsets.
1167
1168 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1169 starting offset points to the beginning of a UTF-8 character was
1170 unnecessarily clumsy. I tidied it up.
1171
1172 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1173 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1174
1175 17. Nobody had reported that the --include_dir option, which was added in
1176 release 7.7 should have been called --include-dir (hyphen, not underscore)
1177 for compatibility with GNU grep. I have changed it to --include-dir, but
1178 left --include_dir as an undocumented synonym, and the same for
1179 --exclude-dir, though that is not available in GNU grep, at least as of
1180 release 2.5.4.
1181
1182 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1183 characters from a string of bytes) have been redefined so as not to use
1184 loops, in order to improve performance in some environments. At the same
1185 time, I abstracted some of the common code into auxiliary macros to save
1186 repetition (this should not affect the compiled code).
1187
1188 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1189 compile-time error is now given if \c is not followed by an ASCII
1190 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1191 different, and any byte value is allowed.)
1192
1193 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1194 START_OPTIMIZE option, which is now allowed at compile time - but just
1195 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1196 to pcregrep and other applications that have no direct access to PCRE
1197 options. The new /Y option in pcretest sets this option when calling
1198 pcre_compile().
1199
1200 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1201 back references. Groups containing recursive back references were forced to
1202 be atomic by that change, but in the case of named groups, the amount of
1203 memory required was incorrectly computed, leading to "Failed: internal
1204 error: code overflow". This has been fixed.
1205
1206 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1207 pcretest.c, to avoid build problems in some Borland environments.
1208
1209
1210 Version 8.10 25-Jun-2010
1211 ------------------------
1212
1213 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1214 THEN.
1215
1216 2. (*ACCEPT) was not working when inside an atomic group.
1217
1218 3. Inside a character class, \B is treated as a literal by default, but
1219 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1220 causes the error). The code is unchanged, but I tidied the documentation.
1221
1222 4. Inside a character class, PCRE always treated \R and \X as literals,
1223 whereas Perl faults them if its -w option is set. I have changed PCRE so
1224 that it faults them when PCRE_EXTRA is set.
1225
1226 5. Added support for \N, which always matches any character other than
1227 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1228
1229 6. When compiling pcregrep with newer versions of gcc which may have
1230 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1231 declared with attribute warn_unused_result" were given. Just casting the
1232 result to (void) does not stop the warnings; a more elaborate fudge is
1233 needed. I've used a macro to implement this.
1234
1235 7. Minor change to pcretest.c to avoid a compiler warning.
1236
1237 8. Added four artifical Unicode properties to help with an option to make
1238 \s etc use properties (see next item). The new properties are: Xan
1239 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1240
1241 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1242 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1243 this option. Modified pcretest to add /W to test this facility. Added
1244 REG_UCP to make it available via the POSIX interface.
1245
1246 10. Added --line-buffered to pcregrep.
1247
1248 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1249 studied, and the match started with a letter with a code point greater than
1250 127 whose first byte was different to the first byte of the other case of
1251 the letter, the other case of this starting letter was not recognized
1252 (#976).
1253
1254 12. If a pattern that was studied started with a repeated Unicode property
1255 test, for example, \p{Nd}+, there was the theoretical possibility of
1256 setting up an incorrect bitmap of starting bytes, but fortunately it could
1257 not have actually happened in practice until change 8 above was made (it
1258 added property types that matched character-matching opcodes).
1259
1260 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1261 possible starting bytes for non-anchored patterns.
1262
1263 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1264 \R, and also a number of cases that involve Unicode properties, both
1265 explicit and implicit when PCRE_UCP is set.
1266
1267 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1268 input, it could crash or give wrong results if characters with values
1269 greater than 0xc0 were present in the subject string. (Detail: it assumed
1270 UTF-8 input when processing these items.)
1271
1272 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1273 size_t is 64-bit (#991).
1274
1275 17. Added a check for running out of memory when PCRE is compiled with
1276 --disable-stack-for-recursion (#990).
1277
1278 18. If the last data line in a file for pcretest does not have a newline on
1279 the end, a newline was missing in the output.
1280
1281 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1282 less than 128) in its various bitmaps. However, there is a facility for
1283 generating tables according to the current locale when PCRE is compiled. It
1284 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1285 characters, are recognized by isspace() and therefore were getting set in
1286 these tables, and indeed these tables seem to approximate to ISO 8859. This
1287 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1288 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1289 which of course cannot start UTF-8 characters. I have changed the code so
1290 that only real ASCII characters (less than 128) and the correct starting
1291 bytes for UTF-8 encodings are set for characters greater than 127 when in
1292 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1293 altogether.)
1294
1295 20. Added the /T option to pcretest so as to be able to run tests with non-
1296 standard character tables, thus making it possible to include the tests
1297 used for 19 above in the standard set of tests.
1298
1299 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1300 reference to a subpattern the other side of a comment that contains an
1301 opening parenthesis caused either an internal compiling error, or a
1302 reference to the wrong subpattern.
1303
1304
1305 Version 8.02 19-Mar-2010
1306 ------------------------
1307
1308 1. The Unicode data tables have been updated to Unicode 5.2.0.
1309
1310 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1311 configured.
1312
1313 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1314 original author of that file, following a query about its status.
1315
1316 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1317 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1318
1319 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1320 quantifier applied to a forward-referencing subroutine call, could compile
1321 incorrect code or give the error "internal error: previously-checked
1322 referenced subpattern not found".
1323
1324 6. Both MS Visual Studio and Symbian OS have problems with initializing
1325 variables to point to external functions. For these systems, therefore,
1326 pcre_malloc etc. are now initialized to local functions that call the
1327 relevant global functions.
1328
1329 7. There were two entries missing in the vectors called coptable and poptable
1330 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1331 I've fixed the data, and added a kludgy way of testing at compile time that
1332 the lengths are correct (equal to the number of opcodes).
1333
1334 8. Following on from 7, I added a similar kludge to check the length of the
1335 eint vector in pcreposix.c.
1336
1337 9. Error texts for pcre_compile() are held as one long string to avoid too
1338 much relocation at load time. To find a text, the string is searched,
1339 counting zeros. There was no check for running off the end of the string,
1340 which could happen if a new error number was added without updating the
1341 string.
1342
1343 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1344
1345 11. \K was not working if it appeared in an atomic group or in a group that
1346 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1347 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1348 the assertion is positive, but not if it is negative.
1349
1350 12. Change 11 fortuitously reduced the size of the stack frame used in the
1351 "match()" function of pcre_exec.c by one pointer. Forthcoming
1352 implementation of support for (*MARK) will need an extra pointer on the
1353 stack; I have reserved it now, so that the stack frame size does not
1354 decrease.
1355
1356 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1357 item in branch that calls a recursion is a subroutine call - as in the
1358 second branch in the above example - was incorrectly given the compile-
1359 time error "recursive call could loop indefinitely" because pcre_compile()
1360 was not correctly checking the subroutine for matching a non-empty string.
1361
1362 14. The checks for overrunning compiling workspace could trigger after an
1363 overrun had occurred. This is a "should never occur" error, but it can be
1364 triggered by pathological patterns such as hundreds of nested parentheses.
1365 The checks now trigger 100 bytes before the end of the workspace.
1366
1367 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1368
1369
1370 Version 8.01 19-Jan-2010
1371 ------------------------
1372
1373 1. If a pattern contained a conditional subpattern with only one branch (in
1374 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1375 computed the wrong minimum data length (which is of course zero for such
1376 subpatterns). This could cause incorrect "no match" results.
1377
1378 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1379 the pattern is reset in the first branch, pcre_compile() failed with
1380 "internal error: code overflow at offset...". This happened only when
1381 the reset was to the original external option setting. (An optimization
1382 abstracts leading options settings into an external setting, which was the
1383 cause of this.)
1384
1385 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1386 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1387 assertion pattern did not match (meaning that the assertion was true), it
1388 was incorrectly treated as false if the SKIP had been reached during the
1389 matching. This also applied to assertions used as conditions.
1390
1391 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1392 assertion subpattern, including such a pattern used as a condition,
1393 unpredictable results occurred, instead of the error return
1394 PCRE_ERROR_DFA_UITEM.
1395
1396 5. The C++ GlobalReplace function was not working like Perl for the special
1397 situation when an empty string is matched. It now does the fancy magic
1398 stuff that is necessary.
1399
1400 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1401 removed. (These were left over from very, very early versions of PCRE.)
1402
1403 7. Some cosmetic changes to the code to make life easier when compiling it
1404 as part of something else:
1405
1406 (a) Change DEBUG to PCRE_DEBUG.
1407
1408 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1409 called "current" as "current_branch", to prevent a collision with the
1410 Linux macro when compiled as a kernel module.
1411
1412 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1413 prevent a collision with the Linux macro when compiled as a kernel
1414 module.
1415
1416 8. In pcre_compile() there are some checks for integer overflows that used to
1417 cast potentially large values to (double). This has been changed to that
1418 when building, a check for int64_t is made, and if it is found, it is used
1419 instead, thus avoiding the use of floating point arithmetic. (There is no
1420 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1421 double.
1422
1423 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1424 2005 (difference between two addresses compared to an unsigned value).
1425
1426 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1427 custom one, because of the following reported problem in Windows:
1428
1429 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1430 under Win32.
1431 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1432 therefore missing the function definition.
1433 - The compiler thus generates a "C" signature for the test function.
1434 - The linker fails to find the "C" function.
1435 - PCRE fails to configure if asked to do so against libbz2.
1436
1437 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1438 messages were output:
1439
1440 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1441 rerunning libtoolize, to keep the correct libtool macros in-tree.
1442 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1443
1444 I have done both of these things.
1445
1446 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1447 most of the time, it *can* run out if it is given a pattern that contains a
1448 runaway infinite recursion. I updated the discussion in the pcrestack man
1449 page.
1450
1451 13. Now that we have gone to the x.xx style of version numbers, the minor
1452 version may start with zero. Using 08 or 09 is a bad idea because users
1453 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1454 interpreted as invalid octal numbers. I've updated the previous comment in
1455 configure.ac, and also added a check that gives an error if 08 or 09 are
1456 used.
1457
1458 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1459 causing partial matching to fail when the end of the subject matched \W
1460 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1461
1462 15. There were some discrepancies between the declarations in pcre_internal.h
1463 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1464 their definitions. The declarations used "const uschar *" and the
1465 definitions used USPTR. Even though USPTR is normally defined as "const
1466 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1467 reported that: "This difference in casting confuses some C++ compilers, for
1468 example, SunCC recognizes above declarations as different functions and
1469 generates broken code for hbpcre." I have changed the declarations to use
1470 USPTR.
1471
1472 16. GNU libtool is named differently on some systems. The autogen.sh script now
1473 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1474 (FreeBSD).
1475
1476 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1477 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1478 comment: "Figure out how to create a longlong from a string: strtoll and
1479 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1480 instance, but it only takes 2 args instead of 3!"
1481
1482 18. A subtle bug concerned with back references has been fixed by a change of
1483 specification, with a corresponding code fix. A pattern such as
1484 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1485 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1486 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1487 same bug. Such groups have to be quantified to be useful, or contained
1488 inside another quantified group. (If there's no repetition, the reference
1489 can never match.) The problem arises because, having left the group and
1490 moved on to the rest of the pattern, a later failure that backtracks into
1491 the group uses the captured value from the final iteration of the group
1492 rather than the correct earlier one. I have fixed this in PCRE by forcing
1493 any group that contains a reference to itself to be an atomic group; that
1494 is, there cannot be any backtracking into it once it has completed. This is
1495 similar to recursive and subroutine calls.
1496
1497
1498 Version 8.00 19-Oct-09
1499 ----------------------
1500
1501 1. The table for translating pcre_compile() error codes into POSIX error codes
1502 was out-of-date, and there was no check on the pcre_compile() error code
1503 being within the table. This could lead to an OK return being given in
1504 error.
1505
1506 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1507 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1508 in a Windows environment.
1509
1510 3. The pcregrep --count option prints the count for each file even when it is
1511 zero, as does GNU grep. However, pcregrep was also printing all files when
1512 --files-with-matches was added. Now, when both options are given, it prints
1513 counts only for those files that have at least one match. (GNU grep just
1514 prints the file name in this circumstance, but including the count seems
1515 more useful - otherwise, why use --count?) Also ensured that the
1516 combination -clh just lists non-zero counts, with no names.
1517
1518 4. The long form of the pcregrep -F option was incorrectly implemented as
1519 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1520 but it seems right to fix it, and I didn't think it was worth preserving
1521 the old behaviour.
1522
1523 5. The command line items --regex=pattern and --regexp=pattern were not
1524 recognized by pcregrep, which required --regex pattern or --regexp pattern
1525 (with a space rather than an '='). The man page documented the '=' forms,
1526 which are compatible with GNU grep; these now work.
1527
1528 6. No libpcreposix.pc file was created for pkg-config; there was just
1529 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1530
1531 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1532 when UCP support is not needed, by modifying the Python script that
1533 generates it from Unicode data files. This should not matter if the module
1534 is correctly used as a library, but I received one complaint about 50K of
1535 unwanted data. My guess is that the person linked everything into his
1536 program rather than using a library. Anyway, it does no harm.
1537
1538 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1539 was a minimum greater than 1 for a wide character in a possessive
1540 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1541 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1542 character. Chaos in the form of incorrect output or a compiling loop could
1543 result.
1544
1545 9. The restrictions on what a pattern can contain when partial matching is
1546 requested for pcre_exec() have been removed. All patterns can now be
1547 partially matched by this function. In addition, if there are at least two
1548 slots in the offset vector, the offset of the earliest inspected character
1549 for the match and the offset of the end of the subject are set in them when
1550 PCRE_ERROR_PARTIAL is returned.
1551
1552 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1553 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1554 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1555 and may be more useful for multi-segment matching.
1556
1557 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1558 used to be given if ever the end of the subject was reached; now it is
1559 given only if matching could not proceed because another character was
1560 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1561 string "Z", which now yields "no match" instead of "partial match". In the
1562 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1563 final character ended with (*FAIL).
1564
1565 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1566 if the pattern had a "must contain" character that was already found in the
1567 earlier partial match, unless partial matching was again requested. For
1568 example, with the pattern /dog.(body)?/, the "must contain" character is
1569 "g". If the first part-match was for the string "dog", restarting with
1570 "sbody" failed. This bug has been fixed.
1571
1572 13. The string returned by pcre_dfa_exec() after a partial match has been
1573 changed so that it starts at the first inspected character rather than the
1574 first character of the match. This makes a difference only if the pattern
1575 starts with a lookbehind assertion or \b or \B (\K is not supported by
1576 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1577 matching functions compatible, and I think it's the right thing to do.
1578
1579 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1580 so that the demonstration program is easily available in environments where
1581 PCRE has not been installed from source.
1582
1583 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1584 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1585 library.
1586
1587 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1588 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1589 is not the first non-POSIX option to be added. Clearly some people find
1590 these options useful.
1591
1592 17. If a caller to the POSIX matching function regexec() passes a non-zero
1593 value for nmatch with a NULL value for pmatch, the value of
1594 nmatch is forced to zero.
1595
1596 18. RunGrepTest did not have a test for the availability of the -u option of
1597 the diff command, as RunTest does. It now checks in the same way as
1598 RunTest, and also checks for the -b option.
1599
1600 19. If an odd number of negated classes containing just a single character
1601 interposed, within parentheses, between a forward reference to a named
1602 subpattern and the definition of the subpattern, compilation crashed with
1603 an internal error, complaining that it could not find the referenced
1604 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1605 [The bug was that it was starting one character too far in when skipping
1606 over the character class, thus treating the ] as data rather than
1607 terminating the class. This meant it could skip too much.]
1608
1609 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1610 /g option in pcretest when the pattern contains \K, which makes it possible
1611 to have an empty string match not at the start, even when the pattern is
1612 anchored. Updated pcretest and pcredemo to use this option.
1613
1614 21. If the maximum number of capturing subpatterns in a recursion was greater
1615 than the maximum at the outer level, the higher number was returned, but
1616 with unset values at the outer level. The correct (outer level) value is
1617 now given.
1618
1619 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1620 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1621 make it do so. The string so far is captured, making this feature
1622 compatible with Perl.
1623
1624 23. The tests have been re-organized, adding tests 11 and 12, to make it
1625 possible to check the Perl 5.10 features against Perl 5.10.
1626
1627 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1628 pattern matches a fixed length string. PCRE did not allow this; now it
1629 does. Neither allows recursion.
1630
1631 25. I finally figured out how to implement a request to provide the minimum
1632 length of subject string that was needed in order to match a given pattern.
1633 (It was back references and recursion that I had previously got hung up
1634 on.) This code has now been added to pcre_study(); it finds a lower bound
1635 to the length of subject needed. It is not necessarily the greatest lower
1636 bound, but using it to avoid searching strings that are too short does give
1637 some useful speed-ups. The value is available to calling programs via
1638 pcre_fullinfo().
1639
1640 26. While implementing 25, I discovered to my embarrassment that pcretest had
1641 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1642 study optimizations had never been tested with that matching function.
1643 Oops. What is worse, even when it was passed study data, there was a bug in
1644 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1645 were also very few tests of studied patterns with pcre_dfa_exec().
1646
1647 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1648 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1649 on the other side of the coin, they are no longer allowed to have different
1650 names, because these cannot be distinguished in PCRE, and this has caused
1651 confusion. (This is a difference from Perl.)
1652
1653 28. When duplicate subpattern names are present (necessarily with different
1654 numbers, as required by 27 above), and a test is made by name in a
1655 conditional pattern, either for a subpattern having been matched, or for
1656 recursion in such a pattern, all the associated numbered subpatterns are
1657 tested, and the overall condition is true if the condition is true for any
1658 one of them. This is the way Perl works, and is also more like the way
1659 testing by number works.
1660
1661
1662 Version 7.9 11-Apr-09
1663 ---------------------
1664
1665 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1666 (pcretest), all targets were linked against these libraries. This included
1667 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1668 libraries. This caused unwanted dependencies to be created. This problem
1669 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1670 pcretest is linked with readline.
1671
1672 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1673 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1674 moved outside it again, because FALSE and TRUE are already defined in AIX,
1675 but BOOL is not.
1676
1677 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1678 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1679
1680 4. The pcregrep documentation said spaces were inserted as well as colons (or
1681 hyphens) following file names and line numbers when outputting matching
1682 lines. This is not true; no spaces are inserted. I have also clarified the
1683 wording for the --colour (or --color) option.
1684
1685 5. In pcregrep, when --colour was used with -o, the list of matching strings
1686 was not coloured; this is different to GNU grep, so I have changed it to be
1687 the same.
1688
1689 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1690 each matching line was coloured. Now it goes on to look for further matches
1691 of any of the test patterns, which is the same behaviour as GNU grep.
1692
1693 7. A pattern that could match an empty string could cause pcregrep to loop; it
1694 doesn't make sense to accept an empty string match in pcregrep, so I have
1695 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1696 seems to be how GNU grep behaves. [But see later change 40 for release
1697 8.33.]
1698
1699 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1700 start or after a newline", because the conditional assertion was not being
1701 correctly handled. The rule now is that both the assertion and what follows
1702 in the first alternative must satisfy the test.
1703
1704 9. If auto-callout was enabled in a pattern with a conditional group whose
1705 condition was an assertion, PCRE could crash during matching, both with
1706 pcre_exec() and pcre_dfa_exec().
1707
1708 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1709 used for matching.
1710
1711 11. Unicode property support in character classes was not working for
1712 characters (bytes) greater than 127 when not in UTF-8 mode.
1713
1714 12. Added the -M command line option to pcretest.
1715
1716 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1717
1718 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1719
1720 16. Added comments and documentation about mis-use of no_arg in the C++
1721 wrapper.
1722
1723 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1724 from Martin Jerabek that uses macro names for all relevant character and
1725 string constants.
1726
1727 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1728 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1729 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1730 these, but not everybody uses configure.
1731
1732 19. A conditional group that had only one branch was not being correctly
1733 recognized as an item that could match an empty string. This meant that an
1734 enclosing group might also not be so recognized, causing infinite looping
1735 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1736 with the subject "ab", where knowledge that the repeated group can match
1737 nothing is needed in order to break the loop.
1738
1739 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1740 exec(), but without supplying a callout function, matching went wrong.
1741
1742 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1743 leak if the size of the offset vector was greater than 30. When the vector
1744 is smaller, the saved offsets during recursion go onto a local stack
1745 vector, but for larger vectors malloc() is used. It was failing to free
1746 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1747 error, in fact).
1748
1749 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1750 heapframe that is used only when UTF-8 support is enabled. This caused no
1751 problem, but was untidy.
1752
1753 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1754 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1755 included within another project.
1756
1757 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1758 slightly modified by me:
1759
1760 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1761 not building pcregrep.
1762
1763 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1764 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1765
1766 25. Forward references, both numeric and by name, in patterns that made use of
1767 duplicate group numbers, could behave incorrectly or give incorrect errors,
1768 because when scanning forward to find the reference group, PCRE was not
1769 taking into account the duplicate group numbers. A pattern such as
1770 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1771
1772 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1773 the feature of a custom pointer more persuasive (as requested by a user).
1774
1775 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1776 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1777 pre-defined.
1778
1779 28. Added support for (*UTF8) at the start of a pattern.
1780
1781 29. Arrange for flags added by the "release type" setting in CMake to be shown
1782 in the configuration summary.
1783
1784
1785 Version 7.8 05-Sep-08
1786 ---------------------
1787
1788 1. Replaced UCP searching code with optimized version as implemented for Ad
1789 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1790 stage table and inline lookup instead of a function, giving speed ups of 2
1791 to 5 times on some simple patterns that I tested. Permission was given to
1792 distribute the MultiStage2.py script that generates the tables (it's not in
1793 the tarball, but is in the Subversion repository).
1794
1795 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1796 scripts.
1797
1798 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1799 a group with a zero qualifier. The result of the study could be incorrect,
1800 or the function might crash, depending on the pattern.
1801
1802 4. Caseless matching was not working for non-ASCII characters in back
1803 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1804 It now works when Unicode Property Support is available.
1805
1806 5. In pcretest, an escape such as \x{de} in the data was always generating
1807 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1808 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1809 truncation.
1810
1811 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1812
1813 7. Added two (int) casts to pcregrep when printing the difference of two
1814 pointers, in case they are 64-bit values.
1815
1816 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1817 test 2 if it fails.
1818
1819 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1820 and a #define of that name to empty if it is not externally set. This is to
1821 allow users of MSVC to set it if necessary.
1822
1823 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1824 the convenience functions in the pcre_get.c source file.
1825
1826 11. An option change at the start of a pattern that had top-level alternatives
1827 could cause overwriting and/or a crash. This command provoked a crash in
1828 some environments:
1829
1830 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1831
1832 This potential security problem was recorded as CVE-2008-2371.
1833
1834 12. For a pattern where the match had to start at the beginning or immediately
1835 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1836 pcre_dfa_exec() could read past the end of the passed subject if there was
1837 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1838 pcretest so that it places the subject at the end of its malloc-ed buffer.
1839
1840 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1841 exec() might read past the end of the data buffer in UTF-8 mode.
1842
1843 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1844 the data contained the byte 0x85 as part of a UTF-8 character within its
1845 first line. This applied both to normal and DFA matching.
1846
1847 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1848 /^[^d]*?$/8 failed to match "abc".
1849
1850 16. Added a missing copyright notice to pcrecpp_internal.h.
1851
1852 17. Make it more clear in the documentation that values returned from
1853 pcre_exec() in ovector are byte offsets, not character counts.
1854
1855 18. Tidied a few places to stop certain compilers from issuing warnings.
1856
1857 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1858 supplied by Stefan Weber. I made a further small update for 7.8 because
1859 there is a change of source arrangements: the pcre_searchfuncs.c module is
1860 replaced by pcre_ucd.c.
1861
1862
1863 Version 7.7 07-May-08
1864 ---------------------
1865
1866 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1867 a string to a long long, pretend we don't even have a long long." This is
1868 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1869
1870 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1871 pre-7.6 versions, which defined a global no_arg variable instead of putting
1872 it in the RE class. (See also #8 below.)
1873
1874 3. Remove a line of dead code, identified by coverity and reported by Nuno
1875 Lopes.
1876
1877 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1878
1879 (1) The include/exclude patterns were being applied to the whole pathnames
1880 of files, instead of just to the final components.
1881
1882 (2) If there was more than one level of directory, the subdirectories were
1883 skipped unless they satisfied the include/exclude conditions. This is
1884 inconsistent with GNU grep (and could even be seen as contrary to the
1885 pcregrep specification - which I improved to make it absolutely clear).
1886 The action now is always to scan all levels of directory, and just
1887 apply the include/exclude patterns to regular files.
1888
1889 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1890 --exclude_dir in the tests to avoid scanning .svn directories.
1891
1892 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1893 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1894 doesn't support NULs in patterns.
1895
1896 7. Added some missing "const"s to declarations of static tables in
1897 pcre_compile.c and pcre_dfa_exec.c.
1898
1899 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1900 caused by fix #2 above. (Subsequently also a second patch to fix the
1901 first patch. And a third patch - this was a messy problem.)
1902
1903 9. Applied Craig's patch to remove the use of push_back().
1904
1905 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1906 matching function regexec().
1907
1908 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1909 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1910 references. PCRE supports relative numbers with this syntax (I don't think
1911 Oniguruma does).
1912
1913 12. Previously, a group with a zero repeat such as (...){0} was completely
1914 omitted from the compiled regex. However, this means that if the group
1915 was called as a subroutine from elsewhere in the pattern, things went wrong
1916 (an internal error was given). Such groups are now left in the compiled
1917 pattern, with a new opcode that causes them to be skipped at execution
1918 time.
1919
1920 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1921 to the way PCRE behaves:
1922
1923 (a) A lone ] character is dis-allowed (Perl treats it as data).
1924
1925 (b) A back reference to an unmatched subpattern matches an empty string
1926 (Perl fails the current match path).
1927
1928 (c) A data ] in a character class must be notated as \] because if the
1929 first data character in a class is ], it defines an empty class. (In
1930 Perl it is not possible to have an empty class.) The empty class []
1931 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1932 The negative empty class [^] matches any one character, independently
1933 of the DOTALL setting.
1934
1935 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1936 non-existent subpattern following a character class starting with ']' and
1937 containing () gave an internal compiling error instead of "reference to
1938 non-existent subpattern". Fortunately, when the pattern did exist, the
1939 compiled code was correct. (When scanning forwards to check for the
1940 existence of the subpattern, it was treating the data ']' as terminating
1941 the class, so got the count wrong. When actually compiling, the reference
1942 was subsequently set up correctly.)
1943
1944 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1945 it was being rejected as not supported by pcre_dfa_exec(), even though
1946 other assertions are supported. I have made pcre_dfa_exec() support
1947 (*FAIL).
1948
1949 16. The implementation of 13c above involved the invention of a new opcode,
1950 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1951 cannot be changed at match time, I realized I could make a small
1952 improvement to matching performance by compiling OP_ALLANY instead of
1953 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1954 on the OP_ANY path.
1955
1956 17. Compiling pcretest on Windows with readline support failed without the
1957 following two fixes: (1) Make the unistd.h include conditional on
1958 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1959
1960 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1961 ncurses library to be included for pcretest when ReadLine support is
1962 requested, but also to allow for it to be overridden. This patch came from
1963 Daniel Bergström.
1964
1965 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1966 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1967 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1968 spotting this.
1969
1970
1971 Version 7.6 28-Jan-08
1972 ---------------------
1973
1974 1. A character class containing a very large number of characters with
1975 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1976 overflow.
1977
1978 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1979 HAVE_LONG_LONG is not defined.
1980
1981 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1982 bring them up to date and include new features. This patch includes:
1983
1984 - Fixed PH's badly added libz and libbz2 support.
1985 - Fixed a problem with static linking.
1986 - Added pcredemo. [But later removed - see 7 below.]
1987 - Fixed dftables problem and added an option.
1988 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1989 HAVE_LONG_LONG.
1990 - Added readline support for pcretest.
1991 - Added an listing of the option settings after cmake has run.
1992
1993 4. A user submitted a patch to Makefile that makes it easy to create
1994 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1995 Makefile.am that cause it to include this special target, without
1996 affecting anything else. Note that the same mingw target plus all
1997 the other distribution libraries and programs are now supported
1998 when configuring with CMake (see 6 below) instead of with
1999 Configure/Make.
2000
2001 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2002 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2003 exported in the Windows port". It has not yet been confirmed that the patch
2004 solves the problem, but it does no harm.
2005
2006 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2007 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2008 with CMake, and also correct the comment about stack recursion.
2009
2010 7. Remove the automatic building of pcredemo from the ./configure system and
2011 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2012 of a program that users should build themselves after PCRE is installed, so
2013 building it automatically is not really right. What is more, it gave
2014 trouble in some build environments.
2015
2016 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2017
2018
2019 Version 7.5 10-Jan-08
2020 ---------------------
2021
2022 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2023 values in parens when parsing an RE using the C++ wrapper."
2024
2025 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2026 Characters greater than 255 were excluded from the class instead of being
2027 included.
2028
2029 3. The same bug as (2) above applied to negated POSIX classes such as
2030 [:^space:].
2031
2032 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2033 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2034 I have changed it.
2035
2036 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2037 first named subpattern) and a construct such as (?&a) would reference the
2038 first named subpattern whose name started with "a" (in other words, the
2039 length check was missing). Both these problems are fixed. "Subpattern name
2040 expected" is now given for (?&) (a zero-length name), and this patch also
2041 makes it give the same error for \k'' (previously it complained that that
2042 was a reference to a non-existent subpattern).
2043
2044 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2045 this is right because (?- can be followed by option settings as well as by
2046 digits. I have, however, made the messages clearer.
2047
2048 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2049 than the number used in the conditional) now cause a compile-time error.
2050 This is actually not compatible with Perl, which accepts such patterns, but
2051 treats the conditional as always being FALSE (as PCRE used to), but it
2052 seems to me that giving a diagnostic is better.
2053
2054 8. Change "alphameric" to the more common word "alphanumeric" in comments
2055 and messages.
2056
2057 9. Fix two occurrences of "backslash" in comments that should have been
2058 "backspace".
2059
2060 10. Remove two redundant lines of code that can never be obeyed (their function
2061 was moved elsewhere).
2062
2063 11. The program that makes PCRE's Unicode character property table had a bug
2064 which caused it to generate incorrect table entries for sequences of
2065 characters that have the same character type, but are in different scripts.
2066 It amalgamated them into a single range, with the script of the first of
2067 them. In other words, some characters were in the wrong script. There were
2068 thirteen such cases, affecting characters in the following ranges:
2069
2070 U+002b0 - U+002c1
2071 U+0060c - U+0060d
2072 U+0061e - U+00612
2073 U+0064b - U+0065e
2074 U+0074d - U+0076d
2075 U+01800 - U+01805
2076 U+01d00 - U+01d77
2077 U+01d9b - U+01dbf
2078 U+0200b - U+0200f
2079 U+030fc - U+030fe
2080 U+03260 - U+0327f
2081 U+0fb46 - U+0fbb1
2082 U+10450 - U+1049d
2083
2084 12. The -o option (show only the matching part of a line) for pcregrep was not
2085 compatible with GNU grep in that, if there was more than one match in a
2086 line, it showed only the first of them. It now behaves in the same way as
2087 GNU grep.
2088
2089 13. If the -o and -v options were combined for pcregrep, it printed a blank
2090 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2091 does the same. The return code can be used to tell if there were any
2092 non-matching lines.
2093
2094 14. Added --file-offsets and --line-offsets to pcregrep.
2095
2096 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2097 infinitely looping recursion. The bug was that positive lookaheads were not
2098 being skipped when checking for a possible empty match (negative lookaheads
2099 and both kinds of lookbehind were skipped).
2100
2101 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2102 inclusion of <windows.h> to before rather than after the definition of
2103 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2104
2105 17. Specifying a possessive quantifier with a specific limit for a Unicode
2106 character property caused pcre_compile() to compile bad code, which led at
2107 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2108 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2109 caused the error; without that there was no problem.
2110
2111 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2112
2113 19. Added --enable-pcretest-libreadline.
2114
2115 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2116 RE::GlobalReplace(). As a result, the number of replacements returned was
2117 double what it should be. I removed one of the increments, but Craig sent a
2118 later patch that removed the other one (the right fix) and added unit tests
2119 that check the return values (which was not done before).
2120
2121 21. Several CMake things:
2122
2123 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2124 the names libpcre and libpcreposix, not just pcre and pcreposix.
2125
2126 (2) The above change means that pcretest and pcregrep are now correctly
2127 linked with the newly-built libraries, not previously installed ones.
2128
2129 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2130
2131 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2132 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2133 UTF-8 newline character). The key issue is that the pattern starts .*;
2134 this means that the match must be either at the beginning, or after a
2135 newline. The bug was in the code for advancing after a failed match and
2136 checking that the new position followed a newline. It was not taking
2137 account of UTF-8 characters correctly.
2138
2139 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2140 character classes. PCRE was not treating the sequence [:...:] as a
2141 character class unless the ... were all letters. Perl, however, seems to
2142 allow any characters between [: and :], though of course it rejects as
2143 unknown any "names" that contain non-letters, because all the known class
2144 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2145 for example, whereas PCRE did not - it did not recognize a POSIX character
2146 class. This seemed a bit dangerous, so the code has been changed to be
2147 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2148 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2149 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2150 Perl does, and where it didn't before.
2151
2152 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2153 Windows environments %n is disabled by default.
2154
2155
2156 Version 7.4 21-Sep-07
2157 ---------------------
2158
2159 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2160 means that a class such as [\s] counted as "explicit reference to CR or
2161 LF". That isn't really right - the whole point of the change was to try to
2162 help when there was an actual mention of one of the two characters. So now
2163 the change happens only if \r or \n (or a literal CR or LF) character is
2164 encountered.
2165
2166 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2167 of both had grown to the point where there were only 3 bits left.
2168 Fortunately, there was spare space in the data structure, and so I have
2169 moved the internal flags into a new 16-bit field to free up more option
2170 bits.
2171
2172 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2173 but did not set the internal JCHANGED flag - either of these is enough to
2174 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2175 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2176 start sets both bits.
2177
2178 4. Added options (at build time, compile time, exec time) to change \R from
2179 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2180
2181 5. doc/pcresyntax.html was missing from the distribution.
2182
2183 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2184 compatibility, even though it is no longer used.
2185
2186 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2187 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2188 windows.h file is present (where different names are used). [This was
2189 reversed later after testing - see 16 below.]
2190
2191 8. Changed all #include <config.h> to #include "config.h". There were also
2192 some further <pcre.h> cases that I changed to "pcre.h".
2193
2194 9. When pcregrep was used with the --colour option, it missed the line ending
2195 sequence off the lines that it output.
2196
2197 10. It was pointed out to me that arrays of string pointers cause lots of
2198 relocations when a shared library is dynamically loaded. A technique of
2199 using a single long string with a table of offsets can drastically reduce
2200 these. I have refactored PCRE in four places to do this. The result is
2201 dramatic:
2202
2203 Originally: 290
2204 After changing UCP table: 187
2205 After changing error message table: 43
2206 After changing table of "verbs" 36
2207 After changing table of Posix names 22
2208
2209 Thanks to the folks working on Gregex for glib for this insight.
2210
2211 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2212 unicode-properties was also set.
2213
2214 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2215
2216 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2217 checked only for CRLF.
2218
2219 14. Added casts to pcretest.c to avoid compiler warnings.
2220
2221 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2222
2223 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2224 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2225 entirely. This removes changes made in 7 above.
2226
2227 17. The CMake files have been updated, and there is now more information about
2228 building with CMake in the NON-UNIX-USE document.
2229
2230
2231 Version 7.3 28-Aug-07
2232 ---------------------
2233
2234 1. In the rejigging of the build system that eventually resulted in 7.1, the
2235 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2236 brackets there is not right, since it causes compilers to look for an
2237 installed pcre.h, not the version that is in the source that is being
2238 compiled (which of course may be different). I have changed it back to:
2239
2240 #include "pcre.h"
2241
2242 I have a vague recollection that the change was concerned with compiling in
2243 different directories, but in the new build system, that is taken care of
2244 by the VPATH setting the Makefile.
2245
2246 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2247 when the subject happened to end in the byte 0x85 (e.g. if the last
2248 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2249 characters but of course it shouldn't be taken as a newline when it is part
2250 of another character. The bug was that, for an unlimited repeat of . in
2251 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2252 characters when looking for a newline.
2253
2254 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2255
2256 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2257 in debug output.
2258
2259 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2260 long printing in the pcrecpp unittest when running under MinGW.
2261
2262 6. ESC_K was left out of the EBCDIC table.
2263
2264 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2265 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2266 limit also applies to "virtual nesting" when a pattern is recursive, and in
2267 this case 1000 isn't so big. I have been able to remove this limit at the
2268 expense of backing off one optimization in certain circumstances. Normally,
2269 when pcre_exec() would call its internal match() function recursively and
2270 immediately return the result unconditionally, it uses a "tail recursion"
2271 feature to save stack. However, when a subpattern that can match an empty
2272 string has an unlimited repetition quantifier, it no longer makes this
2273 optimization. That gives it a stack frame in which to save the data for
2274 checking that an empty string has been matched. Previously this was taken
2275 from the 1000-entry workspace that had been reserved. So now there is no
2276 explicit limit, but more stack is used.
2277
2278 8. Applied Daniel's patches to solve problems with the import/export magic
2279 syntax that is required for Windows, and which was going wrong for the
2280 pcreposix and pcrecpp parts of the library. These were overlooked when this
2281 problem was solved for the main library.
2282
2283 9. There were some crude static tests to avoid integer overflow when computing
2284 the size of patterns that contain repeated groups with explicit upper
2285 limits. As the maximum quantifier is 65535, the maximum group length was
2286 set at 30,000 so that the product of these two numbers did not overflow a
2287 32-bit integer. However, it turns out that people want to use groups that
2288 are longer than 30,000 bytes (though not repeat them that many times).
2289 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2290 made it possible to implement the integer overflow checks in a much more
2291 dynamic way, which I have now done. The artificial limitation on group
2292 length has been removed - we now have only the limit on the total length of
2293 the compiled pattern, which depends on the LINK_SIZE setting.
2294
2295 10. Fixed a bug in the documentation for get/copy named substring when
2296 duplicate names are permitted. If none of the named substrings are set, the
2297 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2298 empty string.
2299
2300 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2301 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2302 because the ] is interpreted as the first data character and the
2303 terminating ] is not found. PCRE has been made compatible with Perl in this
2304 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2305 cause memory overwriting.
2306
2307 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2308 string has been matched (to stop an infinite loop). It was not recognizing
2309 a conditional subpattern that could match an empty string if that
2310 subpattern was within another subpattern. For example, it looped when
2311 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2312 condition was not nested. This bug has been fixed.
2313
2314 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2315 past the start of the subject in the presence of bytes with the top bit
2316 set, for example "\x8aBCD".
2317
2318 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2319 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2320
2321 14. Optimized (?!) to (*FAIL).
2322
2323 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2324 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2325 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2326 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2327 does: it's just the validity check that is more restrictive.
2328
2329 16. Inserted checks for integer overflows during escape sequence (backslash)
2330 processing, and also fixed erroneous offset values for syntax errors during
2331 backslash processing.
2332
2333 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2334 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2335
2336 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2337 caused an overrun.
2338
2339 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2340 something other than just ASCII characters) inside a group that had an
2341 unlimited repeat caused a loop at compile time (while checking to see
2342 whether the group could match an empty string).
2343
2344 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2345 [\P{Any}] did so. (Error in the code for printing property names.)
2346
2347 21. An orphan \E inside a character class could cause a crash.
2348
2349 22. A repeated capturing bracket such as (A)? could cause a wild memory
2350 reference during compilation.
2351
2352 23. There are several functions in pcre_compile() that scan along a compiled
2353 expression for various reasons (e.g. to see if it's fixed length for look
2354 behind). There were bugs in these functions when a repeated \p or \P was
2355 present in the pattern. These operators have additional parameters compared
2356 with \d, etc, and these were not being taken into account when moving along
2357 the compiled data. Specifically:
2358
2359 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2360 length.
2361
2362 (b) An item such as \pL+ within a repeated group could cause crashes or
2363 loops.
2364
2365 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2366 "reference to non-existent subpattern" error.
2367
2368 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2369
2370 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2371 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2372
2373 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2374
2375 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2376 character were causing crashes (broken optimization).
2377
2378 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2379 \p or \P) caused a compile-time loop.
2380
2381 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2382 break. For example, the unstudied pattern [\r\n]A does not match the string
2383 "\r\nA" because change 7.0/46 below moves the current point on by two
2384 characters after failing to match at the start. However, the pattern \nA
2385 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2386 the same is true. There doesn't seem any very clean way out of this, but
2387 what I have chosen to do makes the common cases work: PCRE now takes note
2388 of whether there can be an explicit match for \r or \n anywhere in the
2389 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2390 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2391 pattern has explicit CR or LF references.
2392
2393 29. Added (*CR) etc for changing newline setting at start of pattern.
2394
2395
2396 Version 7.2 19-Jun-07
2397 ---------------------
2398
2399 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2400 which is apparently normally available under Windows.
2401
2402 2. Re-jig the pcregrep tests with different newline settings in an attempt
2403 to make them independent of the local environment's newline setting.
2404
2405 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2406
2407 4. Some of the "internals" tests were previously cut out when the link size
2408 was not 2, because the output contained actual offsets. The recent new
2409 "Z" feature of pcretest means that these can be cut out, making the tests
2410 usable with all link sizes.
2411
2412 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2413 stack recursion. This gives a massive performance boost under BSD, but just
2414 a small improvement under Linux. However, it saves one field in the frame
2415 in all cases.
2416
2417 6. Added more features from the forthcoming Perl 5.10:
2418
2419 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2420 recursion call. It refers to the nth most recently opened parentheses.
2421
2422 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2423 to be opened parentheses.
2424
2425 (c) Conditions that refer to capturing parentheses can be specified
2426 relatively, for example, (?(-2)... or (?(+3)...
2427
2428 (d) \K resets the start of the current match so that everything before
2429 is not part of it.
2430
2431 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2432
2433 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2434 reference syntax.
2435
2436 (g) (?| introduces a group in which the numbering of parentheses in each
2437 alternative starts with the same number.
2438
2439 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2440
2441 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2442 PCRE_INFO_JCHANGED.
2443
2444 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2445 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2446 for detecting groups that can match an empty string.
2447
2448 9. A pattern with a very large number of alternatives (more than several
2449 hundred) was running out of internal workspace during the pre-compile
2450 phase, where pcre_compile() figures out how much memory will be needed. A
2451 bit of new cunning has reduced the workspace needed for groups with
2452 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2453 workspace instead of running out of the 4096 that are available.
2454
2455 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2456
2457 11. Applied patch from Google to remove an optimization that didn't quite work.
2458 The report of the bug said:
2459
2460 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2461 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2462 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2463
2464 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2465 it matched the wrong number of bytes.
2466
2467
2468 Version 7.1 24-Apr-07
2469 ---------------------
2470
2471 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2472 that is more "standard", making use of automake and other Autotools. There
2473 is some re-arrangement of the files and adjustment of comments consequent
2474 on this.
2475
2476 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2477 for recursive directory scanning broke on some systems because the files
2478 are not scanned in any specific order and on different systems the order
2479 was different. A call to "sort" has been inserted into RunGrepTest for the
2480 approprate test as a short-term fix. In the longer term there may be an
2481 alternative.
2482
2483 3. I had an email from Eric Raymond about problems translating some of PCRE's
2484 man pages to HTML (despite the fact that I distribute HTML pages, some
2485 people do their own conversions for various reasons). The problems
2486 concerned the use of low-level troff macros .br and .in. I have therefore
2487 removed all such uses from the man pages (some were redundant, some could
2488 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2489 HTML has been updated to handle .nf/.fi and to complain if it encounters
2490 .br or .in.
2491
2492 4. Updated comments in configure.ac that get placed in config.h.in and also
2493 arranged for config.h to be included in the distribution, with the name
2494 config.h.generic, for the benefit of those who have to compile without
2495 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2496
2497 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2498 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2499 makevp.bat for the current PCRE, using the additional files
2500 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2501
2502 6. A Windows user reported a minor discrepancy with test 2, which turned out
2503 to be caused by a trailing space on an input line that had got lost in his
2504 copy. The trailing space was an accident, so I've just removed it.
2505
2506 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2507 that is needed.
2508
2509 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2510 as "const" (a) because they are and (b) because it helps the PHP
2511 maintainers who have recently made a script to detect big data structures
2512 in the php code that should be moved to the .rodata section. I remembered
2513 to update Builducptable as well, so it won't revert if ucptable.h is ever
2514 re-created.
2515
2516 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2517 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2518 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2519 support is not required. This saves 1.5-2K of code, which is important in
2520 some applications.
2521
2522 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2523 so as not to refer to the tables, even though these functions will never be
2524 called when UTF-8 support is disabled. Otherwise there are problems with a
2525 shared library.
2526
2527 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2528
2529 (a) It was defining its arguments as char * instead of void *.
2530
2531 (b) It was assuming that all moves were upwards in memory; this was true
2532 a long time ago when I wrote it, but is no longer the case.
2533
2534 The emulated memove() is provided for those environments that have neither
2535 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2536 is clearly not the case, as these two bugs were recently reported.
2537
2538 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2539 and Detrail to create the HTML documentation, the .txt form of the man
2540 pages, and it removes trailing spaces from listed files. It also creates
2541 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2542 case, it wraps all the #defines with #ifndefs. This script should be run
2543 before "make dist".
2544
2545 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2546 with Unicode property support.
2547
2548 (a) For a maximizing quantifier, if the two different cases of the
2549 character were of different lengths in their UTF-8 codings (there are
2550 some cases like this - I found 11), and the matching function had to
2551 back up over a mixture of the two cases, it incorrectly assumed they
2552 were both the same length.
2553
2554 (b) When PCRE was configured to use the heap rather than the stack for
2555 recursion during matching, it was not correctly preserving the data for
2556 the other case of a UTF-8 character when checking ahead for a match
2557 while processing a minimizing repeat. If the check also involved
2558 matching a wide character, but failed, corruption could cause an
2559 erroneous result when trying to check for a repeat of the original
2560 character.
2561
2562 13. Some tidying changes to the testing mechanism:
2563
2564 (a) The RunTest script now detects the internal link size and whether there
2565 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2566 values substituted by "configure". (The RunGrepTest script already did
2567 this for UTF-8.) The configure.ac script no longer substitutes the
2568 relevant variables.
2569
2570 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2571 with length and offset values. This means that the output is different
2572 for different internal link sizes. Test 2 is skipped for link sizes
2573 other than 2 because of this, bypassing the problem. Unfortunately,
2574 there was also a test in test 3 (the locale tests) that used /B and
2575 failed for link sizes other than 2. Rather than cut the whole test out,
2576 I have added a new /Z option to pcretest that replaces the length and
2577 offset values with spaces. This is now used to make test 3 independent
2578 of link size. (Test 2 will be tidied up later.)
2579
2580 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2581 segmentation fault instead of returning the appropriate error message.
2582
2583 15. In multiline mode when the newline sequence was set to "any", the pattern
2584 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2585 This doesn't seem right; it now treats the CRLF combination as the line
2586 ending, and so does not match in that case. It's only a pattern such as ^$
2587 that would hit this one: something like ^ABC$ would have failed after \r
2588 and then tried again after \r\n.
2589
2590 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2591 in an attempt to make files that differ only in their line terminators
2592 compare equal. This works on Linux.
2593
2594 17. Under certain error circumstances pcregrep might try to free random memory
2595 as it exited. This is now fixed, thanks to valgrind.
2596
2597 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2598 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2599 was because its rules for how to advance for /g after matching an empty
2600 string at the end of a line did not allow for this case. They now check for
2601 it specially.
2602
2603 20. pcretest is supposed to handle patterns and data of any length, by
2604 extending its buffers when necessary. It was getting this wrong when the
2605 buffer for a data line had to be extended.
2606
2607 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2608 CRLF as a newline sequence.
2609
2610 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2611 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2612 I have nevertheless tidied it up.
2613
2614 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2615
2616 24. Added a man page for pcre-config.
2617
2618
2619 Version 7.0 19-Dec-06
2620 ---------------------
2621
2622 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2623 moving to gcc 4.1.1.
2624
2625 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2626 sys/time.h, which is documented as needed for this function. It doesn't
2627 seem to matter on Linux, but it showed up on some releases of OS X.
2628
2629 3. It seems that there are systems where bytes whose values are greater than
2630 127 match isprint() in the "C" locale. The "C" locale should be the
2631 default when a C program starts up. In most systems, only ASCII printing
2632 characters match isprint(). This difference caused the output from pcretest
2633 to vary, making some of the tests fail. I have changed pcretest so that:
2634
2635 (a) When it is outputting text in the compiled version of a pattern, bytes
2636 other than 32-126 are always shown as hex escapes.
2637
2638 (b) When it is outputting text that is a matched part of a subject string,
2639 it does the same, unless a different locale has been set for the match
2640 (using the /L modifier). In this case, it uses isprint() to decide.
2641
2642 4. Fixed a major bug that caused incorrect computation of the amount of memory
2643 required for a compiled pattern when options that changed within the
2644 pattern affected the logic of the preliminary scan that determines the
2645 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2646 that the computed length was too small. The symptoms of this bug were
2647 either the PCRE error "internal error: code overflow" from pcre_compile(),
2648 or a glibc crash with a message such as "pcretest: free(): invalid next
2649 size (fast)". Examples of patterns that provoked this bug (shown in
2650 pcretest format) are:
2651
2652 /(?-x: )/x
2653 /(?x)(?-x: \s*#\s*)/
2654 /((?i)[\x{c0}])/8
2655 /(?i:[\x{c0}])/8
2656
2657 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2658 is now done differently.
2659
2660 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2661 wrapper classes; (b) implement a new function in the C++ scanner that is
2662 more efficient than the old way of doing things because it avoids levels of
2663 recursion in the regex matching; (c) add a paragraph to the documentation
2664 for the FullMatch() function.
2665
2666 6. The escape sequence \n was being treated as whatever was defined as
2667 "newline". Not only was this contrary to the documentation, which states
2668 that \n is character 10 (hex 0A), but it also went horribly wrong when
2669 "newline" was defined as CRLF. This has been fixed.
2670
2671 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2672 was being set to -1 for the "end of line" case (supposedly a value that no
2673 character can have). Though this value is never used (the check for end of
2674 line is "zero bytes in current character"), it caused compiler complaints.
2675 I've changed it to 0xffffffff.
2676
2677 8. In pcre_version.c, the version string was being built by a sequence of
2678 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2679 string (as it is for production releases) called a macro with an empty
2680 argument. The C standard says the result of this is undefined. The gcc
2681 compiler treats it as an empty string (which was what was wanted) but it is
2682 reported that Visual C gives an error. The source has been hacked around to
2683 avoid this problem.
2684
2685 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2686 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2687 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2688 of them did).
2689
2690 10. Originally, pcretest opened its input and output without "b"; then I was
2691 told that "b" was needed in some environments, so it was added for release
2692 5.0 to both the input and output. (It makes no difference on Unix-like
2693 systems.) Later I was told that it is wrong for the input on Windows. I've
2694 now abstracted the modes into two macros, to make it easier to fiddle with
2695 them, and removed "b" from the input mode under Windows.
2696
2697 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2698
2699 12. Added -help and --help to pcretest as an official way of being reminded
2700 of the options.
2701
2702 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2703 and pcrecpp.cc because they annoy compilers at high warning levels.
2704
2705 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2706
2707 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2708 scripts are not C programs :-) and which was not noticed because it works
2709 on Linux.
2710
2711 16. pcretest is supposed to handle any length of pattern and data line (as one
2712 line or as a continued sequence of lines) by extending its input buffer if
2713 necessary. This feature was broken for very long pattern lines, leading to
2714 a string of junk being passed to pcre_compile() if the pattern was longer
2715 than about 50K.
2716
2717 17. I have done a major re-factoring of the way pcre_compile() computes the
2718 amount of memory needed for a compiled pattern. Previously, there was code
2719 that made a preliminary scan of the pattern in order to do this. That was
2720 OK when PCRE was new, but as the facilities have expanded, it has become
2721 harder and harder to keep it in step with the real compile phase, and there
2722 have been a number of bugs (see for example, 4 above). I have now found a
2723 cunning way of running the real compile function in a "fake" mode that
2724 enables it to compute how much memory it would need, while actually only
2725 ever using a few hundred bytes of working memory and without too many
2726 tests of the mode. This should make future maintenance and development
2727 easier. A side effect of this work is that the limit of 200 on the nesting
2728 depth of parentheses has been removed (though this was never a serious
2729 limitation, I suspect). However, there is a downside: pcre_compile() now
2730 runs more slowly than before (30% or more, depending on the pattern). I
2731 hope this isn't a big issue. There is no effect on runtime performance.
2732
2733 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2734 newline (only possible for the last line of a file) and it was a
2735 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2736
2737 19. Added additional timing features to pcretest. (1) The -tm option now times
2738 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2739 separate command line item, by a number that specifies the number of
2740 repeats to use when timing. The default is 50000; this gives better
2741 precision, but takes uncomfortably long for very large patterns.
2742
2743 20. Extended pcre_study() to be more clever in cases where a branch of a
2744 subpattern has no definite first character. For example, (a*|b*)[cd] would
2745 previously give no result from pcre_study(). Now it recognizes that the
2746 first character must be a, b, c, or d.
2747
2748 21. There was an incorrect error "recursive call could loop indefinitely" if
2749 a subpattern (or the entire pattern) that was being tested for matching an
2750 empty string contained only one non-empty item after a nested subpattern.
2751 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2752 incorrectly, because the \d was being skipped in the check.
2753
2754 22. The pcretest program now has a new pattern option /B and a command line
2755 option -b, which is equivalent to adding /B to every pattern. This causes
2756 it to show the compiled bytecode, without the additional information that
2757 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2758 is the same as /B/I).
2759
2760 23. A new optimization is now able automatically to treat some sequences such
2761 as a*b as a*+b. More specifically, if something simple (such as a character
2762 or a simple class like \d) has an unlimited quantifier, and is followed by
2763 something that cannot possibly match the quantified thing, the quantifier
2764 is automatically "possessified".
2765
2766 24. A recursive reference to a subpattern whose number was greater than 39
2767 went wrong under certain circumstances in UTF-8 mode. This bug could also
2768 have affected the operation of pcre_study().
2769
2770 25. Realized that a little bit of performance could be had by replacing
2771 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2772
2773 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2774
2775 27. Possessive quantifiers such as a++ were previously implemented by turning
2776 them into atomic groups such as ($>a+). Now they have their own opcodes,
2777 which improves performance. This includes the automatically created ones
2778 from 23 above.
2779
2780 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2781 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2782 the first matched character to be a colon. This applied both to named and
2783 numbered groups.
2784
2785 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2786
2787 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2788 building PCRE on Windows easier, so I have included it in the distribution.
2789
2790 31. There is now a check in pcretest against a ridiculously large number being
2791 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2792 loop, the loop is abandoned.
2793
2794 32. Forward references to subpatterns in conditions such as (?(2)...) where
2795 subpattern 2 is defined later cause pcre_compile() to search forwards in
2796 the pattern for the relevant set of parentheses. This search went wrong
2797 when there were unescaped parentheses in a character class, parentheses
2798 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2799
2800 33. "Subroutine" calls and backreferences were previously restricted to
2801 referencing subpatterns earlier in the regex. This restriction has now
2802 been removed.
2803
2804 34. Added a number of extra features that are going to be in Perl 5.10. On the
2805 whole, these are just syntactic alternatives for features that PCRE had
2806 previously implemented using the Python syntax or my own invention. The
2807 other formats are all retained for compatibility.
2808
2809 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2810 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2811 also .NET compatible.
2812
2813 (b) A recursion or subroutine call to a named group can now be defined as
2814 (?&name) as well as (?P>name).
2815
2816 (c) A backreference to a named group can now be defined as \k<name> or
2817 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2818 5.10, are also .NET compatible.
2819
2820 (d) A conditional reference to a named group can now use the syntax
2821 (?(<name>) or (?('name') as well as (?(name).
2822
2823 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2824 groups (named and numbered) that are never evaluated inline, but can be
2825 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2826 is always false. There may be only one alternative in such a group.
2827
2828 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2829 as the simple (?(R). The condition is true only if the most recent
2830 recursion is that of the given number or name. It does not search out
2831 through the entire recursion stack.
2832
2833 (g) The escape \gN or \g{N} has been added, where N is a positive or
2834 negative number, specifying an absolute or relative reference.
2835
2836 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2837 some "unreachable code" warnings.
2838
2839 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2840 things, this adds five new scripts.
2841
2842 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2843 There were also incompatibilities regarding the handling of \Q..\E inside
2844 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2845 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2846
2847 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2848 matches an empty string, and forcibly breaks the loop. There were bugs in
2849 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2850 against aaaa the result was just "a" rather than "aaaa", for example. Two
2851 separate and independent bugs (that affected different cases) have been
2852 fixed.
2853
2854 39. Refactored the code to abolish the use of different opcodes for small
2855 capturing bracket numbers. This is a tidy that I avoided doing when I
2856 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2857 The new approach is not only tidier, it makes it possible to reduce the
2858 memory needed to fix the previous bug (38).
2859
2860 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2861 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2862 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2863 mode.
2864
2865 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2866 report.
2867
2868 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2869 copy construction and assignment in the C++ wrapper.
2870
2871 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2872 couple of bugs that could have given wrong results in the "--newline=crlf"
2873 case.
2874
2875 44. Added a number of casts and did some reorganization of signed/unsigned int
2876 variables following suggestions from Dair Grant. Also renamed the variable
2877 "this" as "item" because it is a C++ keyword.
2878
2879 45. Arranged for dftables to add
2880
2881 #include "pcre_internal.h"
2882
2883 to pcre_chartables.c because without it, gcc 4.x may remove the array
2884 definition from the final binary if PCRE is built into a static library and
2885 dead code stripping is activated.
2886
2887 46. For an unanchored pattern, if a match attempt fails at the start of a
2888 newline sequence, and the newline setting is CRLF or ANY, and the next two
2889 characters are CRLF, advance by two characters instead of one.
2890
2891
2892 Version 6.7 04-Jul-06
2893 ---------------------
2894
2895 1. In order to handle tests when input lines are enormously long, pcretest has
2896 been re-factored so that it automatically extends its buffers when
2897 necessary. The code is crude, but this _is_ just a test program. The
2898 default size has been increased from 32K to 50K.
2899
2900 2. The code in pcre_study() was using the value of the re argument before
2901 testing it for NULL. (Of course, in any sensible call of the function, it
2902 won't be NULL.)
2903
2904 3. The memmove() emulation function in pcre_internal.h, which is used on
2905 systems that lack both memmove() and bcopy() - that is, hardly ever -
2906 was missing a "static" storage class specifier.
2907
2908 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2909 containing an extended class (one that cannot be represented by a bitmap
2910 because it contains high-valued characters or Unicode property items, e.g.
2911 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2912 pattern, but PCRE should not loop if you do not (it no longer does).
2913 [Detail: two cases were found: (a) a repeated subpattern containing an
2914 extended class; (b) a recursive reference to a subpattern that followed a
2915 previous extended class. It wasn't skipping over the extended class
2916 correctly when UTF-8 mode was not set.]
2917
2918 5. A negated single-character class was not being recognized as fixed-length
2919 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2920 compile error "lookbehind assertion is not fixed length".
2921
2922 6. The RunPerlTest auxiliary script was showing an unexpected difference
2923 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2924 write a Perl script that can interpret lines of an input file either as
2925 byte characters or as UTF-8, which is what "perltest" was being required to
2926 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2927 can't do is switch easily at run time between having the "use utf8;" pragma
2928 or not. In the end, I fudged it by using the RunPerlTest script to insert
2929 "use utf8;" explicitly for the UTF-8 tests.
2930
2931 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2932 the end of the subject string, contrary to the documentation and to what
2933 Perl does. This was true of both matching functions. Now it matches only at
2934 the start of the subject and immediately after *internal* newlines.
2935
2936 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2937 a pointer to an int instead of a pointer to an unsigned long int. This
2938 caused problems on 64-bit systems.
2939
2940 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2941 instance of the 'standard' template library not being so standard".
2942
2943 10. There was no check on the number of named subpatterns nor the maximum
2944 length of a subpattern name. The product of these values is used to compute
2945 the size of the memory block for a compiled pattern. By supplying a very
2946 long subpattern name and a large number of named subpatterns, the size
2947 computation could be caused to overflow. This is now prevented by limiting
2948 the length of names to 32 characters, and the number of named subpatterns
2949 to 10,000.
2950
2951 11. Subpatterns that are repeated with specific counts have to be replicated in
2952 the compiled pattern. The size of memory for this was computed from the
2953 length of the subpattern and the repeat count. The latter is limited to
2954 65535, but there was no limit on the former, meaning that integer overflow
2955 could in principle occur. The compiled length of a repeated subpattern is
2956 now limited to 30,000 bytes in order to prevent this.
2957
2958 12. Added the optional facility to have named substrings with the same name.
2959
2960 13. Added the ability to use a named substring as a condition, using the
2961 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2962 are numbers (not recommended). Forward references are permitted.
2963
2964 14. Added forward references in named backreferences (if you see what I mean).
2965
2966 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2967 pattern could run off the end of the subject. For example, the pattern
2968 "(?s)(.{1,5})"8 did this with the subject "ab".
2969
2970 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2971 PCRE_CASELESS was set when matching characters that were quantified with ?
2972 or *.
2973
2974 17. A character class other than a single negated character that had a minimum
2975 but no maximum quantifier - for example [ab]{6,} - was not handled
2976 correctly by pce_dfa_exec(). It would match only one character.
2977
2978 18. A valid (though odd) pattern that looked like a POSIX character
2979 class but used an invalid character after [ (for example [[,abc,]]) caused
2980 pcre_compile() to give the error "Failed: internal error: code overflow" or
2981 in some cases to crash with a glibc free() error. This could even happen if
2982 the pattern terminated after [[ but there just happened to be a sequence of
2983 letters, a binary zero, and a closing ] in the memory that followed.
2984
2985 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2986 over the years. Originally (before any Unicode support), just the bottom 8
2987 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2988 output from "man perlunicode" includes this:
2989
2990 The regular expression compiler produces polymorphic opcodes. That
2991 is, the pattern adapts to the data and automatically switches to
2992 the Unicode character scheme when presented with Unicode data--or
2993 instead uses a traditional byte scheme when presented with byte
2994 data.
2995
2996 Sadly, a wide octal escape does not cause a switch, and in a string with
2997 no other multibyte characters, these octal escapes are treated as before.
2998 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2999 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3000 Unicode string.
3001
3002 I have not perpetrated such confusion in PCRE. Up till now, it took just
3003 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3004 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3005 translate to the appropriate multibyte character.
3006
3007 29. Applied some refactoring to reduce the number of warnings from Microsoft
3008 and Borland compilers. This has included removing the fudge introduced
3009 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3010 a warning about an unused variable.
3011
3012 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3013 characters since release 4.0, because Perl (from release 5.004) does not.
3014 [Or at least, is documented not to: some releases seem to be in conflict
3015 with the documentation.] However, when a pattern was studied with
3016 pcre_study() and all its branches started with \s, PCRE still included VT
3017 as a possible starting character. Of course, this did no harm; it just
3018 caused an unnecessary match attempt.
3019
3020 22. Removed a now-redundant internal flag bit that recorded the fact that case
3021 dependency changed within the pattern. This was once needed for "required
3022 byte" processing, but is no longer used. This recovers a now-scarce options
3023 bit. Also moved the least significant internal flag bit to the most-
3024 significant bit of the word, which was not previously used (hangover from
3025 the days when it was an int rather than a uint) to free up another bit for
3026 the future.
3027
3028 23. Added support for CRLF line endings as well as CR and LF. As well as the
3029 default being selectable at build time, it can now be changed at runtime
3030 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3031 specify that it is scanning data with non-default line endings.
3032
3033 24. Changed the definition of CXXLINK to make it agree with the definition of
3034 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3035
3036 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3037 recursions. This makes a big different to stack usage for some patterns.
3038
3039 26. If a subpattern containing a named recursion or subroutine reference such
3040 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3041 the space required for the compiled pattern went wrong and gave too small a
3042 value. Depending on the environment, this could lead to "Failed: internal
3043 error: code overflow at offset 49" or "glibc detected double free or
3044 corruption" errors.
3045
3046 27. Applied patches from Google (a) to support the new newline modes and (b) to
3047 advance over multibyte UTF-8 characters in GlobalReplace.
3048
3049 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3050 difference for some implementation of PCRE in some Windows version.
3051
3052 29. Added some extra testing facilities to pcretest:
3053
3054 \q<number> in a data line sets the "match limit" value
3055 \Q<number> in a data line sets the "match recursion limt" value
3056 -S <number> sets the stack size, where <number> is in megabytes
3057
3058 The -S option isn't available for Windows.
3059
3060
3061 Version 6.6 06-Feb-06
3062 ---------------------
3063
3064 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3065 in pcreposix.h. I have copied the definition from pcre.h.
3066
3067 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3068 because pcre.h is no longer a built file.
3069
3070 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3071 not normally included in the compiled code.
3072
3073
3074 Version 6.5 01-Feb-06
3075 ---------------------
3076
3077 1. When using the partial match feature with pcre_dfa_exec(), it was not
3078 anchoring the second and subsequent partial matches at the new starting
3079 point. This could lead to incorrect results. For example, with the pattern
3080 /1234/, partially matching against "123" and then "a4" gave a match.
3081
3082 2. Changes to pcregrep:
3083
3084 (a) All non-match returns from pcre_exec() were being treated as failures
3085 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3086 error message is output. Some extra information is given for the
3087 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3088 probably the only errors that are likely to be caused by users (by
3089 specifying a regex that has nested indefinite repeats, for instance).
3090 If there are more than 20 of these errors, pcregrep is abandoned.
3091
3092 (b) A binary zero was treated as data while matching, but terminated the
3093 output line if it was written out. This has been fixed: binary zeroes
3094 are now no different to any other data bytes.
3095
3096 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3097 used to set a locale for matching. The --locale=xxxx long option has
3098 been added (no short equivalent) to specify a locale explicitly on the
3099 pcregrep command, overriding the environment variables.
3100
3101 (d) When -B was used with -n, some line numbers in the output were one less
3102 than they should have been.
3103
3104 (e) Added the -o (--only-matching) option.
3105
3106 (f) If -A or -C was used with -c (count only), some lines of context were
3107 accidentally printed for the final match.
3108
3109 (g) Added the -H (--with-filename) option.
3110
3111 (h) The combination of options -rh failed to suppress file names for files
3112 that were found from directory arguments.
3113
3114 (i) Added the -D (--devices) and -d (--directories) options.
3115
3116 (j) Added the -F (--fixed-strings) option.
3117
3118 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3119
3120 (l) Added the --colo(u)r option.
3121
3122 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3123 is not present by default.
3124
3125 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3126 items such as (?R) or (?1), when the recursion could match a number of
3127 alternatives. If it matched one of the alternatives, but subsequently,
3128 outside the recursion, there was a failure, the code tried to back up into
3129 the recursion. However, because of the way PCRE is implemented, this is not
3130 possible, and the result was an incorrect result from the match.
3131
3132 In order to prevent this happening, the specification of recursion has
3133 been changed so that all such subpatterns are automatically treated as
3134 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3135
3136 4. I had overlooked the fact that, in some locales, there are characters for
3137 which isalpha() is true but neither isupper() nor islower() are true. In
3138 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3139 and ordfeminine) are like this. This affected the treatment of \w and \W
3140 when they appeared in character classes, but not when they appeared outside
3141 a character class. The bit map for "word" characters is now created
3142 separately from the results of isalnum() instead of just taking it from the
3143 upper, lower, and digit maps. (Plus the underscore character, of course.)
3144
3145 5. The above bug also affected the handling of POSIX character classes such as
3146 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3147 permanent tables. Instead, the bit maps for such a class were previously
3148 created as the appropriate unions of the upper, lower, and digit bitmaps.
3149 Now they are created by subtraction from the [[:word:]] class, which has
3150 its own bitmap.
3151
3152 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3153 It is created by subtracting the vertical space characters (\x09, \x0a,
3154 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3155 subtraction was done in the overall bitmap for a character class, meaning
3156 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3157 be recognized. This bug has been fixed.
3158
3159 7. Patches from the folks at Google:
3160
3161 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3162 real life, but is still worth protecting against".
3163
3164 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3165 regular expressions".
3166
3167 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3168 have it.
3169
3170 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3171 "configure" and the latter not, in order to fix a problem somebody had
3172 with compiling the Arg class on HP-UX.
3173
3174 (e) Improve the error-handling of the C++ wrapper a little bit.
3175
3176 (f) New tests for checking recursion limiting.
3177
3178 8. The pcre_memmove() function, which is used only if the environment does not
3179 have a standard memmove() function (and is therefore rarely compiled),
3180 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3181 returning a result (though PCRE never actually uses the result).
3182
3183 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3184 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3185 returned instead of calling malloc() with an overflowing number that would
3186 most likely cause subsequent chaos.
3187
3188 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3189
3190 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3191 with this option is matched, the nmatch and pmatch options of regexec() are
3192 ignored.
3193
3194 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3195 provided in case anyone wants to the the POSIX interface with UTF-8
3196 strings.
3197
3198 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3199 C++ linking (needed for some HP-UX environments).
3200
3201 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3202 (unused parameter) and in the pcre_printint() function (omitted "default"
3203 switch label when the default is to do nothing).
3204
3205 15. Added some code to make it possible, when PCRE is compiled as a C++
3206 library, to replace subject pointers for pcre_exec() with a smart pointer
3207 class, thus making it possible to process discontinuous strings.
3208
3209 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3210 much the same function. They were added by different people who were trying
3211 to make PCRE easy to compile on non-Unix systems. It has been suggested
3212 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3213 for compiling on Windows systems. I have therefore replaced it with
3214 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3215 defaults to "extern" for C or "extern C" for C++, which works fine on
3216 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3217 SCOPE with something explicit in config.h. In addition:
3218
3219 (a) pcreposix.h still had just "extern" instead of either of these macros;
3220 I have replaced it with PCRE_DATA_SCOPE.
3221
3222 (b) Functions such as _pcre_xclass(), which are internal to the library,
3223 but external in the C sense, all had PCRE_EXPORT in their definitions.
3224 This is apparently wrong for the Windows case, so I have removed it.
3225 (It makes no difference on Unix-like systems.)
3226
3227 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3228 of recursive calls to match(). This is different to MATCH_LIMIT because
3229 that limits the total number of calls to match(), not all of which increase
3230 the depth of recursion. Limiting the recursion depth limits the amount of
3231 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3232 when PCRE is compiled, and changed at run time. A patch from Google adds
3233 this functionality to the C++ interface.
3234
3235 18. Changes to the handling of Unicode character properties:
3236
3237 (a) Updated the table to Unicode 4.1.0.
3238
3239 (b) Recognize characters that are not in the table as "Cn" (undefined).
3240
3241 (c) I revised the way the table is implemented to a much improved format
3242 which includes recognition of ranges. It now supports the ranges that
3243 are defined in UnicodeData.txt, and it also amalgamates other
3244 characters into ranges. This has reduced the number of entries in the
3245 table from around 16,000 to around 3,000, thus reducing its size
3246 considerably. I realized I did not need to use a tree structure after
3247 all - a binary chop search is just as efficient. Having reduced the
3248 number of entries, I extended their size from 6 bytes to 8 bytes to
3249 allow for more data.
3250
3251 (d) Added support for Unicode script names via properties such as \p{Han}.
3252
3253 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3254 matching that character.
3255
3256 20. When matching a repeated Unicode property with a minimum greater than zero,
3257 (for example \pL{2,}), PCRE could look past the end of the subject if it
3258 reached it while seeking the minimum number of characters. This could
3259 happen only if some of the characters were more than one byte long, because
3260 there is a check for at least the minimum number of bytes.
3261
3262 21. Refactored the implementation of \p and \P so as to be more general, to
3263 allow for more different types of property in future. This has changed the
3264 compiled form incompatibly. Anybody with saved compiled patterns that use
3265 \p or \P will have to recompile them.
3266
3267 22. Added "Any" and "L&" to the supported property types.
3268
3269 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3270 but give a compile time error if the value is greater than 0xff.
3271
3272 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3273 accidentally not being installed or uninstalled.
3274
3275 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3276 made were to insert the current release number. This seemed silly, because
3277 it made things harder for people building PCRE on systems that don't run
3278 "configure". I have turned pcre.h into a distributed file, no longer built
3279 by "configure", with the version identification directly included. There is
3280 no longer a pcre.h.in file.
3281
3282 However, this change necessitated a change to the pcre-config script as
3283 well. It is built from pcre-config.in, and one of the substitutions was the
3284 release number. I have updated configure.ac so that ./configure now finds
3285 the release number by grepping pcre.h.
3286
3287 26. Added the ability to run the tests under valgrind.
3288
3289
3290 Version 6.4 05-Sep-05
3291 ---------------------
3292
3293 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3294 "--" to be printed when multiple files were scanned, even when none of the
3295 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3296 consider it to be a bug, and have restored the previous behaviour.
3297
3298 2. A couple of code tidies to get rid of compiler warnings.
3299
3300 3. The pcretest program used to cheat by referring to symbols in the library
3301 whose names begin with _pcre_. These are internal symbols that are not
3302 really supposed to be visible externally, and in some environments it is
3303 possible to suppress them. The cheating is now confined to including
3304 certain files from the library's source, which is a bit cleaner.
3305
3306 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3307 file's purpose clearer.
3308
3309 5. Reorganized pcre_ucp_findchar().
3310
3311
3312 Version 6.3 15-Aug-05
3313 ---------------------
3314
3315 1. The file libpcre.pc.in did not have general read permission in the tarball.
3316
3317 2. There were some problems when building without C++ support:
3318
3319 (a) If C++ support was not built, "make install" and "make test" still
3320 tried to test it.
3321
3322 (b) There were problems when the value of CXX was explicitly set. Some
3323 changes have been made to try to fix these, and ...
3324
3325 (c) --disable-cpp can now be used to explicitly disable C++ support.
3326
3327 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3328 backslash in a target when C++ was disabled. This confuses some
3329 versions of "make", apparently. Using an intermediate variable solves
3330 this. (Same for CPP_LOBJ.)
3331
3332 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3333 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3334 necessary on certain architectures.
3335
3336 4. Added a setting of -export-symbols-regex to the link command to remove
3337 those symbols that are exported in the C sense, but actually are local
3338 within the library, and not documented. Their names all begin with
3339 "_pcre_". This is not a perfect job, because (a) we have to except some
3340 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3341 available (and never for static libraries). I have made a note to try to
3342 find a way round (a) in the future.
3343
3344
3345 Version 6.2 01-Aug-05
3346 ---------------------
3347
3348 1. There was no test for integer overflow of quantifier values. A construction
3349 such as {1111111111111111} would give undefined results. What is worse, if
3350 a minimum quantifier for a parenthesized subpattern overflowed and became
3351 negative, the calculation of the memory size went wrong. This could have
3352 led to memory overwriting.
3353
3354 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3355
3356 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3357 operating environments where this matters.
3358
3359 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3360 PCRE options from within the C++ wrapper.
3361
3362 5. Named capturing subpatterns were not being correctly counted when a pattern
3363 was compiled. This caused two problems: (a) If there were more than 100
3364 such subpatterns, the calculation of the memory needed for the whole
3365 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3366 back references of the form \12, where the number was greater than 9, were
3367 not recognized as back references, even though there were sufficient
3368 previous subpatterns.
3369
3370 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3371 versions of gcc, e.g. 2.95.4.
3372
3373
3374 Version 6.1 21-Jun-05
3375 ---------------------
3376
3377 1. There was one reference to the variable "posix" in pcretest.c that was not
3378 surrounded by "#if !defined NOPOSIX".
3379
3380 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3381 the cross-check on the old pcre_info() function, for the benefit of the
3382 cut-down version of PCRE that is currently imported into Exim.
3383
3384 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3385 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3386 patterns that start (?i) but I don't think that matters. The use of (?i) is
3387 just an example; this all applies to the other options as well.
3388
3389 4. Since libtool seems to echo the compile commands it is issuing, the output
3390 from "make" can be reduced a bit by putting "@" in front of each libtool
3391 compile command.
3392
3393 5. Patch from the folks at Google for configure.in to be a bit more thorough
3394 in checking for a suitable C++ installation before trying to compile the
3395 C++ stuff. This should fix a reported problem when a compiler was present,
3396 but no suitable headers.
3397
3398 6. The man pages all had just "PCRE" as their title. I have changed them to
3399 be the relevant file name. I have also arranged that these names are
3400 retained in the file doc/pcre.txt, which is a concatenation in text format
3401 of all the man pages except the little individual ones for each function.
3402
3403 7. The NON-UNIX-USE file had not been updated for the different set of source
3404 files that come with release 6. I also added a few comments about the C++
3405 wrapper.
3406
3407
3408 Version 6.0 07-Jun-05
3409 ---------------------
3410
3411 1. Some minor internal re-organization to help with my DFA experiments.
3412
3413 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3414 didn't matter for the library itself when fully configured, but did matter
3415 when compiling without UCP support, or within Exim, where the ucp files are
3416 not imported.
3417
3418 3. Refactoring of the library code to split up the various functions into
3419 different source modules. The addition of the new DFA matching code (see
3420 below) to a single monolithic source would have made it really too
3421 unwieldy, quite apart from causing all the code to be include in a
3422 statically linked application, when only some functions are used. This is
3423 relevant even without the DFA addition now that patterns can be compiled in
3424 one application and matched in another.
3425
3426 The downside of splitting up is that there have to be some external
3427 functions and data tables that are used internally in different modules of
3428 the library but which are not part of the API. These have all had their
3429 names changed to start with "_pcre_" so that they are unlikely to clash
3430 with other external names.
3431
3432 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3433 a different (DFA) algorithm. Although it is slower than the original
3434 function, it does have some advantages for certain types of matching
3435 problem.
3436
3437 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3438 including restarting after a partial match.
3439
3440 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3441 defined when compiling for Windows was sent to me. I have put it into the
3442 code, though I have no means of testing or verifying it.
3443
3444 7. Added the pcre_refcount() auxiliary function.
3445
3446 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3447 match before or at the first newline in the subject string. In pcretest,
3448 the /f option on a pattern can be used to set this.
3449
3450 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3451 would behave wrongly. This has been present in PCRE since release 4.0.
3452
3453 10. A number of changes to the pcregrep command:
3454
3455 (a) Refactored how -x works; insert ^(...)$ instead of setting
3456 PCRE_ANCHORED and checking the length, in preparation for adding
3457 something similar for -w.
3458
3459 (b) Added the -w (match as a word) option.
3460
3461 (c) Refactored the way lines are read and buffered so as to have more
3462 than one at a time available.
3463
3464 (d) Implemented a pcregrep test script.
3465
3466 (e) Added the -M (multiline match) option. This allows patterns to match
3467 over several lines of the subject. The buffering ensures that at least
3468 8K, or the rest of the document (whichever is the shorter) is available
3469 for matching (and similarly the previous 8K for lookbehind assertions).
3470
3471 (f) Changed the --help output so that it now says
3472
3473 -w, --word-regex(p)
3474
3475 instead of two lines, one with "regex" and the other with "regexp"
3476 because that confused at least one person since the short forms are the
3477 same. (This required a bit of code, as the output is generated
3478 automatically from a table. It wasn't just a text change.)
3479
3480 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3481 option but starts with a hyphen. Could be a pattern or a path name
3482 starting with a hyphen, for instance.
3483
3484 (h) "-" can be given as a file name to represent stdin.
3485
3486 (i) When file names are being printed, "(standard input)" is used for
3487 the standard input, for compatibility with GNU grep. Previously
3488 "<stdin>" was used.
3489
3490 (j) The option --label=xxx can be used to supply a name to be used for
3491 stdin when file names are being printed. There is no short form.
3492
3493 (k) Re-factored the options decoding logic because we are going to add
3494 two more options that take data. Such options can now be given in four
3495 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3496
3497 (l) Added the -A, -B, and -C options for requesting that lines of context
3498 around matches be printed.
3499
3500 (m) Added the -L option to print the names of files that do not contain
3501 any matching lines, that is, the complement of -l.
3502
3503 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3504 continue to scan other files.
3505
3506 (o) The -s option was incorrectly implemented. For compatibility with other
3507 greps, it now suppresses the error message for a non-existent or non-
3508 accessible file (but not the return code). There is a new option called
3509 -q that suppresses the output of matching lines, which was what -s was
3510 previously doing.
3511
3512 (p) Added --include and --exclude options to specify files for inclusion
3513 and exclusion when recursing.
3514
3515 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3516 Hopefully, it now does.
3517
3518 12. Missing cast in pcre_study().
3519
3520 13. Added an "uninstall" target to the makefile.
3521
3522 14. Replaced "extern" in the function prototypes in Makefile.in with
3523 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3524 world, but is set differently for Windows.
3525
3526 15. Added a second compiling function called pcre_compile2(). The only
3527 difference is that it has an extra argument, which is a pointer to an
3528 integer error code. When there is a compile-time failure, this is set
3529 non-zero, in addition to the error test pointer being set to point to an
3530 error message. The new argument may be NULL if no error number is required
3531 (but then you may as well call pcre_compile(), which is now just a
3532 wrapper). This facility is provided because some applications need a
3533 numeric error indication, but it has also enabled me to tidy up the way
3534 compile-time errors are handled in the POSIX wrapper.
3535
3536 16. Added VPATH=.libs to the makefile; this should help when building with one
3537 prefix path and installing with another. (Or so I'm told by someone who
3538 knows more about this stuff than I do.)
3539
3540 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3541 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3542 match everything, including newlines. This is not POSIX-compatible, but
3543 somebody wanted the feature. From pcretest it can be activated by using
3544 both the P and the s flags.
3545
3546 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3547
3548 19. libpcre.pc was being incorrectly installed as executable.
3549
3550 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3551 it now also looks for '\r' so that it will work unmodified on Windows.
3552
3553 21. Added Google's contributed C++ wrapper to the distribution.
3554
3555 22. Added some untidy missing memory free() calls in pcretest, to keep
3556 Electric Fence happy when testing.
3557
3558
3559
3560 Version 5.0 13-Sep-04
3561 ---------------------
3562
3563 1. Internal change: literal characters are no longer packed up into items
3564 containing multiple characters in a single byte-string. Each character
3565 is now matched using a separate opcode. However, there may be more than one
3566 byte in the character in UTF-8 mode.
3567
3568 2. The pcre_callout_block structure has two new fields: pattern_position and
3569 next_item_length. These contain the offset in the pattern to the next match
3570 item, and its length, respectively.
3571
3572 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3573 insertion of callouts before each pattern item. Added the /C option to
3574 pcretest to make use of this.
3575
3576 4. On the advice of a Windows user, the lines
3577
3578 #if defined(_WIN32) || defined(WIN32)
3579 _setmode( _fileno( stdout ), 0x8000 );
3580 #endif /* defined(_WIN32) || defined(WIN32) */
3581
3582 have been added to the source of pcretest. This apparently does useful
3583 magic in relation to line terminators.
3584
3585 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3586 for the benefit of those environments where the "b" makes a difference.
3587
3588 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3589 to know about it. I have put a hack into configure.in that adds in code
3590 to set GCC=yes if CC=icc. This seems to end up at a point in the
3591 generated configure script that is early enough to affect the setting of
3592 compiler options, which is what is needed, but I have no means of testing
3593 whether it really works. (The user who reported this had patched the
3594 generated configure script, which of course I cannot do.)
3595
3596 LATER: After change 22 below (new libtool files), the configure script
3597 seems to know about icc (and also ecc). Therefore, I have commented out
3598 this hack in configure.in.
3599
3600 7. Added support for pkg-config (2 patches were sent in).
3601
3602 8. Negated POSIX character classes that used a combination of internal tables
3603 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3604 [[:^ascii]]. Typically, they would match almost any characters. The other
3605 POSIX classes were not broken in this way.
3606
3607 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3608 to find the match, as PCRE was deluded into thinking that the match had to
3609 start at the start point or following a newline. The same bug applied to
3610 patterns with negative forward assertions or any backward assertions
3611 preceding ".*" at the start, unless the pattern required a fixed first
3612 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3613
3614 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3615 starting at the last subject character, bytes beyond the end of the subject
3616 string were read.
3617
3618 11. Renamed the variable "class" as "classbits" to make life easier for C++
3619 users. (Previously there was a macro definition, but it apparently wasn't
3620 enough.)
3621
3622 12. Added the new field "tables" to the extra data so that tables can be passed
3623 in at exec time, or the internal tables can be re-selected. This allows
3624 a compiled regex to be saved and re-used at a later time by a different
3625 program that might have everything at different addresses.
3626
3627 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3628 -R library as well as a -L library.
3629
3630 14. The debugging options of pcretest (-d on the command line or D on a
3631 pattern) showed incorrect output for anything following an extended class
3632 that contained multibyte characters and which was followed by a quantifier.
3633
3634 15. Added optional support for general category Unicode character properties
3635 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3636 support. It adds about 90K to the size of the library. The meanings of the
3637 inbuilt class escapes such as \d and \s have NOT been changed.
3638
3639 16. Updated pcredemo.c to include calls to free() to release the memory for the
3640 compiled pattern.
3641
3642 17. The generated file chartables.c was being created in the source directory
3643 instead of in the building directory. This caused the build to fail if the
3644 source directory was different from the building directory, and was
3645 read-only.
3646
3647 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3648 file. No doubt somebody will tell me if they don't make sense... Also added
3649 Dan Mooney's comments about building on OpenVMS.
3650
3651 19. Added support for partial matching via the PCRE_PARTIAL option for
3652 pcre_exec() and the \P data escape in pcretest.
3653
3654 20. Extended pcretest with 3 new pattern features:
3655
3656 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3657 write the compiled pattern to the file whose name is "rest-of-line".
3658 This is a straight binary dump of the data, with the saved pointer to
3659 the character tables forced to be NULL. The study data, if any, is
3660 written too. After writing, pcretest reads a new pattern.
3661
3662 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3663 compiled pattern from the given file. There must not be any
3664 occurrences of "<" in the file name (pretty unlikely); if there are,
3665 pcretest will instead treat the initial "<" as a pattern delimiter.
3666 After reading in the pattern, pcretest goes on to read data lines as
3667 usual.
3668
3669 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3670 and 16-bit fields in a compiled pattern, to simulate a pattern that
3671 was compiled on a host of opposite endianness.
3672
3673 21. The pcre-exec() function can now cope with patterns that were compiled on
3674 hosts of opposite endianness, with this restriction:
3675
3676 As for any compiled expression that is saved and used later, the tables
3677 pointer field cannot be preserved; the extra_data field in the arguments
3678 to pcre_exec() should be used to pass in a tables address if a value
3679 other than the default internal tables were used at compile time.
3680
3681 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3682 now diagnosed as an error. Previously, most of the time, a negative number
3683 would have been treated as zero, but if in addition "ovector" was passed as
3684 NULL, a crash could occur.
3685
3686 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3687 new versions from the libtool 1.5 distribution (the last one is a copy of
3688 a file called libtool.m4). This seems to have fixed the need to patch
3689 "configure" to support Darwin 1.3 (which I used to do). However, I still
3690 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3691 workstation).
3692
3693 24. Changed the PCRE licence to be the more standard "BSD" licence.
3694
3695
3696 Version 4.5 01-Dec-03
3697 ---------------------
3698
3699 1. There has been some re-arrangement of the code for the match() function so
3700 that it can be compiled in a version that does not call itself recursively.
3701 Instead, it keeps those local variables that need separate instances for
3702 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3703 needs to "recurse". Keeping track of where control must go is done by means
3704 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3705 hide most of the details from the main code, and operates only if
3706 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3707 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3708 operating.
3709
3710 To make it easier for callers to provide specially tailored get/free
3711 functions for this usage, two new functions, pcre_stack_malloc, and
3712 pcre_stack_free, are used. They are always called in strict stacking order,
3713 and the size of block requested is always the same.
3714
3715 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3716 PCRE has been compiled to use the stack or the heap for recursion. The
3717 -C option of pcretest uses this to show which version is compiled.
3718
3719 A new data escape \S, is added to pcretest; it causes the amounts of store
3720 obtained and freed by both kinds of malloc/free at match time to be added
3721 to the output.
3722
3723 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3724 what's available on my current Linux desktop machine.
3725
3726 3. When matching a UTF-8 string, the test for a valid string at the start has
3727 been extended. If start_offset is not zero, PCRE now checks that it points
3728 to a byte that is the start of a UTF-8 character. If not, it returns
3729 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3730 this is necessary because there may be backward assertions in the pattern.
3731 When matching the same subject several times, it may save resources to use
3732 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3733
3734 4. The code for checking the validity of UTF-8 strings has been tightened so
3735 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3736 containing "overlong sequences".
3737
3738 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3739 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3740 should have been "&", but it just so happened that all the cases this let
3741 through by mistake were picked up later in the function.
3742
3743 6. I had used a variable called "isblank" - this is a C99 function, causing
3744 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3745
3746 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3747 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3748 so the version is shown; (c) stop "make test" echoing "./RunTest".
3749
3750 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3751
3752 9. The prototype for memmove() for systems that don't have it was using
3753 size_t, but the inclusion of the header that defines size_t was later. I've
3754 moved the #includes for the C headers earlier to avoid this.
3755
3756 10. Added some adjustments to the code to make it easier to compiler on certain
3757 special systems:
3758
3759 (a) Some "const" qualifiers were missing.
3760 (b) Added the macro EXPORT before all exported functions; by default this
3761 is defined to be empty.
3762 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3763 that it reads its output file name as an argument instead of writing
3764 to the standard output and assuming this can be redirected.
3765
3766 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3767 class containing characters with values greater than 255, PCRE compilation
3768 went into a loop.
3769
3770 12. A recursive reference to a subpattern that was within another subpattern
3771 that had a minimum quantifier of zero caused PCRE to crash. For example,
3772 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3773 recursion. If the recursively-called subpattern itself had a zero repeat,
3774 that was OK.
3775
3776 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3777 buffer into which it was copied (for escape processing) was still set at
3778 1024, so long lines caused crashes.
3779
3780 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3781 "internal error: code overflow...". This applied to any character class
3782 that was followed by a possessive quantifier.
3783
3784 15. Modified the Makefile to add libpcre.la as a prerequisite for
3785 libpcreposix.la because I was told this is needed for a parallel build to
3786 work.
3787
3788 16. If a pattern that contained .* following optional items at the start was
3789 studied, the wrong optimizing data was generated, leading to matching
3790 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3791 matching string must start with a or b or c. The correct conclusion for
3792 this pattern is that a match can start with any character.
3793
3794
3795 Version 4.4 13-Aug-03
3796 ---------------------
3797
3798 1. In UTF-8 mode, a character class containing characters with values between
3799 127 and 255 was not handled correctly if the compiled pattern was studied.
3800 In fixing this, I have also improved the studying algorithm for such
3801 classes (slightly).
3802
3803 2. Three internal functions had redundant arguments passed to them. Removal
3804 might give a very teeny performance improvement.
3805
3806 3. Documentation bug: the value of the capture_top field in a callout is *one
3807 more than* the number of the hightest numbered captured substring.
3808
3809 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3810 in incorrectly linking with a previously installed version. They now link
3811 explicitly with libpcre.la.
3812
3813 5. configure.in no longer needs to recognize Cygwin specially.
3814
3815 6. A problem in pcre.in for Windows platforms is fixed.
3816
3817 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3818 pcretest, it used to include the size of the study block as part of its
3819 output. Unfortunately, the structure contains a field that has a different
3820 size on different hardware architectures. This meant that the tests that
3821 showed this size failed. As the block is currently always of a fixed size,
3822 this information isn't actually particularly useful in pcretest output, so
3823 I have just removed it.
3824
3825 8. Three pre-processor statements accidentally did not start in column 1.
3826 Sadly, there are *still* compilers around that complain, even though
3827 standard C has not required this for well over a decade. Sigh.
3828
3829 9. In pcretest, the code for checking callouts passed small integers in the
3830 callout_data field, which is a void * field. However, some picky compilers
3831 complained about the casts involved for this on 64-bit systems. Now
3832 pcretest passes the address of the small integer instead, which should get
3833 rid of the warnings.
3834
3835 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3836 both compile and run time, and gives an error if an invalid UTF-8 sequence
3837 is found. There is a option for disabling this check in cases where the
3838 string is known to be correct and/or the maximum performance is wanted.
3839
3840 11. In response to a bug report, I changed one line in Makefile.in from
3841
3842 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3843 to
3844 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3845
3846 to look similar to other lines, but I have no way of telling whether this
3847 is the right thing to do, as I do not use Windows. No doubt I'll get told
3848 if it's wrong...
3849
3850
3851 Version 4.3 21-May-03
3852 ---------------------
3853
3854 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3855 Makefile.
3856
3857 2. Some refactoring to improve the quality of the code:
3858
3859 (i) The utf8_table... variables are now declared "const".
3860
3861 (ii) The code for \cx, which used the "case flipping" table to upper case
3862 lower case letters, now just substracts 32. This is ASCII-specific,
3863 but the whole concept of \cx is ASCII-specific, so it seems
3864 reasonable.
3865
3866 (iii) PCRE was using its character types table to recognize decimal and
3867 hexadecimal digits in the pattern. This is silly, because it handles
3868 only 0-9, a-f, and A-F, but the character types table is locale-
3869 specific, which means strange things might happen. A private
3870 table is now used for this - though it costs 256 bytes, a table is
3871 much faster than multiple explicit tests. Of course, the standard
3872 character types table is still used for matching digits in subject
3873 strings against \d.
3874
3875 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3876 ending in _t are). So I've renamed it as ESC_tee.
3877
3878 3. The first argument for regexec() in the POSIX wrapper should have been
3879 defined as "const".
3880
3881 4. Changed pcretest to use malloc() for its buffers so that they can be
3882 Electric Fenced for debugging.
3883
3884 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3885 to read one or more bytes before the start of the subject string. Often this
3886 had no effect on PCRE's behaviour, but in some circumstances it could
3887 provoke a segmentation fault.
3888
3889 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3890 to try to read one or more bytes before the start of the subject string.
3891
3892 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3893 UTF-8 support could misbehave in various ways if the subject string
3894 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3895 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3896 back over UTF-8 characters.)
3897
3898
3899 Version 4.2 14-Apr-03
3900 ---------------------
3901
3902 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3903
3904 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3905 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3906 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3907 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3908 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3909 and BUILD_EXEEXT
3910 Note: automatic setting of the BUILD variables is not yet working
3911 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3912 compile-time but not at link-time
3913 [LINK]: use for linking executables only
3914 make different versions for Windows and non-Windows
3915 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3916 libraries
3917 [LINK_FOR_BUILD]: new variable
3918 [OBJEXT]: use throughout
3919 [EXEEXT]: use throughout
3920 <winshared>: new target
3921 <wininstall>: new target
3922 <dftables.o>: use native compiler
3923 <dftables>: use native linker
3924 <install>: handle Windows platform correctly
3925 <clean>: ditto
3926 <check>: ditto
3927 copy DLL to top builddir before testing
3928
3929 As part of these changes, -no-undefined was removed again. This was reported
3930 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3931 in any case.
3932
3933 3. Some tidies to get rid of compiler warnings:
3934
3935 . In the match_data structure, match_limit was an unsigned long int, whereas
3936 match_call_count was an int. I've made them both unsigned long ints.
3937
3938 . In pcretest the fact that a const uschar * doesn't automatically cast to
3939 a void * provoked a warning.
3940
3941 . Turning on some more compiler warnings threw up some "shadow" variables
3942 and a few more missing casts.
3943
3944 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3945 option, a class that contained a single character with a value between 128
3946 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3947
3948 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3949 option, a class that contained several characters, but with at least one
3950 whose value was between 128 and 255 caused PCRE to crash.
3951
3952
3953 Version 4.1 12-Mar-03
3954 ---------------------
3955
3956 1. Compiling with gcc -pedantic found a couple of places where casts were
3957 needed, and a string in dftables.c that was longer than standard compilers are
3958 required to support.
3959
3960 2. Compiling with Sun's compiler found a few more places where the code could
3961 be tidied up in order to avoid warnings.
3962
3963 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3964 first of these names is deprecated in the latest Autoconf in favour of the name
3965 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3966 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3967 analogy I have changed it to CFLAGS_FOR_BUILD.
3968
3969 4. Added -no-undefined to the linking command in the Makefile, because this is
3970 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3971 linking step for the pcreposix library.
3972
3973 5. PCRE was failing to diagnose the case of two named groups with the same
3974 name.
3975
3976 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3977 literal character that is needed in the subject for a match, and scans along to
3978 ensure that it is present before embarking on the full matching process. This
3979 saves time in cases of nested unlimited repeats that are never going to match.
3980 Problem: the scan can take a lot of time if the subject is very long (e.g.
3981 megabytes), thus penalizing straightforward matches. It is now done only if the
3982 amount of subject to be scanned is less than 1000 bytes.
3983
3984 7. A lesser problem with the same optimization is that it was recording the
3985 first character of an anchored pattern as "needed", thus provoking a search
3986 right along the subject, even when the first match of the pattern was going to
3987 fail. The "needed" character is now not set for anchored patterns, unless it
3988 follows something in the pattern that is of non-fixed length. Thus, it still
3989 fulfils its original purpose of finding quick non-matches in cases of nested
3990 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3991
3992
3993 Version 4.0 17-Feb-03
3994 ---------------------
3995
3996 1. If a comment in an extended regex that started immediately after a meta-item
3997 extended to the end of string, PCRE compiled incorrect data. This could lead to
3998 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3999
4000 2. Moved to autoconf 2.53 and libtool 1.4.2.
4001
4002 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4003 the special perltest8 script is no longer needed - all the tests can be run
4004 from a single perltest script.
4005
4006 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4007 by \s. It has now been removed in PCRE. This means it isn't recognized as
4008 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4009 class [:space:] *does* include VT, thereby creating a mess.
4010
4011 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4012 space and tab.
4013
4014 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4015 its new features into the main test script, reducing the number of scripts.
4016
4017 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4018 were backward compatible, and made the (?i) apply to the whole pattern, as if
4019 /i were given. Now it behaves more logically, and applies the option setting
4020 only to what follows. PCRE has been changed to follow suit. However, if it
4021 finds options settings right at the start of the pattern, it extracts them into
4022 the global options, as before. Thus, they show up in the info data.
4023
4024 8. Added support for the \Q...\E escape sequence. Characters in between are
4025 treated as literals. This is slightly different from Perl in that $ and @ are
4026 also handled as literals inside the quotes. In Perl, they will cause variable
4027 interpolation. Note the following examples:
4028
4029 Pattern PCRE matches Perl matches
4030
4031 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4032 \Qabc\$xyz\E abc\$xyz abc\$xyz
4033 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4034
4035 For compatibility with Perl, \Q...\E sequences are recognized inside character
4036 classes as well as outside them.
4037
4038 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4039 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4040 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4041 signed/unsigned warnings.
4042
4043 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4044 option for pcretest, so I've replaced it by a simple function that does just
4045 that job.
4046
4047 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4048 "pcregrep -".
4049
4050 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4051 Java package. This provides some syntactic sugar for simple cases of what my
4052 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4053 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4054 item, you can use this simplified notation. Note that only makes sense with
4055 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4056 greediness, whatever the setting of the PCRE_UNGREEDY option.
4057
4058 13. A change of greediness default within a pattern was not taking effect at
4059 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4060 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4061 was abstracted outside.
4062
4063 14. PCRE now supports the \G assertion. It is true when the current matching
4064 position is at the start point of the match. This differs from \A when the
4065 starting offset is non-zero. Used with the /g option of pcretest (or similar
4066 code), it works in the same way as it does for Perl's /g option. If all
4067 alternatives of a regex begin with \G, the expression is anchored to the start
4068 match position, and the "anchored" flag is set in the compiled expression.
4069
4070 15. Some bugs concerning the handling of certain option changes within patterns
4071 have been fixed. These applied to options other than (?ims). For example,
4072 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4073 been the other way round. Some of this was related to change 7 above.
4074
4075 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4076 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4077 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4078 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4079
4080 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4081 mode. Unlike ".", it always matches newline, whatever the setting of
4082 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4083 assertions. Perl allows it, but it doesn't (in general) work because it can't
4084 calculate the length of the lookbehind. At least, that's the case for Perl
4085 5.8.0 - I've been told they are going to document that it doesn't work in
4086 future.
4087
4088 18. Added an error diagnosis for escapes that PCRE does not support: these are
4089 \L, \l, \N, \P, \p, \U, \u, and \X.
4090
4091 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4092 reading past the end of the pattern in cases such as /[abcd/.
4093
4094 20. PCRE was getting more memory than necessary for patterns with classes that
4095 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4096
4097 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4098 compiling PCRE for use with Virtual Pascal.
4099
4100 22. Small fix to the Makefile to make it work properly if the build is done
4101 outside the source tree.
4102
4103 23. Added a new extension: a condition to go with recursion. If a conditional
4104 subpattern starts with (?(R) the "true" branch is used if recursion has
4105 happened, whereas the "false" branch is used only at the top level.
4106
4107 24. When there was a very long string of literal characters (over 255 bytes
4108 without UTF support, over 250 bytes with UTF support), the computation of how
4109 much memory was required could be incorrect, leading to segfaults or other
4110 strange effects.
4111
4112 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4113 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4114 there was a subsequent back reference to those brackets. This meant that, for
4115 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4116 possible to check for precisely this case. All we can do is abandon the
4117 optimization if .* occurs inside capturing brackets when there are any back
4118 references whatsoever. (See below for a better fix that came later.)
4119
4120 26. The handling of the optimization for finding the first character of a
4121 non-anchored pattern, and for finding a character that is required later in the
4122 match were failing in some cases. This didn't break the matching; it just
4123 failed to optimize when it could. The way this is done has been re-implemented.
4124
4125 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4126
4127 28. Added a new feature that provides some of the functionality that Perl
4128 provides with (?{...}). The facility is termed a "callout". The way it is done
4129 in PCRE is for the caller to provide an optional function, by setting
4130 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4131 global variable. By default it is unset, which disables all calling out. To get
4132 the function called, the regex must include (?C) at appropriate points. This
4133 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4134 This provides a means of identifying different callout points. When PCRE
4135 reaches such a point in the regex, if pcre_callout has been set, the external
4136 function is called. It is provided with data in a structure called
4137 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4138 matching continues; if it returns a non-zero value, the match at the current
4139 point fails. However, backtracking will occur if possible. [This was changed
4140 later and other features added - see item 49 below.]
4141
4142 29. pcretest is upgraded to test the callout functionality. It provides a
4143 callout function that displays information. By default, it shows the start of
4144 the match and the current position in the text. There are some new data escapes
4145 to vary what happens:
4146
4147 \C+ in addition, show current contents of captured substrings
4148 \C- do not supply a callout function
4149 \C!n return 1 when callout number n is reached
4150 \C!n!m return 1 when callout number n is reached for the mth time
4151
4152 30. If pcregrep was called with the -l option and just a single file name, it
4153 output "<stdin>" if a match was found, instead of the file name.
4154
4155 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4156 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4157 pcre_exec(). This saves a malloc/free per call. The default value of
4158 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4159 when configuring.
4160
4161 32. The default maximum size of a compiled pattern is 64K. There have been a
4162 few cases of people hitting this limit. The code now uses macros to handle the
4163 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4164 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4165 configuring. Tests 2 and 5 work only with 2-byte links because they output
4166 debugging information about compiled patterns.
4167
4168 33. Internal code re-arrangements:
4169
4170 (a) Moved the debugging function for printing out a compiled regex into
4171 its own source file (printint.c) and used #include to pull it into
4172 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4173 separate copies.
4174
4175 (b) Defined the list of op-code names for debugging as a macro in
4176 internal.h so that it is next to the definition of the opcodes.
4177
4178 (c) Defined a table of op-code lengths for simpler skipping along compiled
4179 code. This is again a macro in internal.h so that it is next to the
4180 definition of the opcodes.
4181
4182 34. Added support for recursive calls to individual subpatterns, along the
4183 lines of Robin Houston's patch (but implemented somewhat differently).
4184
4185 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4186 allow it to read and process whole directories in Win32. This code was
4187 contributed by Lionel Fourquaux; it has not been tested by me.
4188
4189 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4190 used to name a group. Names consist of alphanumerics and underscores, and must
4191 be unique. Back references use the syntax (?P=name) and recursive calls use
4192 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4193 numbers. The function pcre_fullinfo() can be used after compilation to extract
4194 a name/number map. There are three relevant calls:
4195
4196 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4197 PCRE_INFO_NAMECOUNT yields the number of entries
4198 PCRE_INFO_NAMETABLE yields a pointer to the map.
4199
4200 The map is a vector of fixed-size entries. The size of each entry depends on
4201 the length of the longest name used. The first two bytes of each entry are the
4202 group number, most significant byte first. There follows the corresponding
4203 name, zero terminated. The names are in alphabetical order.
4204
4205 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4206 case instead of 255. Making it the same both with and without UTF-8 support
4207 means that the same test output works with both.
4208
4209 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4210 calling malloc() with a zero argument.
4211
4212 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4213 optimization. I've improved things by keeping a bitmap of backreferences with
4214 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4215 fact referenced, the optimization can be applied. It is unlikely that a
4216 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4217 the match to follow \n) will appear inside brackets with a number greater than
4218 31, but if it does, any back reference > 31 suppresses the optimization.
4219
4220 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4221 of disabling numbered capturing parentheses. Any opening parenthesis that is
4222 not followed by ? behaves as if it were followed by ?: but named parentheses
4223 can still be used for capturing (and they will acquire numbers in the usual
4224 way).
4225
4226 41. Redesigned the return codes from the match() function into yes/no/error so
4227 that errors can be passed back from deep inside the nested calls. A malloc
4228 failure while inside a recursive subpattern call now causes the
4229 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4230
4231 42. It is now possible to set a limit on the number of times the match()
4232 function is called in a call to pcre_exec(). This facility makes it possible to
4233 limit the amount of recursion and backtracking, though not in a directly
4234 obvious way, because the match() function is used in a number of different
4235 circumstances. The count starts from zero for each position in the subject
4236 string (for non-anchored patterns). The default limit is, for compatibility, a
4237 large number, namely 10 000 000. You can change this in two ways:
4238
4239 (a) When configuring PCRE before making, you can use --with-match-limit=n
4240 to set a default value for the compiled library.
4241
4242 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4243 a different value is set. See 45 below.
4244
4245 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4246
4247 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4248 of things that can be changed at compile time. The first argument specifies
4249 what is wanted and the second points to where the information is to be placed.
4250 The current list of available information is:
4251
4252 PCRE_CONFIG_UTF8
4253
4254 The output is an integer that is set to one if UTF-8 support is available;
4255 otherwise it is set to zero.
4256
4257 PCRE_CONFIG_NEWLINE
4258
4259 The output is an integer that it set to the value of the code that is used for
4260 newline. It is either LF (10) or CR (13).
4261
4262 PCRE_CONFIG_LINK_SIZE
4263
4264 The output is an integer that contains the number of bytes used for internal
4265 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4266
4267 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4268
4269 The output is an integer that contains the threshold above which the POSIX
4270 interface uses malloc() for output vectors. See item 31 above.
4271
4272 PCRE_CONFIG_MATCH_LIMIT
4273
4274 The output is an unsigned integer that contains the default limit of the number
4275 of match() calls in a pcre_exec() execution. See 42 above.
4276
4277 44. pcretest has been upgraded by the addition of the -C option. This causes it
4278 to extract all the available output from the new pcre_config() function, and to
4279 output it. The program then exits immediately.
4280
4281 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4282 order to support additional features. One way would have been to define
4283 pcre_exec2() (for example) with extra arguments, but this would not have been
4284 extensible, and would also have required all calls to the original function to
4285 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4286 is used for passing in "extra" data from pcre_study().
4287
4288 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4289 contains the following fields:
4290
4291 flags a bitmap indicating which of the following fields are set
4292 study_data opaque data from pcre_study()
4293 match_limit a way of specifying a limit on match() calls for a specific
4294 call to pcre_exec()
4295 callout_data data for callouts (see 49 below)
4296
4297 The flag bits are also defined in pcre.h, and are
4298
4299 PCRE_EXTRA_STUDY_DATA
4300 PCRE_EXTRA_MATCH_LIMIT
4301 PCRE_EXTRA_CALLOUT_DATA
4302
4303 The pcre_study() function now returns one of these new pcre_extra blocks, with
4304 the actual study data pointed to by the study_data field, and the
4305 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4306 before. That is, this change is entirely upwards-compatible and requires no
4307 change to existing code.
4308
4309 If you want to pass in additional data to pcre_exec(), you can either place it
4310 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4311 block.
4312
4313 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4314 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4315 times with different match limits, until it finds the minimum value needed for
4316 pcre_exec() to complete. The value is then output. This can be instructive; for
4317 most simple matches the number is quite small, but for pathological cases it
4318 gets very large very quickly.
4319
4320 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4321 returns the size of the data block pointed to by the study_data field in a
4322 pcre_extra block, that is, the value that was passed as the argument to
4323 pcre_malloc() when PCRE was getting memory in which to place the information
4324 created by pcre_study(). The fourth argument should point to a size_t variable.
4325 pcretest has been extended so that this information is shown after a successful
4326 pcre_study() call when information about the compiled regex is being displayed.
4327
4328 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4329 because what follows is always an absolute path. (Later: it turns out that this
4330 is more than cosmetic for MinGW, because it doesn't like empty path
4331 components.)
4332
4333 49. Some changes have been made to the callout feature (see 28 above):
4334
4335 (i) A callout function now has three choices for what it returns:
4336
4337 0 => success, carry on matching
4338 > 0 => failure at this point, but backtrack if possible
4339 < 0 => serious error, return this value from pcre_exec()
4340
4341 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4342 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4343 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4344 use by callout functions. It will never be used by PCRE itself.
4345
4346 (ii) The pcre_extra structure (see 45 above) has a void * field called
4347 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4348 pcre_callout_block structure has a field of the same name. The contents of
4349 the field passed in the pcre_extra structure are passed to the callout
4350 function in the corresponding field in the callout block. This makes it
4351 easier to use the same callout-containing regex from multiple threads. For
4352 testing, the pcretest program has a new data escape
4353
4354 \C*n pass the number n (may be negative) as callout_data
4355
4356 If the callout function in pcretest receives a non-zero value as
4357 callout_data, it returns that value.
4358
4359 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4360 there were some redundant $(CFLAGS) in commands that are now specified as
4361 $(LINK), which already includes $(CFLAGS).
4362
4363 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4364 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4365 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4366 one-byte characters throughout. Note that case-insensitive matching applies
4367 only to characters whose values are less than 256. PCRE doesn't support the
4368 notion of cases for higher-valued characters.
4369
4370 (i) A character class whose characters are all within 0-255 is handled as
4371 a bit map, and the map is inverted for negative classes. Previously, a
4372 character > 255 always failed to match such a class; however it should
4373 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4374
4375 (ii) A negated character class with a single character < 255 is coded as
4376 "not this character" (OP_NOT). This wasn't working properly when the test
4377 character was multibyte, either singly or repeated.
4378
4379 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4380 mode, for example: \x{100}{2,3}.
4381
4382 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4383 singly or repeated) now correctly test multibyte characters. However,
4384 PCRE doesn't recognize any characters with values greater than 255 as
4385 digits, spaces, or word characters. Such characters always match \D, \S,
4386 and \W, and never match \d, \s, or \w.
4387
4388 (v) Classes may now contain characters and character ranges with values
4389 greater than 255. For example: [ab\x{100}-\x{400}].
4390
4391 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4392 PCRE in UTF-8 mode.
4393
4394 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4395 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4396 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4397 value.)
4398
4399 53. The single man page has become too large. I have therefore split it up into
4400 a number of separate man pages. These also give rise to individual HTML pages;
4401 these are now put in a separate directory, and there is an index.html page that
4402 lists them all. Some hyperlinking between the pages has been installed.
4403
4404 54. Added convenience functions for handling named capturing parentheses.
4405
4406 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4407 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4408 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4409 are faulted.
4410
4411 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4412 calling configure. These values are used when compiling the dftables.c program
4413 which is run to generate the source of the default character tables. They
4414 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4415 you will need to set these values.
4416
4417 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4418
4419
4420 Version 3.9 02-Jan-02
4421 ---------------------
4422
4423 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4424
4425 2. If --disable-static was given, the building process failed when trying to
4426 build pcretest and pcregrep. (For some reason it was using libtool to compile
4427 them, which is not right, as they aren't part of the library.)
4428
4429
4430 Version 3.8 18-Dec-01
4431 ---------------------
4432
4433 1. The experimental UTF-8 code was completely screwed up. It was packing the
4434 bytes in the wrong order. How dumb can you get?
4435
4436
4437 Version 3.7 29-Oct-01
4438 ---------------------
4439
4440 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4441 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4442 this didn't happen under Solaris 8, where I normally test things.
4443
4444 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4445 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4446 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4447 this fix an hour or so after the initial 3.7 release.)
4448
4449
4450 Version 3.6 23-Oct-01
4451 ---------------------
4452
4453 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4454 offsets passed as NULL with zero offset count.
4455
4456 2. The config.guess and config.sub files had not been updated when I moved to
4457 the latest autoconf.
4458
4459
4460 Version 3.5 15-Aug-01
4461 ---------------------
4462
4463 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4464 had been forgotten.
4465
4466 2. By using declared but undefined structures, we can avoid using "void"
4467 definitions in pcre.h while keeping the internal definitions of the structures
4468 private.
4469
4470 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4471 user point of view, this means that both static and shared libraries are built
4472 by default, but this can be individually controlled. More of the work of
4473 handling this static/shared cases is now inside libtool instead of PCRE's make
4474 file.
4475
4476 4. The pcretest utility is now installed along with pcregrep because it is
4477 useful for users (to test regexs) and by doing this, it automatically gets
4478 relinked by libtool. The documentation has been turned into a man page, so
4479 there are now .1, .txt, and .html versions in /doc.
4480
4481 5. Upgrades to pcregrep:
4482 (i) Added long-form option names like gnu grep.
4483 (ii) Added --help to list all options with an explanatory phrase.
4484 (iii) Added -r, --recursive to recurse into sub-directories.
4485 (iv) Added -f, --file to read patterns from a file.
4486
4487 6. pcre_exec() was referring to its "code" argument before testing that
4488 argument for NULL (and giving an error if it was NULL).
4489
4490 7. Upgraded Makefile.in to allow for compiling in a different directory from
4491 the source directory.
4492
4493 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4494 options bits, the pointer it was passed was to an int instead of to an unsigned
4495 long int. This mattered only on 64-bit systems.
4496
4497 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4498 generated) instead of pcre.in, which it its source. Also made the same change
4499 in several of the .c files.
4500
4501 10. A new release of gcc defines printf() as a macro, which broke pcretest
4502 because it had an ifdef in the middle of a string argument for printf(). Fixed
4503 by using separate calls to printf().
4504
4505 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4506 script, to force use of CR or LF instead of \n in the source. On non-Unix
4507 systems, the value can be set in config.h.
4508
4509 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4510 absolute limit. Changed the text of the error message to make this clear, and
4511 likewise updated the man page.
4512
4513 13. The limit of 99 on the number of capturing subpatterns has been removed.
4514 The new limit is 65535, which I hope will not be a "real" limit.
4515
4516
4517 Version 3.4 22-Aug-00
4518 ---------------------
4519
4520 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4521
4522 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4523
4524
4525 Version 3.3 01-Aug-00
4526 ---------------------
4527
4528 1. If an octal character was given, but the value was greater than \377, it
4529 was not getting masked to the least significant bits, as documented. This could
4530 lead to crashes in some systems.
4531
4532 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4533 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4534
4535 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4536 These just pass their arguments on to (pcre_free)(), but they are provided
4537 because some uses of PCRE bind it to non-C systems that can call its functions,
4538 but cannot call free() or pcre_free() directly.
4539
4540 4. Add "make test" as a synonym for "make check". Corrected some comments in
4541 the Makefile.
4542
4543 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4544 Makefile.
4545
4546 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4547 command called pgrep for grepping around the active processes.
4548
4549 7. Added the beginnings of support for UTF-8 character strings.
4550
4551 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4552 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4553 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4554 out for the ar command.)
4555
4556
4557 Version 3.2 12-May-00
4558 ---------------------
4559
4560 This is purely a bug fixing release.
4561
4562 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4563 of ZA. This was just one example of several cases that could provoke this bug,
4564 which was introduced by change 9 of version 2.00. The code for breaking
4565 infinite loops after an iteration that matches an empty string was't working
4566 correctly.
4567
4568 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4569 when matched against abbab (for example). After matching an empty string, it
4570 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4571 caused it to match further down the string than it should.
4572
4573 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4574 was there because it doesn't seem to be needed, and it causes trouble on some
4575 systems, as it is not a Standard C header. It has been removed.
4576
4577 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4578 were reported on the Macintosh. The changes were from
4579
4580 while ((c = *(++ptr)) != 0 && c != '\n');
4581 to
4582 while ((c = *(++ptr)) != 0 && c != '\n') ;
4583
4584 Totally extraordinary, but if that's what it takes...
4585
4586 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4587 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4588 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4589 assumes the way PCRE uses memmove() (always moving upwards).
4590
4591 6. PCRE is being used in one environment where strchr() is not available. There
4592 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4593 faster code anyway.
4594
4595
4596 Version 3.1 09-Feb-00
4597 ---------------------
4598
4599 The only change in this release is the fixing of some bugs in Makefile.in for
4600 the "install" target:
4601
4602 (1) It was failing to install pcreposix.h.
4603
4604 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4605
4606
4607 Version 3.0 01-Feb-00
4608 ---------------------
4609
4610 1. Add support for the /+ modifier to perltest (to output $` like it does in
4611 pcretest).
4612
4613 2. Add support for the /g modifier to perltest.
4614
4615 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4616 matches null strings.
4617
4618 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4619 pattern. Perl treats empty patterns specially - it reuses the most recent
4620 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4621 effect.
4622
4623 5. The POSIX interface was broken in that it was just handing over the POSIX
4624 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4625 required a bigger vector, with some working space on the end. This means that
4626 the POSIX wrapper now has to get and free some memory, and copy the results.
4627
4628 6. Added some simple autoconf support, placing the test data and the
4629 documentation in separate directories, re-organizing some of the
4630 information files, and making it build pcre-config (a GNU standard). Also added
4631 libtool support for building PCRE as a shared library, which is now the
4632 default.
4633
4634 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4635 09 are not valid octal constants. Single digits will be used for minor values
4636 less than 10.
4637
4638 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4639 existing programs that set these in the POSIX interface can use PCRE without
4640 modification.
4641
4642 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4643 return all that pcre_info() returns, plus additional data. The pcre_info()
4644 function is retained for compatibility, but is considered to be obsolete.
4645
4646 10. Added experimental recursion feature (?R) to handle one common case that
4647 Perl 5.6 will be able to do with (?p{...}).
4648
4649 11. Added support for POSIX character classes like [:alpha:], which Perl is
4650 adopting.
4651
4652
4653 Version 2.08 31-Aug-99
4654 ----------------------
4655
4656 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4657 trying to match at the startoffset position, but instead was moving forward to
4658 the next newline as if a previous match had failed.
4659
4660 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4661 and could get into a loop if a null string was matched other than at the start
4662 of the subject.
4663
4664 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4665 be distinguished at compile time, and for completeness also added PCRE_DATE.
4666
4667 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4668 in GnuWin32 environments.
4669
4670
4671 Version 2.07 29-Jul-99
4672 ----------------------
4673
4674 1. The documentation is now supplied in plain text form and HTML as well as in
4675 the form of man page sources.
4676
4677 2. C++ compilers don't like assigning (void *) values to other pointer types.
4678 In particular this affects malloc(). Although there is no problem in Standard
4679 C, I've put in casts to keep C++ compilers happy.
4680
4681 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4682 should be (const char *).
4683
4684 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4685 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4686 However, I haven't made this a standard facility. The documentation doesn't
4687 mention it, and the Makefile doesn't support it.
4688
4689 5. The Makefile now contains an "install" target, with editable destinations at
4690 the top of the file. The pcretest program is not installed.
4691
4692 6. pgrep -V now gives the PCRE version number and date.
4693
4694 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4695 causing the entire string to be ignored, instead of just the last character.
4696
4697 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4698 non-matching string, it can take a very, very long time, even for strings of
4699 quite modest length, because of the nested recursion. PCRE now does better in
4700 some of these cases. It does this by remembering the last required literal
4701 character in the pattern, and pre-searching the subject to ensure it is present
4702 before running the real match. In other words, it applies a heuristic to detect
4703 some types of certain failure quickly, and in the above example, if presented
4704 with a string that has no trailing " it gives "no match" very quickly.
4705
4706 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4707 other alternatives are tried instead.
4708
4709
4710 Version 2.06 09-Jun-99
4711 ----------------------
4712
4713 1. Change pcretest's output for amount of store used to show just the code
4714 space, because the remainder (the data block) varies in size between 32-bit and
4715 64-bit systems.
4716
4717 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4718 start matching at. This allows lookbehinds to work when searching for multiple
4719 occurrences in a string.
4720
4721 3. Added additional options to pcretest for testing multiple occurrences:
4722
4723 /+ outputs the rest of the string that follows a match
4724 /g loops for multiple occurrences, using the new startoffset argument
4725 /G loops for multiple occurrences by passing an incremented pointer
4726
4727 4. PCRE wasn't doing the "first character" optimization for patterns starting
4728 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4729 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4730 the letter 'x'. On long subject strings, this gives a significant speed-up.
4731
4732
4733 Version 2.05 21-Apr-99
4734 ----------------------
4735
4736 1. Changed the type of magic_number from int to long int so that it works
4737 properly on 16-bit systems.
4738
4739 2. Fixed a bug which caused patterns starting with .* not to work correctly
4740 when the subject string contained newline characters. PCRE was assuming
4741 anchoring for such patterns in all cases, which is not correct because .* will
4742 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4743 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4744 must be retried after every newline in the subject.
4745
4746
4747 Version 2.04 18-Feb-99
4748 ----------------------
4749
4750 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4751 computation of the store needed to hold the pattern was incorrect (too large).
4752 If such patterns were nested a few deep, this could multiply and become a real
4753 problem.
4754
4755 2. Added /M option to pcretest to show the memory requirement of a specific
4756 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4757
4758 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4759 compiled in such a way that the backtracking after subsequent failure was
4760 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4761 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4762
4763
4764 Version 2.03 02-Feb-99
4765 ----------------------
4766
4767 1. Fixed typo and small mistake in man page.
4768
4769 2. Added 4th condition (GPL supersedes if conflict) and created separate
4770 LICENCE file containing the conditions.
4771
4772 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4773 Perl, that is the internal \ allows the delimiter to be included in the
4774 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4775 the final delimiter, add \ to the end of the pattern (to test the error).
4776
4777 4. Added the convenience functions for extracting substrings after a successful
4778 match. Updated pcretest to make it able to test these functions.
4779
4780
4781 Version 2.02 14-Jan-99
4782 ----------------------
4783
4784 1. Initialized the working variables associated with each extraction so that
4785 their saving and restoring doesn't refer to uninitialized store.
4786
4787 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4788 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4789 fix the problem.
4790
4791 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4792 calls, and wasn't printing the correct value for compiling calls. Increased the
4793 default value of LOOPREPEAT, and the number of significant figures in the
4794 times.
4795
4796 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4797
4798 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4799 a building problem on Windows NT with a FAT file system.
4800
4801
4802 Version 2.01 21-Oct-98
4803 ----------------------
4804
4805 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4806 to character tables built by pcre_maketables() in the current locale. If NULL
4807 is passed, the default tables are used.
4808
4809
4810 Version 2.00 24-Sep-98
4811 ----------------------
4812
4813 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4814 it any more.
4815
4816 2. Allow quantification of (?>) groups, and make it work correctly.
4817
4818 3. The first character computation wasn't working for (?>) groups.
4819
4820 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4821 end of the subject) and add 5.005's \z, which really does match only at the
4822 very end of the subject.
4823
4824 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4825
4826 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4827 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4828 localized options. All options to pcre_study() were also removed.
4829
4830 7. Add other new features from 5.005:
4831
4832 $(?<= positive lookbehind
4833 $(?<! negative lookbehind
4834 (?imsx-imsx) added the unsetting capability
4835 such a setting is global if at outer level; local otherwise
4836 (?imsx-imsx:) non-capturing groups with option setting
4837 (?(cond)re|re) conditional pattern matching
4838
4839 A backreference to itself in a repeated group matches the previous
4840 captured string.
4841
4842 8. General tidying up of studying (both automatic and via "study")
4843 consequential on the addition of new assertions.
4844
4845 9. As in 5.005, unlimited repeated groups that could match an empty substring
4846 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4847 runtime if any iteration does actually match an empty substring.
4848
4849 10. Include the RunTest script in the distribution.
4850
4851 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4852 discrepancies, some of which were old and were also with respect to 5.004. They
4853 have now been fixed.
4854
4855
4856 Version 1.09 28-Apr-98
4857 ----------------------
4858
4859 1. A negated single character class followed by a quantifier with a minimum
4860 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4861 program crashes, or just wrong answers. This did not apply to negated classes
4862 containing more than one character, or to minima other than one.
4863
4864
4865 Version 1.08 27-Mar-98
4866 ----------------------
4867
4868 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4869
4870 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4871 latter must appear before anything that relies on it in the pattern.
4872
4873
4874 Version 1.07 16-Feb-98
4875 ----------------------
4876
4877 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4878 repeat of a potentially empty string).
4879
4880
4881 Version 1.06 23-Jan-98
4882 ----------------------
4883
4884 1. Added Markus Oberhumer's little patches for C++.
4885
4886 2. Literal strings longer than 255 characters were broken.
4887
4888
4889 Version 1.05 23-Dec-97
4890 ----------------------
4891
4892 1. Negated character classes containing more than one character were failing if
4893 PCRE_CASELESS was set at run time.
4894
4895
4896 Version 1.04 19-Dec-97
4897 ----------------------
4898
4899 1. Corrected the man page, where some "const" qualifiers had been omitted.
4900
4901 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4902 input syntax.
4903
4904 3. Fixed memory leak which occurred when a regex with back references was
4905 matched with an offsets vector that wasn't big enough. The temporary memory
4906 that is used in this case wasn't being freed if the match failed.
4907
4908 4. Tidied pcretest to ensure it frees memory that it gets.
4909
4910 5. Temporary memory was being obtained in the case where the passed offsets
4911 vector was exactly big enough.
4912
4913 6. Corrected definition of offsetof() from change 5 below.
4914
4915 7. I had screwed up change 6 below and broken the rules for the use of
4916 setjmp(). Now fixed.
4917
4918
4919 Version 1.03 18-Dec-97
4920 ----------------------
4921
4922 1. A erroneous regex with a missing opening parenthesis was correctly
4923 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4924 on some systems.
4925
4926 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4927 it was reported that one broken compiler failed on the former because "code" is
4928 also an independent variable.
4929
4930 3. The erroneous regex a[]b caused an array overrun reference.
4931
4932 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4933 fail on data ending with that character. (It was going on too far, and checking
4934 the next character, typically a binary zero.) This was specific to the
4935 optimized code for single-character negative classes.
4936
4937 5. Added a contributed patch from the TIN world which does the following:
4938
4939 + Add an undef for memmove, in case the the system defines a macro for it.
4940
4941 + Add a definition of offsetof(), in case there isn't one. (I don't know
4942 the reason behind this - offsetof() is part of the ANSI standard - but
4943 it does no harm).
4944
4945 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4946 most of the places where whitespace preceded '#'. I have given up and
4947 allowed the remaining 2 cases to be at the margin.
4948
4949 + Rename some variables in pcre to eliminate shadowing. This seems very
4950 pedantic, but does no harm, of course.
4951
4952 6. Moved the call to setjmp() into its own function, to get rid of warnings
4953 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4954
4955 7. Constructs such as \d{8,} were compiling into the equivalent of
4956 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4957 outcome, but in this particular case used more store than had been allocated,
4958 which caused the bug to be discovered because it threw up an internal error.
4959
4960 8. The debugging code in both pcre and pcretest for outputting the compiled
4961 form of a regex was going wrong in the case of back references followed by
4962 curly-bracketed repeats.
4963
4964
4965 Version 1.02 12-Dec-97
4966 ----------------------
4967
4968 1. Typos in pcre.3 and comments in the source fixed.
4969
4970 2. Applied a contributed patch to get rid of places where it used to remove
4971 'const' from variables, and fixed some signed/unsigned and uninitialized
4972 variable warnings.
4973
4974 3. Added the "runtest" target to Makefile.
4975
4976 4. Set default compiler flag to -O2 rather than just -O.
4977
4978
4979 Version 1.01 19-Nov-97
4980 ----------------------
4981
4982 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4983 like /([ab]*)*/, that is, for classes with more than one character in them.
4984
4985 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4986 as /((?>a*))*/ (a PCRE_EXTRA facility).
4987
4988
4989 Version 1.00 18-Nov-97
4990 ----------------------
4991
4992 1. Added compile-time macros to support systems such as SunOS4 which don't have
4993 memmove() or strerror() but have other things that can be used instead.
4994
4995 2. Arranged that "make clean" removes the executables.
4996
4997
4998 Version 0.99 27-Oct-97
4999 ----------------------
5000
5001 1. Fixed bug in code for optimizing classes with only one character. It was
5002 initializing a 32-byte map regardless, which could cause it to run off the end
5003 of the memory it had got.
5004
5005 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5006
5007
5008 Version 0.98 22-Oct-97
5009 ----------------------
5010
5011 1. Fixed bug in code for handling temporary memory usage when there are more
5012 back references than supplied space in the ovector. This could cause segfaults.
5013
5014
5015 Version 0.97 21-Oct-97
5016 ----------------------
5017
5018 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5019
5020 2. Optimized negated single characters not to use a bit map.
5021
5022 3. Brought error texts together as macro definitions; clarified some of them;
5023 fixed one that was wrong - it said "range out of order" when it meant "invalid
5024 escape sequence".
5025
5026 4. Changed some char * arguments to const char *.
5027
5028 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5029
5030 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5031 pcretest.
5032
5033
5034 Version 0.96 16-Oct-97
5035 ----------------------
5036
5037 1. Added a simple "pgrep" utility to the distribution.
5038
5039 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5040 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5041 where "ddd" means "one or more decimal digits".
5042
5043 3. Fixed serious bug. If a pattern had a back reference, but the call to
5044 pcre_exec() didn't supply a large enough ovector to record the related
5045 identifying subpattern, the match always failed. PCRE now remembers the number
5046 of the largest back reference, and gets some temporary memory in which to save
5047 the offsets during matching if necessary, in order to ensure that
5048 backreferences always work.
5049
5050 4. Increased the compatibility with Perl in a number of ways:
5051
5052 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5053 to request this handling. The option can be set at compile or exec time.
5054
5055 (b) $ matches before a terminating newline by default; an option
5056 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5057 mode). The option can be set at compile or exec time.
5058
5059 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5060 the same as Perl's. If the decimal number it represents is less than 10
5061 or there aren't that many previous left capturing parentheses, an octal
5062 escape is read. Inside a character class, it's always an octal escape,
5063 even if it is a single digit.
5064
5065 (d) An escaped but undefined alphabetic character is taken as a literal,
5066 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5067 escapes.
5068
5069 (e) {0} is now permitted. (The previous item is removed from the compiled
5070 pattern).
5071
5072 5. Changed all the names of code files so that the basic parts are no longer
5073 than 10 characters, and abolished the teeny "globals.c" file.
5074
5075 6. Changed the handling of character classes; they are now done with a 32-byte
5076 bit map always.
5077
5078 7. Added the -d and /D options to pcretest to make it possible to look at the
5079 internals of compilation without having to recompile pcre.
5080
5081
5082 Version 0.95 23-Sep-97
5083 ----------------------
5084
5085 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5086 \x20 at the start of a run of normal characters. These were being treated as
5087 real characters, instead of the source characters being re-checked.
5088
5089
5090 Version 0.94 18-Sep-97
5091 ----------------------
5092
5093 1. The functions are now thread-safe, with the caveat that the global variables
5094 containing pointers to malloc() and free() or alternative functions are the
5095 same for all threads.
5096
5097 2. Get pcre_study() to generate a bitmap of initial characters for non-
5098 anchored patterns when this is possible, and use it if passed to pcre_exec().
5099
5100
5101 Version 0.93 15-Sep-97
5102 ----------------------
5103
5104 1. /(b)|(:+)/ was computing an incorrect first character.
5105
5106 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5107 but not actually doing anything yet.
5108
5109 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5110 as Perl does (e.g. [-az] or [az-]).
5111
5112 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5113 all possible positions.
5114
5115 5. Split up into different modules to avoid including unneeded functions in a
5116 compiled binary. However, compile and exec are still in one module. The "study"
5117 function is split off.
5118
5119 6. The character tables are now in a separate module whose source is generated
5120 by an auxiliary program - but can then be edited by hand if required. There are
5121 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5122 toupper() in the code.
5123
5124 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5125 make them global. Abolish the function for setting them, as the caller can now
5126 set them directly.
5127
5128
5129 Version 0.92 11-Sep-97
5130 ----------------------
5131
5132 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5133 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5134
5135 2. Caseless matching was not working in character classes if the characters in
5136 the pattern were in upper case.
5137
5138 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5139
5140 4. Make PCRE_ANCHORED public and accept as a compile option.
5141
5142 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5143 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5144 pass them.
5145
5146 6. Give an error if bad option bits passed at compile or run time.
5147
5148 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5149 pcretest to cause it to pass that flag.
5150
5151 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5152 options, and the first character, if set.
5153
5154 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5155
5156
5157 Version 0.91 10-Sep-97
5158 ----------------------
5159
5160 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5161 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5162
5163 2. PCRE was looping on encountering an indefinitely repeated back reference to
5164 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5165 Perl does - treats the match as successful.
5166
5167 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5