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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1325 - (show annotations)
Fri May 10 14:03:18 2013 UTC (6 years, 6 months ago) by ph10
File size: 246951 byte(s)
Fix spelling mistakes in comments.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.33 28-April-2013
5 --------------------------
6
7 1. Added 'U' to some constants that are compared to unsigned integers, to
8 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
9 variables that are added to signed variables, to ensure the result is
10 signed and can be negated.
11
12 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
13 CMake config files.
14
15 3. Revise the creation of config.h.generic so that all boolean macros are
16 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
17 overriding via -D on the command line possible.
18
19 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
20 to unsigned int is reported to make a quite noticeable speed difference in
21 a specific Windows environment. Testing on Linux did also appear to show
22 some benefit (and it is clearly not harmful). Also fixed the definition of
23 Xop which should be unsigned.
24
25 5. Related to (4), changing the definition of the intermediate variable cc
26 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
27 improvements.
28
29 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
30 unnecessary spaces.
31
32 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
33 and later.
34
35 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
36
37 9. Optimizing fast_forward_start_bits in JIT.
38
39 10. Adding experimental support for callouts in JIT, and fixing some
40 issues revealed during this work. Namely:
41
42 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
43
44 (b) Minimum length was not checked before the matching is started.
45
46 11. The value of capture_last that is passed to callouts was incorrect in some
47 cases when there was a capture on one path that was subsequently abandoned
48 after a backtrack. Also, the capture_last value is now reset after a
49 recursion, since all captures are also reset in this case.
50
51 12. The interpreter no longer returns the "too many substrings" error in the
52 case when an overflowing capture is in a branch that is subsequently
53 abandoned after a backtrack.
54
55 13. In the pathological case when an offset vector of size 2 is used, pcretest
56 now prints out the matched string after a yield of 0 or 1.
57
58 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
59 Only supported by the JIT compiler at the moment.
60
61 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
62
63 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
64 offset of the starting point of the matching process, provided the offsets
65 vector is large enough.
66
67 17. The \A escape now records a lookbehind value of 1, though its execution
68 does not actually inspect the previous character. This is to ensure that,
69 in partial multi-segment matching, at least one character from the old
70 segment is retained when a new segment is processed. Otherwise, if there
71 are no lookbehinds in the pattern, \A might match incorrectly at the start
72 of a new segment.
73
74 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
75
76 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
77 gives some modest performance improvement in 8-bit mode.
78
79 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
80 be expressed in certain programming languages using Universal Character
81 Names.
82
83 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
84 which points out that "non characters" are not "characters that may not
85 appear in Unicode strings" but rather "characters that are reserved for
86 internal use and have only local meaning".
87
88 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
89 there was a conditional group that depended on an assertion, if the
90 assertion was false, the callout that immediately followed the alternation
91 in the condition was skipped when pcre_exec() was used for matching.
92
93 23. Allow an explicit callout to be inserted before an assertion that is the
94 condition for a conditional group, for compatibility with automatic
95 callouts, which always insert a callout at this point.
96
97 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
98 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
99
100 25. (*PRUNE) is now supported by the JIT compiler.
101
102 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
103
104 27. Fix the case where there are two or more SKIPs with arguments that may be
105 ignored.
106
107 28. (*SKIP) is now supported by the JIT compiler.
108
109 29. (*THEN) is now supported by the JIT compiler.
110
111 30. Update RunTest with additional test selector options.
112
113 31. The way PCRE handles backtracking verbs has been changed in two ways.
114
115 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
116 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
117 backtracking. In some cases this makes it more Perl-compatible, but Perl's
118 rather obscure rules do not always do the same thing.
119
120 (2) Previously, backtracking verbs were confined within assertions. This is
121 no longer the case for positive assertions, except for (*ACCEPT). Again,
122 this sometimes improves Perl compatibility, and sometimes does not.
123
124 32. A number of tests that were in test 2 because Perl did things differently
125 have been moved to test 1, because either Perl or PCRE has changed, and
126 these tests are now compatible.
127
128 32. Control verbs are handled in the same way in JIT and interpreter.
129
130 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
131 contained a forward subroutine reference caused a compile error.
132
133 34. Auto-detect and optimize limited repetitions in JIT.
134
135 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
136 blocking (*UTF) etc.
137
138 36. In the interpreter, maximizing pattern repetitions for characters and
139 character types now use tail recursion, which reduces stack usage.
140
141 37. The value of the max lookbehind was not correctly preserved if a compiled
142 and saved regex was reloaded on a host of different endianness.
143
144 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
145 of the compiled pattern block, expand the flags field from 16 to 32 bits
146 because it was almost full.
147
148 39. Try madvise first before posix_madvise.
149
150 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
151 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
152 original change locked out all matches of empty strings. This has been
153 changed so that one match of an empty string per line is recognized.
154 Subsequent searches on the same line (for colouring or for --only-matching,
155 for example) do not recognize empty strings.
156
157 41. Applied a user patch to fix a number of spelling mistakes in comments.
158
159
160 Version 8.32 30-November-2012
161 -----------------------------
162
163 1. Improved JIT compiler optimizations for first character search and single
164 character iterators.
165
166 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
167 Patch by Daniel Richard G.
168
169 3. Single character iterator optimizations in the JIT compiler.
170
171 4. Improved JIT compiler optimizations for character ranges.
172
173 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
174 Reported by Giuseppe D'Angelo.
175
176 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
177 of a line, was being set incorrectly in cases where .* appeared inside
178 atomic brackets at the start of a pattern, or where there was a subsequent
179 *PRUNE or *SKIP.
180
181 7. Improved instruction cache flush for POWER/PowerPC.
182 Patch by Daniel Richard G.
183
184 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
185 grep:
186
187 (a) There is now no limit to the number of patterns to be matched.
188
189 (b) An error is given if a pattern is too long.
190
191 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
192 are now supported.
193
194 (d) --exclude-from and --include-from (multiple use) have been added.
195
196 (e) Exclusions and inclusions now apply to all files and directories, not
197 just to those obtained from scanning a directory recursively.
198
199 (f) Multiple uses of -f and --file-list are now supported.
200
201 (g) In a Windows environment, the default for -d has been changed from
202 "read" (the GNU grep default) to "skip", because otherwise the presence
203 of a directory in the file list provokes an error.
204
205 (h) The documentation has been revised and clarified in places.
206
207 9. Improve the matching speed of capturing brackets.
208
209 10. Changed the meaning of \X so that it now matches a Unicode extended
210 grapheme cluster.
211
212 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
213 out POSIX threads when JIT support is configured.
214
215 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
216
217 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
218 structure could go wrong in environments where size_t is not the same size
219 as int.
220
221 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
222 set.
223
224 15. The EBCDIC support had decayed; later updates to the code had included
225 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
226 general tidy up of EBCDIC-related issues, and the documentation was also
227 not quite right. There is now a test that can be run on ASCII systems to
228 check some of the EBCDIC-related things (but is it not a full test).
229
230 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
231 in a small tidy to the code.
232
233 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
234
235 18. If the --only-matching (-o) option in pcregrep is specified multiple
236 times, each one causes appropriate output. For example, -o1 -o2 outputs the
237 substrings matched by the 1st and 2nd capturing parentheses. A separating
238 string can be specified by --om-separator (default empty).
239
240 19. Improving the first n character searches.
241
242 20. Turn case lists for horizontal and vertical white space into macros so that
243 they are defined only once.
244
245 21. This set of changes together give more compatible Unicode case-folding
246 behaviour for characters that have more than one other case when UCP
247 support is available.
248
249 (a) The Unicode property table now has offsets into a new table of sets of
250 three or more characters that are case-equivalent. The MultiStage2.py
251 script that generates these tables (the pcre_ucd.c file) now scans
252 CaseFolding.txt instead of UnicodeData.txt for character case
253 information.
254
255 (b) The code for adding characters or ranges of characters to a character
256 class has been abstracted into a generalized function that also handles
257 case-independence. In UTF-mode with UCP support, this uses the new data
258 to handle characters with more than one other case.
259
260 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
261 whose other case is greater than 256 are now correctly matched
262 caselessly. Previously, the high codepoint matched the low one, but not
263 vice versa.
264
265 (d) The processing of \h, \H, \v, and \ in character classes now makes use
266 of the new class addition function, using character lists defined as
267 macros alongside the case definitions of 20 above.
268
269 (e) Caseless back references now work with characters that have more than
270 one other case.
271
272 (f) General caseless matching of characters with more than one other case
273 is supported.
274
275 22. Unicode character properties were updated from Unicode 6.2.0
276
277 23. Improved CMake support under Windows. Patch by Daniel Richard G.
278
279 24. Add support for 32-bit character strings, and UTF-32
280
281 25. Major JIT compiler update (code refactoring and bugfixing).
282 Experimental Sparc 32 support is added.
283
284 26. Applied a modified version of Daniel Richard G's patch to create
285 pcre.h.generic and config.h.generic by "make" instead of in the
286 PrepareRelease script.
287
288 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
289 pcre_compile.c when checking for a zero character.
290
291 28. Introducing a native interface for JIT. Through this interface, the compiled
292 machine code can be directly executed. The purpose of this interface is to
293 provide fast pattern matching, so several sanity checks are not performed.
294 However, feature tests are still performed. The new interface provides
295 1.4x speedup compared to the old one.
296
297 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
298 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
299 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
300
301 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
302 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
303 the "old" RFC 2279). Instead, it ended up passing a negative length to
304 pcre_exec().
305
306 31. Add support for GCC's visibility feature to hide internal functions.
307
308 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
309 "unknown -C option" after outputting 0 or 1.
310
311 33. There is now support for generating a code coverage report for the test
312 suite in environments where gcc is the compiler and lcov is installed. This
313 is mainly for the benefit of the developers.
314
315 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
316 unaddressable using valgrind annotations, allowing valgrind to detect
317 invalid memory accesses. This is mainly for the benefit of the developers.
318
319 25. (*UTF) can now be used to start a pattern in any of the three libraries.
320
321 26. Give configure error if --enable-cpp but no C++ compiler found.
322
323
324 Version 8.31 06-July-2012
325 -------------------------
326
327 1. Fixing a wrong JIT test case and some compiler warnings.
328
329 2. Removed a bashism from the RunTest script.
330
331 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
332 to unsigned type, result still unsigned" that was given by an MS compiler
333 on encountering the code "-sizeof(xxx)".
334
335 4. Partial matching support is added to the JIT compiler.
336
337 5. Fixed several bugs concerned with partial matching of items that consist
338 of more than one character:
339
340 (a) /^(..)\1/ did not partially match "aba" because checking references was
341 done on an "all or nothing" basis. This also applied to repeated
342 references.
343
344 (b) \R did not give a hard partial match if \r was found at the end of the
345 subject.
346
347 (c) \X did not give a hard partial match after matching one or more
348 characters at the end of the subject.
349
350 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
351 a partial match for the string "\r".
352
353 (e) When newline was set to CRLF, the metacharacter "." did not recognize
354 a partial match for a CR character at the end of the subject string.
355
356 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
357 running pcretest, the text "(JIT)" added to the output whenever JIT is
358 actually used to run the match.
359
360 7. Individual JIT compile options can be set in pcretest by following -s+[+]
361 or /S+[+] with a digit between 1 and 7.
362
363 8. OP_NOT now supports any UTF character not just single-byte ones.
364
365 9. (*MARK) control verb is now supported by the JIT compiler.
366
367 10. The command "./RunTest list" lists the available tests without actually
368 running any of them. (Because I keep forgetting what they all are.)
369
370 11. Add PCRE_INFO_MAXLOOKBEHIND.
371
372 12. Applied a (slightly modified) user-supplied patch that improves performance
373 when the heap is used for recursion (compiled with --disable-stack-for-
374 recursion). Instead of malloc and free for each heap frame each time a
375 logical recursion happens, frames are retained on a chain and re-used where
376 possible. This sometimes gives as much as 30% improvement.
377
378 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
379 call.
380
381 14. As documented, (*COMMIT) is now confined to within a positive assertion.
382
383 15. It is now possible to link pcretest with libedit as an alternative to
384 libreadline.
385
386 16. (*COMMIT) control verb is now supported by the JIT compiler.
387
388 17. The Unicode data tables have been updated to Unicode 6.1.0.
389
390 18. Added --file-list option to pcregrep.
391
392 19. Added binary file support to pcregrep, including the -a, --binary-files,
393 -I, and --text options.
394
395 20. The madvise function is renamed for posix_madvise for QNX compatibility
396 reasons. Fixed by Giuseppe D'Angelo.
397
398 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
399
400 22. Changed the option for creating symbolic links for 16-bit man pages from
401 -s to -sf so that re-installing does not cause issues.
402
403 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
404
405 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
406 matches in certain environments (the workspace was not being correctly
407 retained). Also added to pcre_dfa_exec() a simple plausibility check on
408 some of the workspace data at the beginning of a restart.
409
410 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
411 was not doing so when it should - probably a typo introduced by SVN 528
412 (change 8.10/14).
413
414 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
415 \w+ when the character tables indicated that \x{c4} was a word character.
416 There were several related cases, all because the tests for doing a table
417 lookup were testing for characters less than 127 instead of 255.
418
419 27. If a pattern contains capturing parentheses that are not used in a match,
420 their slots in the ovector are set to -1. For those that are higher than
421 any matched groups, this happens at the end of processing. In the case when
422 there were back references that the ovector was too small to contain
423 (causing temporary malloc'd memory to be used during matching), and the
424 highest capturing number was not used, memory off the end of the ovector
425 was incorrectly being set to -1. (It was using the size of the temporary
426 memory instead of the true size.)
427
428 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
429 ovector size, it uses memory at the end of the block that it has got.
430
431 29. Check for an overlong MARK name and give an error at compile time. The
432 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
433
434 30. JIT compiler update.
435
436 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
437 Rill for the patch.
438
439 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
440
441 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
442
443 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
444 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
445 was enabled.
446
447 35. Improve JIT code generation for greedy plus quantifier.
448
449 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
450 1 to "aa" instead of to an empty string. The bug affected repeated groups
451 that could potentially match an empty string.
452
453 37. Optimizing single character iterators in JIT.
454
455 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
456 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
457 codepoints that are too big for the mode are faulted, and in a UTF mode,
458 disallowed codepoints are also faulted.
459
460 39. If PCRE was compiled with UTF support, in three places in the DFA
461 matcher there was code that should only have been obeyed in UTF mode, but
462 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
463 processing when bytes with values greater than 127 were present. In 16-bit
464 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
465 both cases the values are those that cannot be the first data item in a UTF
466 character. The three items that might have provoked this were recursions,
467 possessively repeated groups, and atomic groups.
468
469 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
470 and pcregrep, because some OS require shared objects to be explicitly
471 passed to ld, causing the link step to fail if they are not.
472
473 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
474 mode, patterns that started with \h* or \R* might be incorrectly matched.
475
476
477 Version 8.30 04-February-2012
478 -----------------------------
479
480 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
481 name is defined in ctype.h.
482
483 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
484 only in quite long subpatterns.
485
486 3. Removed the function pcre_info(), which has been obsolete and deprecated
487 since it was replaced by pcre_fullinfo() in February 2000.
488
489 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
490 match a (*MARK), and the match failed at the start of the subject, a
491 reference to memory before the start of the subject could occur. This bug
492 was introduced by fix 17 of release 8.21.
493
494 5. A reference to an unset group with zero minimum repetition was giving
495 totally wrong answers (in non-JavaScript-compatibility mode). For example,
496 /(another)?(\1?)test/ matched against "hello world test". This bug was
497 introduced in release 8.13.
498
499 6. Add support for 16-bit character strings (a large amount of work involving
500 many changes and refactorings).
501
502 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
503 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
504 from a file.
505
506 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
507 rounding is not applied in this particular case).
508
509 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
510 if they appear, or are escaped, in patterns.
511
512 10. Get rid of a number of -Wunused-but-set-variable warnings.
513
514 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
515 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
516 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
517 also returns the mark "x". This bug applied to capturing parentheses,
518 non-capturing parentheses, and atomic parentheses. It also applied to some
519 assertions.
520
521 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
522 information out of configure.ac instead of relying on pcre.h.generic, which
523 is not stored in the repository.
524
525 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
526 -lreadline.
527
528 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
529
530 15. Applied Graycode's patch to put the top-level frame on the stack rather
531 than the heap when not using the stack for recursion. This gives a
532 performance improvement in many cases when recursion is not deep.
533
534 16. Experimental code added to "pcretest -C" to output the stack frame size.
535
536
537 Version 8.21 12-Dec-2011
538 ------------------------
539
540 1. Updating the JIT compiler.
541
542 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
543 are added as well.
544
545 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
546 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
547 calling _pcre_jit_exec. Some extra comments are added.
548
549 4. (*MARK) settings inside atomic groups that do not contain any capturing
550 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
551 was introduced by change 18 for 8.20.
552
553 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
554 ECMA-262 standard.
555
556 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
557 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
558 This bug was probably introduced by change 9 of 8.13.
559
560 7. While fixing 6 above, I noticed that a number of other items were being
561 incorrectly rejected as "not fixed length". This arose partly because newer
562 opcodes had not been added to the fixed-length checking code. I have (a)
563 corrected the bug and added tests for these items, and (b) arranged for an
564 error to occur if an unknown opcode is encountered while checking for fixed
565 length instead of just assuming "not fixed length". The items that were
566 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
567 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
568 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
569
570 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
571 being incorrectly compiled and would have given unpredicatble results.
572
573 9. A possessively repeated subpattern with minimum repeat count greater than
574 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
575 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
576 the first (A) could occur when it should not.
577
578 10. Add a cast and remove a redundant test from the code.
579
580 11. JIT should use pcre_malloc/pcre_free for allocation.
581
582 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
583 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
584 is anything other than /usr, -L is still shown).
585
586 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
587
588 14. Perl does not support \N without a following name in a [] class; PCRE now
589 also gives an error.
590
591 15. If a forward reference was repeated with an upper limit of around 2000,
592 it caused the error "internal error: overran compiling workspace". The
593 maximum number of forward references (including repeats) was limited by the
594 internal workspace, and dependent on the LINK_SIZE. The code has been
595 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
596 the default depends on LINK_SIZE. There is a new upper limit (for safety)
597 of around 200,000 forward references. While doing this, I also speeded up
598 the filling in of repeated forward references.
599
600 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
601 incorrectly expecting the subject to contain another "a" after the start.
602
603 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
604 in the match, the SKIP should be ignored. This was not happening; instead
605 the SKIP was being treated as NOMATCH. For patterns such as
606 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
607 tested.
608
609 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
610 now much more compatible with Perl, in particular in cases where the result
611 is a non-match for a non-anchored pattern. For example, if
612 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
613 "m", where previously it did not return a name. A side effect of this
614 change is that for partial matches, the last encountered mark name is
615 returned, as for non matches. A number of tests that were previously not
616 Perl-compatible have been moved into the Perl-compatible test files. The
617 refactoring has had the pleasing side effect of removing one argument from
618 the match() function, thus reducing its stack requirements.
619
620 19. If the /S+ option was used in pcretest to study a pattern using JIT,
621 subsequent uses of /S (without +) incorrectly behaved like /S+.
622
623 21. Retrieve executable code size support for the JIT compiler and fixing
624 some warnings.
625
626 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
627 not work when the shorter character appeared right at the end of the
628 subject string.
629
630 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
631 systems.
632
633 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
634 output it when the /M option is used in pcretest.
635
636 25. The CheckMan script was not being included in the distribution. Also, added
637 an explicit "perl" to run Perl scripts from the PrepareRelease script
638 because this is reportedly needed in Windows.
639
640 26. If study data was being save in a file and studying had not found a set of
641 "starts with" bytes for the pattern, the data written to the file (though
642 never used) was taken from uninitialized memory and so caused valgrind to
643 complain.
644
645 27. Updated RunTest.bat as provided by Sheri Pierce.
646
647 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
648
649 29. Computation of memory usage for the table of capturing group names was
650 giving an unnecessarily large value.
651
652
653 Version 8.20 21-Oct-2011
654 ------------------------
655
656 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
657 a POSIX class. After further experiments with Perl, which convinced me that
658 Perl has bugs and confusions, a closing square bracket is no longer allowed
659 in a POSIX name. This bug also affected patterns with classes that started
660 with full stops.
661
662 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
663 captured substring, but while checking the failing first alternative,
664 substring 1 is temporarily captured. If the output vector supplied to
665 pcre_exec() was not big enough for this capture, the yield of the function
666 was still zero ("insufficient space for captured substrings"). This cannot
667 be totally fixed without adding another stack variable, which seems a lot
668 of expense for a edge case. However, I have improved the situation in cases
669 such as /(a)(b)x|abc/ matched against "abc", where the return code
670 indicates that fewer than the maximum number of slots in the ovector have
671 been set.
672
673 3. Related to (2) above: when there are more back references in a pattern than
674 slots in the output vector, pcre_exec() uses temporary memory during
675 matching, and copies in the captures as far as possible afterwards. It was
676 using the entire output vector, but this conflicts with the specification
677 that only 2/3 is used for passing back captured substrings. Now it uses
678 only the first 2/3, for compatibility. This is, of course, another edge
679 case.
680
681 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
682 main code base, and can be used by building with --enable-jit. When this is
683 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
684 runtime --no-jit option is given.
685
686 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
687 ovector, the return from the function was zero, implying that there were
688 other matches that did not fit. The correct "exactly full" value is now
689 returned.
690
691 6. If a subpattern that was called recursively or as a subroutine contained
692 (*PRUNE) or any other control that caused it to give a non-standard return,
693 invalid errors such as "Error -26 (nested recursion at the same subject
694 position)" or even infinite loops could occur.
695
696 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
697 computing the minimum length on reaching *ACCEPT, and so ended up with the
698 wrong value of 1 rather than 0. Further investigation indicates that
699 computing a minimum subject length in the presence of *ACCEPT is difficult
700 (think back references, subroutine calls), and so I have changed the code
701 so that no minimum is registered for a pattern that contains *ACCEPT.
702
703 8. If (*THEN) was present in the first (true) branch of a conditional group,
704 it was not handled as intended. [But see 16 below.]
705
706 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
707 Sheri Pierce.
708
709 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
710 the first byte in a match must be "a".
711
712 11. Change 17 for 8.13 increased the recursion depth for patterns like
713 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
714 pattern contains any instances of (*THEN). If it does not, the old
715 optimizations are restored. It would be nice to do this on a per-group
716 basis, but at the moment that is not feasible.
717
718 12. In some environments, the output of pcretest -C is CRLF terminated. This
719 broke RunTest's code that checks for the link size. A single white space
720 character after the value is now allowed for.
721
722 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
723 For "fr", it uses the Windows-specific input and output files.
724
725 14. If (*THEN) appeared in a group that was called recursively or as a
726 subroutine, it did not work as intended. [But see next item.]
727
728 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
729 pattern fragments (but not containing any | characters). If A and B are
730 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
731 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
732 D. In other words, Perl considers parentheses that do not contain any |
733 characters to be part of a surrounding alternative, whereas PCRE was
734 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
735 differently. PCRE now behaves in the same way as Perl, except in the case
736 of subroutine/recursion calls such as (?1) which have in any case always
737 been different (but PCRE had them first :-).
738
739 16. Related to 15 above: Perl does not treat the | in a conditional group as
740 creating alternatives. Such a group is treated in the same way as an
741 ordinary group without any | characters when processing (*THEN). PCRE has
742 been changed to match Perl's behaviour.
743
744 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
745 RunGrepTest script failed.
746
747 18. Change 22 for version 13 caused atomic groups to use more stack. This is
748 inevitable for groups that contain captures, but it can lead to a lot of
749 stack use in large patterns. The old behaviour has been restored for atomic
750 groups that do not contain any capturing parentheses.
751
752 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
753 suppress the check for a minimum subject length at run time. (If it was
754 given to pcre_exec() or pcre_dfa_exec() it did work.)
755
756 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
757 fail to work when decoding hex characters in data strings in EBCDIC
758 environments.
759
760 21. It appears that in at least one Mac OS environment, the isxdigit() function
761 is implemented as a macro that evaluates to its argument more than once,
762 contravening the C 90 Standard (I haven't checked a later standard). There
763 was an instance in pcretest which caused it to go wrong when processing
764 \x{...} escapes in subject strings. The has been rewritten to avoid using
765 things like p++ in the argument of isxdigit().
766
767
768 Version 8.13 16-Aug-2011
769 ------------------------
770
771 1. The Unicode data tables have been updated to Unicode 6.0.0.
772
773 2. Two minor typos in pcre_internal.h have been fixed.
774
775 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
776 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
777 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
778
779 4. There were a number of related bugs in the code for matching backrefences
780 caselessly in UTF-8 mode when codes for the characters concerned were
781 different numbers of bytes. For example, U+023A and U+2C65 are an upper
782 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
783 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
784 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
785 2-byte code at the end of the subject (it thought there wasn't enough data
786 left).
787
788 5. Comprehensive information about what went wrong is now returned by
789 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
790 as the output vector has at least 2 elements. The offset of the start of
791 the failing character and a reason code are placed in the vector.
792
793 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
794 now returned is for the first byte of the failing character, instead of the
795 last byte inspected. This is an incompatible change, but I hope it is small
796 enough not to be a problem. It makes the returned offset consistent with
797 pcre_exec() and pcre_dfa_exec().
798
799 7. pcretest now gives a text phrase as well as the error number when
800 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
801 failure, the offset and reason code are output.
802
803 8. When \R was used with a maximizing quantifier it failed to skip backwards
804 over a \r\n pair if the subsequent match failed. Instead, it just skipped
805 back over a single character (\n). This seems wrong (because it treated the
806 two characters as a single entity when going forwards), conflicts with the
807 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
808 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
809 has been changed.
810
811 9. Some internal refactoring has changed the processing so that the handling
812 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
813 time (the PCRE_DOTALL option was changed this way some time ago: version
814 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
815 which was always a bit of a fudge. It also means that there is one less
816 argument for the match() function, which reduces its stack requirements
817 slightly. This change also fixes an incompatibility with Perl: the pattern
818 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
819
820 10. More internal refactoring has drastically reduced the number of recursive
821 calls to match() for possessively repeated groups such as (abc)++ when
822 using pcre_exec().
823
824 11. While implementing 10, a number of bugs in the handling of groups were
825 discovered and fixed:
826
827 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
828 (a|)*(?1) gave a compile-time internal error.
829 ((a|)+)+ did not notice that the outer group could match an empty string.
830 (^a|^)+ was not marked as anchored.
831 (.*a|.*)+ was not marked as matching at start or after a newline.
832
833 12. Yet more internal refactoring has removed another argument from the match()
834 function. Special calls to this function are now indicated by setting a
835 value in a variable in the "match data" data block.
836
837 13. Be more explicit in pcre_study() instead of relying on "default" for
838 opcodes that mean there is no starting character; this means that when new
839 ones are added and accidentally left out of pcre_study(), testing should
840 pick them up.
841
842 14. The -s option of pcretest has been documented for ages as being an old
843 synonym of -m (show memory usage). I have changed it to mean "force study
844 for every regex", that is, assume /S for every regex. This is similar to -i
845 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
846 using it. It makes it easier to run collections of tests with and without
847 study enabled, and thereby test pcre_study() more easily. All the standard
848 tests are now run with and without -s (but some patterns can be marked as
849 "never study" - see 20 below).
850
851 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
852 restoration of the capturing data to the outer values was not happening
853 correctly.
854
855 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
856 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
857 pattern had matched an empty string, and so incorrectly returned a no
858 match.
859
860 17. There was optimizing code for the last branch of non-capturing parentheses,
861 and also for the obeyed branch of a conditional subexpression, which used
862 tail recursion to cut down on stack usage. Unfortunately, now that there is
863 the possibility of (*THEN) occurring in these branches, tail recursion is
864 no longer possible because the return has to be checked for (*THEN). These
865 two optimizations have therefore been removed. [But see 8.20/11 above.]
866
867 18. If a pattern containing \R was studied, it was assumed that \R always
868 matched two bytes, thus causing the minimum subject length to be
869 incorrectly computed because \R can also match just one byte.
870
871 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
872 was incorrectly computed.
873
874 20. If /S is present twice on a test pattern in pcretest input, it now
875 *disables* studying, thereby overriding the use of -s on the command line
876 (see 14 above). This is necessary for one or two tests to keep the output
877 identical in both cases.
878
879 21. When (*ACCEPT) was used in an assertion that matched an empty string and
880 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
881
882 22. When an atomic group that contained a capturing parenthesis was
883 successfully matched, but the branch in which it appeared failed, the
884 capturing was not being forgotten if a higher numbered group was later
885 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
886 group 1 to "a", when in fact it should be unset. This applied to multi-
887 branched capturing and non-capturing groups, repeated or not, and also to
888 positive assertions (capturing in negative assertions does not happen
889 in PCRE) and also to nested atomic groups.
890
891 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
892 subject after a captured substring, to make it easier to tell which of a
893 number of identical substrings has been captured.
894
895 24. The way atomic groups are processed by pcre_exec() has been changed so that
896 if they are repeated, backtracking one repetition now resets captured
897 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
898 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
899 "aaa". Previously, it would have been "a". As part of this code
900 refactoring, the way recursive calls are handled has also been changed.
901
902 25. If an assertion condition captured any substrings, they were not passed
903 back unless some other capturing happened later. For example, if
904 (?(?=(a))a) was matched against "a", no capturing was returned.
905
906 26. When studying a pattern that contained subroutine calls or assertions,
907 the code for finding the minimum length of a possible match was handling
908 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
909 group 1 called group 2 while simultaneously a separate group 2 called group
910 1). A stack overflow occurred in this case. I have fixed this by limiting
911 the recursion depth to 10.
912
913 27. Updated RunTest.bat in the distribution to the version supplied by Tom
914 Fortmann. This supports explicit test numbers on the command line, and has
915 argument validation and error reporting.
916
917 28. An instance of \X with an unlimited repeat could fail if at any point the
918 first character it looked at was a mark character.
919
920 29. Some minor code refactoring concerning Unicode properties and scripts
921 should reduce the stack requirement of match() slightly.
922
923 30. Added the '=' option to pcretest to check the setting of unused capturing
924 slots at the end of the pattern, which are documented as being -1, but are
925 not included in the return count.
926
927 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
928 compiled something random. Now it gives a compile-time error (as does
929 Perl).
930
931 32. A *MARK encountered during the processing of a positive assertion is now
932 recorded and passed back (compatible with Perl).
933
934 33. If --only-matching or --colour was set on a pcregrep call whose pattern
935 had alternative anchored branches, the search for a second match in a line
936 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
937 matched the line "0102" twice. The same bug affected patterns that started
938 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
939 twice.
940
941 34. Previously, PCRE did not allow quantification of assertions. However, Perl
942 does, and because of capturing effects, quantifying parenthesized
943 assertions may at times be useful. Quantifiers are now allowed for
944 parenthesized assertions.
945
946 35. A minor code tidy in pcre_compile() when checking options for \R usage.
947
948 36. \g was being checked for fancy things in a character class, when it should
949 just be a literal "g".
950
951 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
952 appearance of a nested POSIX class supersedes an apparent external class.
953 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
954 unescaped square brackets may also appear as part of class names. For
955 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
956 more like Perl. (But see 8.20/1 above.)
957
958 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
959 was because it thought it was \N{name}, which is not supported).
960
961 39. Add minix to OS list not supporting the -S option in pcretest.
962
963 40. PCRE tries to detect cases of infinite recursion at compile time, but it
964 cannot analyze patterns in sufficient detail to catch mutual recursions
965 such as ((?1))((?2)). There is now a runtime test that gives an error if a
966 subgroup is called recursively as a subpattern for a second time at the
967 same position in the subject string. In previous releases this might have
968 been caught by the recursion limit, or it might have run out of stack.
969
970 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
971 happen only once. PCRE was, however incorrectly giving a compile time error
972 "recursive call could loop indefinitely" because it cannot analyze the
973 pattern in sufficient detail. The compile time test no longer happens when
974 PCRE is compiling a conditional subpattern, but actual runaway loops are
975 now caught at runtime (see 40 above).
976
977 42. It seems that Perl allows any characters other than a closing parenthesis
978 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
979 has been changed to be the same.
980
981 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
982 as not to get warnings when autogen.sh is called. Also changed
983 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
984
985 44. To help people who use pcregrep to scan files containing exceedingly long
986 lines, the following changes have been made:
987
988 (a) The default value of the buffer size parameter has been increased from
989 8K to 20K. (The actual buffer used is three times this size.)
990
991 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
992 PCRE is built.
993
994 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
995 to be set at run time.
996
997 (d) Numerical values in pcregrep options can be followed by K or M, for
998 example --buffer-size=50K.
999
1000 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1001 given and the return code is set to 2.
1002
1003 45. Add a pointer to the latest mark to the callout data block.
1004
1005 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1006 partial match of an empty string instead of no match. This was specific to
1007 the use of ".".
1008
1009 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1010 complete match instead of a partial match. This bug was dependent on both
1011 the PCRE_UTF8 and PCRE_DOTALL options being set.
1012
1013 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1014 starting byte set, because \b was not being ignored.
1015
1016
1017 Version 8.12 15-Jan-2011
1018 ------------------------
1019
1020 1. Fixed some typos in the markup of the man pages, and wrote a script that
1021 checks for such things as part of the documentation building process.
1022
1023 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1024 --match-limit and --recursion-limit options (added for 8.11). In
1025 particular, this made one of the standard tests fail. (The integer value
1026 went into the wrong half of a long int.)
1027
1028 3. If the --colour option was given to pcregrep with -v (invert match), it
1029 did strange things, either producing crazy output, or crashing. It should,
1030 of course, ignore a request for colour when reporting lines that do not
1031 match.
1032
1033 4. Another pcregrep bug caused similar problems if --colour was specified with
1034 -M (multiline) and the pattern match finished with a line ending.
1035
1036 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1037 matched in multiline mode, the following line was shown as part of the
1038 match. This seems wrong, so I have changed it.
1039
1040 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1041 the check for further matches in the same line (so they could be coloured)
1042 to overrun the end of the current line. If another match was found, it was
1043 incorrectly shown (and then shown again when found in the next line).
1044
1045 7. If pcregrep was compiled under Windows, there was a reference to the
1046 function pcregrep_exit() before it was defined. I am assuming this was
1047 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1048 reported by a user. I've moved the definition above the reference.
1049
1050
1051 Version 8.11 10-Dec-2010
1052 ------------------------
1053
1054 1. (*THEN) was not working properly if there were untried alternatives prior
1055 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1056 backtracked to try for "b" instead of moving to the next alternative branch
1057 at the same level (in this case, to look for "c"). The Perl documentation
1058 is clear that when (*THEN) is backtracked onto, it goes to the "next
1059 alternative in the innermost enclosing group".
1060
1061 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1062 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1063 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1064 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1065 (*THEN).
1066
1067 3. If \s appeared in a character class, it removed the VT character from
1068 the class, even if it had been included by some previous item, for example
1069 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1070 of \s, but is part of the POSIX "space" class.)
1071
1072 4. A partial match never returns an empty string (because you can always
1073 match an empty string at the end of the subject); however the checking for
1074 an empty string was starting at the "start of match" point. This has been
1075 changed to the "earliest inspected character" point, because the returned
1076 data for a partial match starts at this character. This means that, for
1077 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1078 (previously it gave "no match").
1079
1080 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1081 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1082 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1083 has an implication that the given string is incomplete (because a partial
1084 match is preferred over a full match). For this reason, these items now
1085 give a partial match in this situation. [Aside: previously, the one case
1086 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1087 match rather than a full match, which was wrong by the old rules, but is
1088 now correct.]
1089
1090 6. There was a bug in the handling of #-introduced comments, recognized when
1091 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1092 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1093 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1094 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1095 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1096 places in pcre_compile().
1097
1098 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1099 comments when looking ahead for named forward references to subpatterns,
1100 the only newline sequence it recognized was NL. It now handles newlines
1101 according to the set newline convention.
1102
1103 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1104 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1105 cater for a lack of strerror(). These oversights have been fixed.
1106
1107 9. Added --match-limit and --recursion-limit to pcregrep.
1108
1109 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1110
1111 11. When the -o option was used, pcregrep was setting a return code of 1, even
1112 when matches were found, and --line-buffered was not being honoured.
1113
1114 12. Added an optional parentheses number to the -o and --only-matching options
1115 of pcregrep.
1116
1117 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1118 can match an empty string. The code to do it in pcretest and pcredemo
1119 needed fixing:
1120
1121 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1122 only one byte after an empty string match just before CRLF (this case
1123 just got forgotten; "any" and "anycrlf" were OK).
1124
1125 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1126 mode when an empty string match preceded an ASCII character followed by
1127 a non-ASCII character. (The code for advancing by one character rather
1128 than one byte was nonsense.)
1129
1130 (c) The pcredemo.c sample program did not have any code at all to handle
1131 the cases when CRLF is a valid newline sequence.
1132
1133 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1134 as a starting offset was within the subject string. There is now a new
1135 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1136 negative or greater than the length of the string. In order to test this,
1137 pcretest is extended to allow the setting of negative starting offsets.
1138
1139 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1140 starting offset points to the beginning of a UTF-8 character was
1141 unnecessarily clumsy. I tidied it up.
1142
1143 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1144 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1145
1146 17. Nobody had reported that the --include_dir option, which was added in
1147 release 7.7 should have been called --include-dir (hyphen, not underscore)
1148 for compatibility with GNU grep. I have changed it to --include-dir, but
1149 left --include_dir as an undocumented synonym, and the same for
1150 --exclude-dir, though that is not available in GNU grep, at least as of
1151 release 2.5.4.
1152
1153 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1154 characters from a string of bytes) have been redefined so as not to use
1155 loops, in order to improve performance in some environments. At the same
1156 time, I abstracted some of the common code into auxiliary macros to save
1157 repetition (this should not affect the compiled code).
1158
1159 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1160 compile-time error is now given if \c is not followed by an ASCII
1161 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1162 different, and any byte value is allowed.)
1163
1164 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1165 START_OPTIMIZE option, which is now allowed at compile time - but just
1166 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1167 to pcregrep and other applications that have no direct access to PCRE
1168 options. The new /Y option in pcretest sets this option when calling
1169 pcre_compile().
1170
1171 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1172 back references. Groups containing recursive back references were forced to
1173 be atomic by that change, but in the case of named groups, the amount of
1174 memory required was incorrectly computed, leading to "Failed: internal
1175 error: code overflow". This has been fixed.
1176
1177 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1178 pcretest.c, to avoid build problems in some Borland environments.
1179
1180
1181 Version 8.10 25-Jun-2010
1182 ------------------------
1183
1184 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1185 THEN.
1186
1187 2. (*ACCEPT) was not working when inside an atomic group.
1188
1189 3. Inside a character class, \B is treated as a literal by default, but
1190 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1191 causes the error). The code is unchanged, but I tidied the documentation.
1192
1193 4. Inside a character class, PCRE always treated \R and \X as literals,
1194 whereas Perl faults them if its -w option is set. I have changed PCRE so
1195 that it faults them when PCRE_EXTRA is set.
1196
1197 5. Added support for \N, which always matches any character other than
1198 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1199
1200 6. When compiling pcregrep with newer versions of gcc which may have
1201 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1202 declared with attribute warn_unused_result" were given. Just casting the
1203 result to (void) does not stop the warnings; a more elaborate fudge is
1204 needed. I've used a macro to implement this.
1205
1206 7. Minor change to pcretest.c to avoid a compiler warning.
1207
1208 8. Added four artifical Unicode properties to help with an option to make
1209 \s etc use properties (see next item). The new properties are: Xan
1210 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1211
1212 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1213 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1214 this option. Modified pcretest to add /W to test this facility. Added
1215 REG_UCP to make it available via the POSIX interface.
1216
1217 10. Added --line-buffered to pcregrep.
1218
1219 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1220 studied, and the match started with a letter with a code point greater than
1221 127 whose first byte was different to the first byte of the other case of
1222 the letter, the other case of this starting letter was not recognized
1223 (#976).
1224
1225 12. If a pattern that was studied started with a repeated Unicode property
1226 test, for example, \p{Nd}+, there was the theoretical possibility of
1227 setting up an incorrect bitmap of starting bytes, but fortunately it could
1228 not have actually happened in practice until change 8 above was made (it
1229 added property types that matched character-matching opcodes).
1230
1231 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1232 possible starting bytes for non-anchored patterns.
1233
1234 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1235 \R, and also a number of cases that involve Unicode properties, both
1236 explicit and implicit when PCRE_UCP is set.
1237
1238 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1239 input, it could crash or give wrong results if characters with values
1240 greater than 0xc0 were present in the subject string. (Detail: it assumed
1241 UTF-8 input when processing these items.)
1242
1243 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1244 size_t is 64-bit (#991).
1245
1246 17. Added a check for running out of memory when PCRE is compiled with
1247 --disable-stack-for-recursion (#990).
1248
1249 18. If the last data line in a file for pcretest does not have a newline on
1250 the end, a newline was missing in the output.
1251
1252 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1253 less than 128) in its various bitmaps. However, there is a facility for
1254 generating tables according to the current locale when PCRE is compiled. It
1255 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1256 characters, are recognized by isspace() and therefore were getting set in
1257 these tables, and indeed these tables seem to approximate to ISO 8859. This
1258 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1259 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1260 which of course cannot start UTF-8 characters. I have changed the code so
1261 that only real ASCII characters (less than 128) and the correct starting
1262 bytes for UTF-8 encodings are set for characters greater than 127 when in
1263 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1264 altogether.)
1265
1266 20. Added the /T option to pcretest so as to be able to run tests with non-
1267 standard character tables, thus making it possible to include the tests
1268 used for 19 above in the standard set of tests.
1269
1270 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1271 reference to a subpattern the other side of a comment that contains an
1272 opening parenthesis caused either an internal compiling error, or a
1273 reference to the wrong subpattern.
1274
1275
1276 Version 8.02 19-Mar-2010
1277 ------------------------
1278
1279 1. The Unicode data tables have been updated to Unicode 5.2.0.
1280
1281 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1282 configured.
1283
1284 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1285 original author of that file, following a query about its status.
1286
1287 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1288 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1289
1290 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1291 quantifier applied to a forward-referencing subroutine call, could compile
1292 incorrect code or give the error "internal error: previously-checked
1293 referenced subpattern not found".
1294
1295 6. Both MS Visual Studio and Symbian OS have problems with initializing
1296 variables to point to external functions. For these systems, therefore,
1297 pcre_malloc etc. are now initialized to local functions that call the
1298 relevant global functions.
1299
1300 7. There were two entries missing in the vectors called coptable and poptable
1301 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1302 I've fixed the data, and added a kludgy way of testing at compile time that
1303 the lengths are correct (equal to the number of opcodes).
1304
1305 8. Following on from 7, I added a similar kludge to check the length of the
1306 eint vector in pcreposix.c.
1307
1308 9. Error texts for pcre_compile() are held as one long string to avoid too
1309 much relocation at load time. To find a text, the string is searched,
1310 counting zeros. There was no check for running off the end of the string,
1311 which could happen if a new error number was added without updating the
1312 string.
1313
1314 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1315
1316 11. \K was not working if it appeared in an atomic group or in a group that
1317 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1318 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1319 the assertion is positive, but not if it is negative.
1320
1321 12. Change 11 fortuitously reduced the size of the stack frame used in the
1322 "match()" function of pcre_exec.c by one pointer. Forthcoming
1323 implementation of support for (*MARK) will need an extra pointer on the
1324 stack; I have reserved it now, so that the stack frame size does not
1325 decrease.
1326
1327 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1328 item in branch that calls a recursion is a subroutine call - as in the
1329 second branch in the above example - was incorrectly given the compile-
1330 time error "recursive call could loop indefinitely" because pcre_compile()
1331 was not correctly checking the subroutine for matching a non-empty string.
1332
1333 14. The checks for overrunning compiling workspace could trigger after an
1334 overrun had occurred. This is a "should never occur" error, but it can be
1335 triggered by pathological patterns such as hundreds of nested parentheses.
1336 The checks now trigger 100 bytes before the end of the workspace.
1337
1338 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1339
1340
1341 Version 8.01 19-Jan-2010
1342 ------------------------
1343
1344 1. If a pattern contained a conditional subpattern with only one branch (in
1345 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1346 computed the wrong minimum data length (which is of course zero for such
1347 subpatterns). This could cause incorrect "no match" results.
1348
1349 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1350 the pattern is reset in the first branch, pcre_compile() failed with
1351 "internal error: code overflow at offset...". This happened only when
1352 the reset was to the original external option setting. (An optimization
1353 abstracts leading options settings into an external setting, which was the
1354 cause of this.)
1355
1356 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1357 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1358 assertion pattern did not match (meaning that the assertion was true), it
1359 was incorrectly treated as false if the SKIP had been reached during the
1360 matching. This also applied to assertions used as conditions.
1361
1362 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1363 assertion subpattern, including such a pattern used as a condition,
1364 unpredictable results occurred, instead of the error return
1365 PCRE_ERROR_DFA_UITEM.
1366
1367 5. The C++ GlobalReplace function was not working like Perl for the special
1368 situation when an empty string is matched. It now does the fancy magic
1369 stuff that is necessary.
1370
1371 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1372 removed. (These were left over from very, very early versions of PCRE.)
1373
1374 7. Some cosmetic changes to the code to make life easier when compiling it
1375 as part of something else:
1376
1377 (a) Change DEBUG to PCRE_DEBUG.
1378
1379 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1380 called "current" as "current_branch", to prevent a collision with the
1381 Linux macro when compiled as a kernel module.
1382
1383 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1384 prevent a collision with the Linux macro when compiled as a kernel
1385 module.
1386
1387 8. In pcre_compile() there are some checks for integer overflows that used to
1388 cast potentially large values to (double). This has been changed to that
1389 when building, a check for int64_t is made, and if it is found, it is used
1390 instead, thus avoiding the use of floating point arithmetic. (There is no
1391 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1392 double.
1393
1394 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1395 2005 (difference between two addresses compared to an unsigned value).
1396
1397 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1398 custom one, because of the following reported problem in Windows:
1399
1400 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1401 under Win32.
1402 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1403 therefore missing the function definition.
1404 - The compiler thus generates a "C" signature for the test function.
1405 - The linker fails to find the "C" function.
1406 - PCRE fails to configure if asked to do so against libbz2.
1407
1408 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1409 messages were output:
1410
1411 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1412 rerunning libtoolize, to keep the correct libtool macros in-tree.
1413 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1414
1415 I have done both of these things.
1416
1417 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1418 most of the time, it *can* run out if it is given a pattern that contains a
1419 runaway infinite recursion. I updated the discussion in the pcrestack man
1420 page.
1421
1422 13. Now that we have gone to the x.xx style of version numbers, the minor
1423 version may start with zero. Using 08 or 09 is a bad idea because users
1424 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1425 interpreted as invalid octal numbers. I've updated the previous comment in
1426 configure.ac, and also added a check that gives an error if 08 or 09 are
1427 used.
1428
1429 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1430 causing partial matching to fail when the end of the subject matched \W
1431 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1432
1433 15. There were some discrepancies between the declarations in pcre_internal.h
1434 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1435 their definitions. The declarations used "const uschar *" and the
1436 definitions used USPTR. Even though USPTR is normally defined as "const
1437 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1438 reported that: "This difference in casting confuses some C++ compilers, for
1439 example, SunCC recognizes above declarations as different functions and
1440 generates broken code for hbpcre." I have changed the declarations to use
1441 USPTR.
1442
1443 16. GNU libtool is named differently on some systems. The autogen.sh script now
1444 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1445 (FreeBSD).
1446
1447 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1448 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1449 comment: "Figure out how to create a longlong from a string: strtoll and
1450 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1451 instance, but it only takes 2 args instead of 3!"
1452
1453 18. A subtle bug concerned with back references has been fixed by a change of
1454 specification, with a corresponding code fix. A pattern such as
1455 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1456 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1457 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1458 same bug. Such groups have to be quantified to be useful, or contained
1459 inside another quantified group. (If there's no repetition, the reference
1460 can never match.) The problem arises because, having left the group and
1461 moved on to the rest of the pattern, a later failure that backtracks into
1462 the group uses the captured value from the final iteration of the group
1463 rather than the correct earlier one. I have fixed this in PCRE by forcing
1464 any group that contains a reference to itself to be an atomic group; that
1465 is, there cannot be any backtracking into it once it has completed. This is
1466 similar to recursive and subroutine calls.
1467
1468
1469 Version 8.00 19-Oct-09
1470 ----------------------
1471
1472 1. The table for translating pcre_compile() error codes into POSIX error codes
1473 was out-of-date, and there was no check on the pcre_compile() error code
1474 being within the table. This could lead to an OK return being given in
1475 error.
1476
1477 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1478 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1479 in a Windows environment.
1480
1481 3. The pcregrep --count option prints the count for each file even when it is
1482 zero, as does GNU grep. However, pcregrep was also printing all files when
1483 --files-with-matches was added. Now, when both options are given, it prints
1484 counts only for those files that have at least one match. (GNU grep just
1485 prints the file name in this circumstance, but including the count seems
1486 more useful - otherwise, why use --count?) Also ensured that the
1487 combination -clh just lists non-zero counts, with no names.
1488
1489 4. The long form of the pcregrep -F option was incorrectly implemented as
1490 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1491 but it seems right to fix it, and I didn't think it was worth preserving
1492 the old behaviour.
1493
1494 5. The command line items --regex=pattern and --regexp=pattern were not
1495 recognized by pcregrep, which required --regex pattern or --regexp pattern
1496 (with a space rather than an '='). The man page documented the '=' forms,
1497 which are compatible with GNU grep; these now work.
1498
1499 6. No libpcreposix.pc file was created for pkg-config; there was just
1500 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1501
1502 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1503 when UCP support is not needed, by modifying the Python script that
1504 generates it from Unicode data files. This should not matter if the module
1505 is correctly used as a library, but I received one complaint about 50K of
1506 unwanted data. My guess is that the person linked everything into his
1507 program rather than using a library. Anyway, it does no harm.
1508
1509 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1510 was a minimum greater than 1 for a wide character in a possessive
1511 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1512 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1513 character. Chaos in the form of incorrect output or a compiling loop could
1514 result.
1515
1516 9. The restrictions on what a pattern can contain when partial matching is
1517 requested for pcre_exec() have been removed. All patterns can now be
1518 partially matched by this function. In addition, if there are at least two
1519 slots in the offset vector, the offset of the earliest inspected character
1520 for the match and the offset of the end of the subject are set in them when
1521 PCRE_ERROR_PARTIAL is returned.
1522
1523 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1524 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1525 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1526 and may be more useful for multi-segment matching.
1527
1528 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1529 used to be given if ever the end of the subject was reached; now it is
1530 given only if matching could not proceed because another character was
1531 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1532 string "Z", which now yields "no match" instead of "partial match". In the
1533 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1534 final character ended with (*FAIL).
1535
1536 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1537 if the pattern had a "must contain" character that was already found in the
1538 earlier partial match, unless partial matching was again requested. For
1539 example, with the pattern /dog.(body)?/, the "must contain" character is
1540 "g". If the first part-match was for the string "dog", restarting with
1541 "sbody" failed. This bug has been fixed.
1542
1543 13. The string returned by pcre_dfa_exec() after a partial match has been
1544 changed so that it starts at the first inspected character rather than the
1545 first character of the match. This makes a difference only if the pattern
1546 starts with a lookbehind assertion or \b or \B (\K is not supported by
1547 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1548 matching functions compatible, and I think it's the right thing to do.
1549
1550 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1551 so that the demonstration program is easily available in environments where
1552 PCRE has not been installed from source.
1553
1554 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1555 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1556 library.
1557
1558 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1559 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1560 is not the first non-POSIX option to be added. Clearly some people find
1561 these options useful.
1562
1563 17. If a caller to the POSIX matching function regexec() passes a non-zero
1564 value for nmatch with a NULL value for pmatch, the value of
1565 nmatch is forced to zero.
1566
1567 18. RunGrepTest did not have a test for the availability of the -u option of
1568 the diff command, as RunTest does. It now checks in the same way as
1569 RunTest, and also checks for the -b option.
1570
1571 19. If an odd number of negated classes containing just a single character
1572 interposed, within parentheses, between a forward reference to a named
1573 subpattern and the definition of the subpattern, compilation crashed with
1574 an internal error, complaining that it could not find the referenced
1575 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1576 [The bug was that it was starting one character too far in when skipping
1577 over the character class, thus treating the ] as data rather than
1578 terminating the class. This meant it could skip too much.]
1579
1580 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1581 /g option in pcretest when the pattern contains \K, which makes it possible
1582 to have an empty string match not at the start, even when the pattern is
1583 anchored. Updated pcretest and pcredemo to use this option.
1584
1585 21. If the maximum number of capturing subpatterns in a recursion was greater
1586 than the maximum at the outer level, the higher number was returned, but
1587 with unset values at the outer level. The correct (outer level) value is
1588 now given.
1589
1590 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1591 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1592 make it do so. The string so far is captured, making this feature
1593 compatible with Perl.
1594
1595 23. The tests have been re-organized, adding tests 11 and 12, to make it
1596 possible to check the Perl 5.10 features against Perl 5.10.
1597
1598 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1599 pattern matches a fixed length string. PCRE did not allow this; now it
1600 does. Neither allows recursion.
1601
1602 25. I finally figured out how to implement a request to provide the minimum
1603 length of subject string that was needed in order to match a given pattern.
1604 (It was back references and recursion that I had previously got hung up
1605 on.) This code has now been added to pcre_study(); it finds a lower bound
1606 to the length of subject needed. It is not necessarily the greatest lower
1607 bound, but using it to avoid searching strings that are too short does give
1608 some useful speed-ups. The value is available to calling programs via
1609 pcre_fullinfo().
1610
1611 26. While implementing 25, I discovered to my embarrassment that pcretest had
1612 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1613 study optimizations had never been tested with that matching function.
1614 Oops. What is worse, even when it was passed study data, there was a bug in
1615 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1616 were also very few tests of studied patterns with pcre_dfa_exec().
1617
1618 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1619 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1620 on the other side of the coin, they are no longer allowed to have different
1621 names, because these cannot be distinguished in PCRE, and this has caused
1622 confusion. (This is a difference from Perl.)
1623
1624 28. When duplicate subpattern names are present (necessarily with different
1625 numbers, as required by 27 above), and a test is made by name in a
1626 conditional pattern, either for a subpattern having been matched, or for
1627 recursion in such a pattern, all the associated numbered subpatterns are
1628 tested, and the overall condition is true if the condition is true for any
1629 one of them. This is the way Perl works, and is also more like the way
1630 testing by number works.
1631
1632
1633 Version 7.9 11-Apr-09
1634 ---------------------
1635
1636 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1637 (pcretest), all targets were linked against these libraries. This included
1638 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1639 libraries. This caused unwanted dependencies to be created. This problem
1640 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1641 pcretest is linked with readline.
1642
1643 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1644 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1645 moved outside it again, because FALSE and TRUE are already defined in AIX,
1646 but BOOL is not.
1647
1648 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1649 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1650
1651 4. The pcregrep documentation said spaces were inserted as well as colons (or
1652 hyphens) following file names and line numbers when outputting matching
1653 lines. This is not true; no spaces are inserted. I have also clarified the
1654 wording for the --colour (or --color) option.
1655
1656 5. In pcregrep, when --colour was used with -o, the list of matching strings
1657 was not coloured; this is different to GNU grep, so I have changed it to be
1658 the same.
1659
1660 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1661 each matching line was coloured. Now it goes on to look for further matches
1662 of any of the test patterns, which is the same behaviour as GNU grep.
1663
1664 7. A pattern that could match an empty string could cause pcregrep to loop; it
1665 doesn't make sense to accept an empty string match in pcregrep, so I have
1666 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1667 seems to be how GNU grep behaves. [But see later change 40 for release
1668 8.33.]
1669
1670 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1671 start or after a newline", because the conditional assertion was not being
1672 correctly handled. The rule now is that both the assertion and what follows
1673 in the first alternative must satisfy the test.
1674
1675 9. If auto-callout was enabled in a pattern with a conditional group whose
1676 condition was an assertion, PCRE could crash during matching, both with
1677 pcre_exec() and pcre_dfa_exec().
1678
1679 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1680 used for matching.
1681
1682 11. Unicode property support in character classes was not working for
1683 characters (bytes) greater than 127 when not in UTF-8 mode.
1684
1685 12. Added the -M command line option to pcretest.
1686
1687 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1688
1689 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1690
1691 16. Added comments and documentation about mis-use of no_arg in the C++
1692 wrapper.
1693
1694 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1695 from Martin Jerabek that uses macro names for all relevant character and
1696 string constants.
1697
1698 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1699 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1700 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1701 these, but not everybody uses configure.
1702
1703 19. A conditional group that had only one branch was not being correctly
1704 recognized as an item that could match an empty string. This meant that an
1705 enclosing group might also not be so recognized, causing infinite looping
1706 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1707 with the subject "ab", where knowledge that the repeated group can match
1708 nothing is needed in order to break the loop.
1709
1710 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1711 exec(), but without supplying a callout function, matching went wrong.
1712
1713 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1714 leak if the size of the offset vector was greater than 30. When the vector
1715 is smaller, the saved offsets during recursion go onto a local stack
1716 vector, but for larger vectors malloc() is used. It was failing to free
1717 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1718 error, in fact).
1719
1720 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1721 heapframe that is used only when UTF-8 support is enabled. This caused no
1722 problem, but was untidy.
1723
1724 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1725 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1726 included within another project.
1727
1728 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1729 slightly modified by me:
1730
1731 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1732 not building pcregrep.
1733
1734 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1735 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1736
1737 25. Forward references, both numeric and by name, in patterns that made use of
1738 duplicate group numbers, could behave incorrectly or give incorrect errors,
1739 because when scanning forward to find the reference group, PCRE was not
1740 taking into account the duplicate group numbers. A pattern such as
1741 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1742
1743 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1744 the feature of a custom pointer more persuasive (as requested by a user).
1745
1746 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1747 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1748 pre-defined.
1749
1750 28. Added support for (*UTF8) at the start of a pattern.
1751
1752 29. Arrange for flags added by the "release type" setting in CMake to be shown
1753 in the configuration summary.
1754
1755
1756 Version 7.8 05-Sep-08
1757 ---------------------
1758
1759 1. Replaced UCP searching code with optimized version as implemented for Ad
1760 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1761 stage table and inline lookup instead of a function, giving speed ups of 2
1762 to 5 times on some simple patterns that I tested. Permission was given to
1763 distribute the MultiStage2.py script that generates the tables (it's not in
1764 the tarball, but is in the Subversion repository).
1765
1766 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1767 scripts.
1768
1769 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1770 a group with a zero qualifier. The result of the study could be incorrect,
1771 or the function might crash, depending on the pattern.
1772
1773 4. Caseless matching was not working for non-ASCII characters in back
1774 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1775 It now works when Unicode Property Support is available.
1776
1777 5. In pcretest, an escape such as \x{de} in the data was always generating
1778 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1779 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1780 truncation.
1781
1782 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1783
1784 7. Added two (int) casts to pcregrep when printing the difference of two
1785 pointers, in case they are 64-bit values.
1786
1787 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1788 test 2 if it fails.
1789
1790 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1791 and a #define of that name to empty if it is not externally set. This is to
1792 allow users of MSVC to set it if necessary.
1793
1794 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1795 the convenience functions in the pcre_get.c source file.
1796
1797 11. An option change at the start of a pattern that had top-level alternatives
1798 could cause overwriting and/or a crash. This command provoked a crash in
1799 some environments:
1800
1801 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1802
1803 This potential security problem was recorded as CVE-2008-2371.
1804
1805 12. For a pattern where the match had to start at the beginning or immediately
1806 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1807 pcre_dfa_exec() could read past the end of the passed subject if there was
1808 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1809 pcretest so that it places the subject at the end of its malloc-ed buffer.
1810
1811 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1812 exec() might read past the end of the data buffer in UTF-8 mode.
1813
1814 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1815 the data contained the byte 0x85 as part of a UTF-8 character within its
1816 first line. This applied both to normal and DFA matching.
1817
1818 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1819 /^[^d]*?$/8 failed to match "abc".
1820
1821 16. Added a missing copyright notice to pcrecpp_internal.h.
1822
1823 17. Make it more clear in the documentation that values returned from
1824 pcre_exec() in ovector are byte offsets, not character counts.
1825
1826 18. Tidied a few places to stop certain compilers from issuing warnings.
1827
1828 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1829 supplied by Stefan Weber. I made a further small update for 7.8 because
1830 there is a change of source arrangements: the pcre_searchfuncs.c module is
1831 replaced by pcre_ucd.c.
1832
1833
1834 Version 7.7 07-May-08
1835 ---------------------
1836
1837 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1838 a string to a long long, pretend we don't even have a long long." This is
1839 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1840
1841 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1842 pre-7.6 versions, which defined a global no_arg variable instead of putting
1843 it in the RE class. (See also #8 below.)
1844
1845 3. Remove a line of dead code, identified by coverity and reported by Nuno
1846 Lopes.
1847
1848 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1849
1850 (1) The include/exclude patterns were being applied to the whole pathnames
1851 of files, instead of just to the final components.
1852
1853 (2) If there was more than one level of directory, the subdirectories were
1854 skipped unless they satisfied the include/exclude conditions. This is
1855 inconsistent with GNU grep (and could even be seen as contrary to the
1856 pcregrep specification - which I improved to make it absolutely clear).
1857 The action now is always to scan all levels of directory, and just
1858 apply the include/exclude patterns to regular files.
1859
1860 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1861 --exclude_dir in the tests to avoid scanning .svn directories.
1862
1863 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1864 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1865 doesn't support NULs in patterns.
1866
1867 7. Added some missing "const"s to declarations of static tables in
1868 pcre_compile.c and pcre_dfa_exec.c.
1869
1870 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1871 caused by fix #2 above. (Subsequently also a second patch to fix the
1872 first patch. And a third patch - this was a messy problem.)
1873
1874 9. Applied Craig's patch to remove the use of push_back().
1875
1876 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1877 matching function regexec().
1878
1879 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1880 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1881 references. PCRE supports relative numbers with this syntax (I don't think
1882 Oniguruma does).
1883
1884 12. Previously, a group with a zero repeat such as (...){0} was completely
1885 omitted from the compiled regex. However, this means that if the group
1886 was called as a subroutine from elsewhere in the pattern, things went wrong
1887 (an internal error was given). Such groups are now left in the compiled
1888 pattern, with a new opcode that causes them to be skipped at execution
1889 time.
1890
1891 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1892 to the way PCRE behaves:
1893
1894 (a) A lone ] character is dis-allowed (Perl treats it as data).
1895
1896 (b) A back reference to an unmatched subpattern matches an empty string
1897 (Perl fails the current match path).
1898
1899 (c) A data ] in a character class must be notated as \] because if the
1900 first data character in a class is ], it defines an empty class. (In
1901 Perl it is not possible to have an empty class.) The empty class []
1902 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1903 The negative empty class [^] matches any one character, independently
1904 of the DOTALL setting.
1905
1906 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1907 non-existent subpattern following a character class starting with ']' and
1908 containing () gave an internal compiling error instead of "reference to
1909 non-existent subpattern". Fortunately, when the pattern did exist, the
1910 compiled code was correct. (When scanning forwards to check for the
1911 existence of the subpattern, it was treating the data ']' as terminating
1912 the class, so got the count wrong. When actually compiling, the reference
1913 was subsequently set up correctly.)
1914
1915 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1916 it was being rejected as not supported by pcre_dfa_exec(), even though
1917 other assertions are supported. I have made pcre_dfa_exec() support
1918 (*FAIL).
1919
1920 16. The implementation of 13c above involved the invention of a new opcode,
1921 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1922 cannot be changed at match time, I realized I could make a small
1923 improvement to matching performance by compiling OP_ALLANY instead of
1924 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1925 on the OP_ANY path.
1926
1927 17. Compiling pcretest on Windows with readline support failed without the
1928 following two fixes: (1) Make the unistd.h include conditional on
1929 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1930
1931 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1932 ncurses library to be included for pcretest when ReadLine support is
1933 requested, but also to allow for it to be overridden. This patch came from
1934 Daniel Bergström.
1935
1936 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1937 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1938 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1939 spotting this.
1940
1941
1942 Version 7.6 28-Jan-08
1943 ---------------------
1944
1945 1. A character class containing a very large number of characters with
1946 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1947 overflow.
1948
1949 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1950 HAVE_LONG_LONG is not defined.
1951
1952 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1953 bring them up to date and include new features. This patch includes:
1954
1955 - Fixed PH's badly added libz and libbz2 support.
1956 - Fixed a problem with static linking.
1957 - Added pcredemo. [But later removed - see 7 below.]
1958 - Fixed dftables problem and added an option.
1959 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1960 HAVE_LONG_LONG.
1961 - Added readline support for pcretest.
1962 - Added an listing of the option settings after cmake has run.
1963
1964 4. A user submitted a patch to Makefile that makes it easy to create
1965 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1966 Makefile.am that cause it to include this special target, without
1967 affecting anything else. Note that the same mingw target plus all
1968 the other distribution libraries and programs are now supported
1969 when configuring with CMake (see 6 below) instead of with
1970 Configure/Make.
1971
1972 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1973 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1974 exported in the Windows port". It has not yet been confirmed that the patch
1975 solves the problem, but it does no harm.
1976
1977 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1978 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1979 with CMake, and also correct the comment about stack recursion.
1980
1981 7. Remove the automatic building of pcredemo from the ./configure system and
1982 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1983 of a program that users should build themselves after PCRE is installed, so
1984 building it automatically is not really right. What is more, it gave
1985 trouble in some build environments.
1986
1987 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1988
1989
1990 Version 7.5 10-Jan-08
1991 ---------------------
1992
1993 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1994 values in parens when parsing an RE using the C++ wrapper."
1995
1996 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1997 Characters greater than 255 were excluded from the class instead of being
1998 included.
1999
2000 3. The same bug as (2) above applied to negated POSIX classes such as
2001 [:^space:].
2002
2003 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2004 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2005 I have changed it.
2006
2007 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2008 first named subpattern) and a construct such as (?&a) would reference the
2009 first named subpattern whose name started with "a" (in other words, the
2010 length check was missing). Both these problems are fixed. "Subpattern name
2011 expected" is now given for (?&) (a zero-length name), and this patch also
2012 makes it give the same error for \k'' (previously it complained that that
2013 was a reference to a non-existent subpattern).
2014
2015 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2016 this is right because (?- can be followed by option settings as well as by
2017 digits. I have, however, made the messages clearer.
2018
2019 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2020 than the number used in the conditional) now cause a compile-time error.
2021 This is actually not compatible with Perl, which accepts such patterns, but
2022 treats the conditional as always being FALSE (as PCRE used to), but it
2023 seems to me that giving a diagnostic is better.
2024
2025 8. Change "alphameric" to the more common word "alphanumeric" in comments
2026 and messages.
2027
2028 9. Fix two occurrences of "backslash" in comments that should have been
2029 "backspace".
2030
2031 10. Remove two redundant lines of code that can never be obeyed (their function
2032 was moved elsewhere).
2033
2034 11. The program that makes PCRE's Unicode character property table had a bug
2035 which caused it to generate incorrect table entries for sequences of
2036 characters that have the same character type, but are in different scripts.
2037 It amalgamated them into a single range, with the script of the first of
2038 them. In other words, some characters were in the wrong script. There were
2039 thirteen such cases, affecting characters in the following ranges:
2040
2041 U+002b0 - U+002c1
2042 U+0060c - U+0060d
2043 U+0061e - U+00612
2044 U+0064b - U+0065e
2045 U+0074d - U+0076d
2046 U+01800 - U+01805
2047 U+01d00 - U+01d77
2048 U+01d9b - U+01dbf
2049 U+0200b - U+0200f
2050 U+030fc - U+030fe
2051 U+03260 - U+0327f
2052 U+0fb46 - U+0fbb1
2053 U+10450 - U+1049d
2054
2055 12. The -o option (show only the matching part of a line) for pcregrep was not
2056 compatible with GNU grep in that, if there was more than one match in a
2057 line, it showed only the first of them. It now behaves in the same way as
2058 GNU grep.
2059
2060 13. If the -o and -v options were combined for pcregrep, it printed a blank
2061 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2062 does the same. The return code can be used to tell if there were any
2063 non-matching lines.
2064
2065 14. Added --file-offsets and --line-offsets to pcregrep.
2066
2067 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2068 infinitely looping recursion. The bug was that positive lookaheads were not
2069 being skipped when checking for a possible empty match (negative lookaheads
2070 and both kinds of lookbehind were skipped).
2071
2072 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2073 inclusion of <windows.h> to before rather than after the definition of
2074 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2075
2076 17. Specifying a possessive quantifier with a specific limit for a Unicode
2077 character property caused pcre_compile() to compile bad code, which led at
2078 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2079 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2080 caused the error; without that there was no problem.
2081
2082 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2083
2084 19. Added --enable-pcretest-libreadline.
2085
2086 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2087 RE::GlobalReplace(). As a result, the number of replacements returned was
2088 double what it should be. I removed one of the increments, but Craig sent a
2089 later patch that removed the other one (the right fix) and added unit tests
2090 that check the return values (which was not done before).
2091
2092 21. Several CMake things:
2093
2094 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2095 the names libpcre and libpcreposix, not just pcre and pcreposix.
2096
2097 (2) The above change means that pcretest and pcregrep are now correctly
2098 linked with the newly-built libraries, not previously installed ones.
2099
2100 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2101
2102 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2103 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2104 UTF-8 newline character). The key issue is that the pattern starts .*;
2105 this means that the match must be either at the beginning, or after a
2106 newline. The bug was in the code for advancing after a failed match and
2107 checking that the new position followed a newline. It was not taking
2108 account of UTF-8 characters correctly.
2109
2110 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2111 character classes. PCRE was not treating the sequence [:...:] as a
2112 character class unless the ... were all letters. Perl, however, seems to
2113 allow any characters between [: and :], though of course it rejects as
2114 unknown any "names" that contain non-letters, because all the known class
2115 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2116 for example, whereas PCRE did not - it did not recognize a POSIX character
2117 class. This seemed a bit dangerous, so the code has been changed to be
2118 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2119 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2120 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2121 Perl does, and where it didn't before.
2122
2123 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2124 Windows environments %n is disabled by default.
2125
2126
2127 Version 7.4 21-Sep-07
2128 ---------------------
2129
2130 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2131 means that a class such as [\s] counted as "explicit reference to CR or
2132 LF". That isn't really right - the whole point of the change was to try to
2133 help when there was an actual mention of one of the two characters. So now
2134 the change happens only if \r or \n (or a literal CR or LF) character is
2135 encountered.
2136
2137 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2138 of both had grown to the point where there were only 3 bits left.
2139 Fortunately, there was spare space in the data structure, and so I have
2140 moved the internal flags into a new 16-bit field to free up more option
2141 bits.
2142
2143 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2144 but did not set the internal JCHANGED flag - either of these is enough to
2145 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2146 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2147 start sets both bits.
2148
2149 4. Added options (at build time, compile time, exec time) to change \R from
2150 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2151
2152 5. doc/pcresyntax.html was missing from the distribution.
2153
2154 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2155 compatibility, even though it is no longer used.
2156
2157 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2158 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2159 windows.h file is present (where different names are used). [This was
2160 reversed later after testing - see 16 below.]
2161
2162 8. Changed all #include <config.h> to #include "config.h". There were also
2163 some further <pcre.h> cases that I changed to "pcre.h".
2164
2165 9. When pcregrep was used with the --colour option, it missed the line ending
2166 sequence off the lines that it output.
2167
2168 10. It was pointed out to me that arrays of string pointers cause lots of
2169 relocations when a shared library is dynamically loaded. A technique of
2170 using a single long string with a table of offsets can drastically reduce
2171 these. I have refactored PCRE in four places to do this. The result is
2172 dramatic:
2173
2174 Originally: 290
2175 After changing UCP table: 187
2176 After changing error message table: 43
2177 After changing table of "verbs" 36
2178 After changing table of Posix names 22
2179
2180 Thanks to the folks working on Gregex for glib for this insight.
2181
2182 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2183 unicode-properties was also set.
2184
2185 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2186
2187 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2188 checked only for CRLF.
2189
2190 14. Added casts to pcretest.c to avoid compiler warnings.
2191
2192 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2193
2194 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2195 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2196 entirely. This removes changes made in 7 above.
2197
2198 17. The CMake files have been updated, and there is now more information about
2199 building with CMake in the NON-UNIX-USE document.
2200
2201
2202 Version 7.3 28-Aug-07
2203 ---------------------
2204
2205 1. In the rejigging of the build system that eventually resulted in 7.1, the
2206 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2207 brackets there is not right, since it causes compilers to look for an
2208 installed pcre.h, not the version that is in the source that is being
2209 compiled (which of course may be different). I have changed it back to:
2210
2211 #include "pcre.h"
2212
2213 I have a vague recollection that the change was concerned with compiling in
2214 different directories, but in the new build system, that is taken care of
2215 by the VPATH setting the Makefile.
2216
2217 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2218 when the subject happened to end in the byte 0x85 (e.g. if the last
2219 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2220 characters but of course it shouldn't be taken as a newline when it is part
2221 of another character. The bug was that, for an unlimited repeat of . in
2222 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2223 characters when looking for a newline.
2224
2225 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2226
2227 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2228 in debug output.
2229
2230 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2231 long printing in the pcrecpp unittest when running under MinGW.
2232
2233 6. ESC_K was left out of the EBCDIC table.
2234
2235 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2236 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2237 limit also applies to "virtual nesting" when a pattern is recursive, and in
2238 this case 1000 isn't so big. I have been able to remove this limit at the
2239 expense of backing off one optimization in certain circumstances. Normally,
2240 when pcre_exec() would call its internal match() function recursively and
2241 immediately return the result unconditionally, it uses a "tail recursion"
2242 feature to save stack. However, when a subpattern that can match an empty
2243 string has an unlimited repetition quantifier, it no longer makes this
2244 optimization. That gives it a stack frame in which to save the data for
2245 checking that an empty string has been matched. Previously this was taken
2246 from the 1000-entry workspace that had been reserved. So now there is no
2247 explicit limit, but more stack is used.
2248
2249 8. Applied Daniel's patches to solve problems with the import/export magic
2250 syntax that is required for Windows, and which was going wrong for the
2251 pcreposix and pcrecpp parts of the library. These were overlooked when this
2252 problem was solved for the main library.
2253
2254 9. There were some crude static tests to avoid integer overflow when computing
2255 the size of patterns that contain repeated groups with explicit upper
2256 limits. As the maximum quantifier is 65535, the maximum group length was
2257 set at 30,000 so that the product of these two numbers did not overflow a
2258 32-bit integer. However, it turns out that people want to use groups that
2259 are longer than 30,000 bytes (though not repeat them that many times).
2260 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2261 made it possible to implement the integer overflow checks in a much more
2262 dynamic way, which I have now done. The artificial limitation on group
2263 length has been removed - we now have only the limit on the total length of
2264 the compiled pattern, which depends on the LINK_SIZE setting.
2265
2266 10. Fixed a bug in the documentation for get/copy named substring when
2267 duplicate names are permitted. If none of the named substrings are set, the
2268 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2269 empty string.
2270
2271 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2272 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2273 because the ] is interpreted as the first data character and the
2274 terminating ] is not found. PCRE has been made compatible with Perl in this
2275 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2276 cause memory overwriting.
2277
2278 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2279 string has been matched (to stop an infinite loop). It was not recognizing
2280 a conditional subpattern that could match an empty string if that
2281 subpattern was within another subpattern. For example, it looped when
2282 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2283 condition was not nested. This bug has been fixed.
2284
2285 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2286 past the start of the subject in the presence of bytes with the top bit
2287 set, for example "\x8aBCD".
2288
2289 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2290 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2291
2292 14. Optimized (?!) to (*FAIL).
2293
2294 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2295 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2296 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2297 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2298 does: it's just the validity check that is more restrictive.
2299
2300 16. Inserted checks for integer overflows during escape sequence (backslash)
2301 processing, and also fixed erroneous offset values for syntax errors during
2302 backslash processing.
2303
2304 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2305 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2306
2307 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2308 caused an overrun.
2309
2310 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2311 something other than just ASCII characters) inside a group that had an
2312 unlimited repeat caused a loop at compile time (while checking to see
2313 whether the group could match an empty string).
2314
2315 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2316 [\P{Any}] did so. (Error in the code for printing property names.)
2317
2318 21. An orphan \E inside a character class could cause a crash.
2319
2320 22. A repeated capturing bracket such as (A)? could cause a wild memory
2321 reference during compilation.
2322
2323 23. There are several functions in pcre_compile() that scan along a compiled
2324 expression for various reasons (e.g. to see if it's fixed length for look
2325 behind). There were bugs in these functions when a repeated \p or \P was
2326 present in the pattern. These operators have additional parameters compared
2327 with \d, etc, and these were not being taken into account when moving along
2328 the compiled data. Specifically:
2329
2330 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2331 length.
2332
2333 (b) An item such as \pL+ within a repeated group could cause crashes or
2334 loops.
2335
2336 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2337 "reference to non-existent subpattern" error.
2338
2339 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2340
2341 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2342 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2343
2344 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2345
2346 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2347 character were causing crashes (broken optimization).
2348
2349 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2350 \p or \P) caused a compile-time loop.
2351
2352 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2353 break. For example, the unstudied pattern [\r\n]A does not match the string
2354 "\r\nA" because change 7.0/46 below moves the current point on by two
2355 characters after failing to match at the start. However, the pattern \nA
2356 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2357 the same is true. There doesn't seem any very clean way out of this, but
2358 what I have chosen to do makes the common cases work: PCRE now takes note
2359 of whether there can be an explicit match for \r or \n anywhere in the
2360 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2361 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2362 pattern has explicit CR or LF references.
2363
2364 29. Added (*CR) etc for changing newline setting at start of pattern.
2365
2366
2367 Version 7.2 19-Jun-07
2368 ---------------------
2369
2370 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2371 which is apparently normally available under Windows.
2372
2373 2. Re-jig the pcregrep tests with different newline settings in an attempt
2374 to make them independent of the local environment's newline setting.
2375
2376 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2377
2378 4. Some of the "internals" tests were previously cut out when the link size
2379 was not 2, because the output contained actual offsets. The recent new
2380 "Z" feature of pcretest means that these can be cut out, making the tests
2381 usable with all link sizes.
2382
2383 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2384 stack recursion. This gives a massive performance boost under BSD, but just
2385 a small improvement under Linux. However, it saves one field in the frame
2386 in all cases.
2387
2388 6. Added more features from the forthcoming Perl 5.10:
2389
2390 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2391 recursion call. It refers to the nth most recently opened parentheses.
2392
2393 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2394 to be opened parentheses.
2395
2396 (c) Conditions that refer to capturing parentheses can be specified
2397 relatively, for example, (?(-2)... or (?(+3)...
2398
2399 (d) \K resets the start of the current match so that everything before
2400 is not part of it.
2401
2402 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2403
2404 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2405 reference syntax.
2406
2407 (g) (?| introduces a group in which the numbering of parentheses in each
2408 alternative starts with the same number.
2409
2410 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2411
2412 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2413 PCRE_INFO_JCHANGED.
2414
2415 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2416 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2417 for detecting groups that can match an empty string.
2418
2419 9. A pattern with a very large number of alternatives (more than several
2420 hundred) was running out of internal workspace during the pre-compile
2421 phase, where pcre_compile() figures out how much memory will be needed. A
2422 bit of new cunning has reduced the workspace needed for groups with
2423 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2424 workspace instead of running out of the 4096 that are available.
2425
2426 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2427
2428 11. Applied patch from Google to remove an optimization that didn't quite work.
2429 The report of the bug said:
2430
2431 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2432 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2433 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2434
2435 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2436 it matched the wrong number of bytes.
2437
2438
2439 Version 7.1 24-Apr-07
2440 ---------------------
2441
2442 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2443 that is more "standard", making use of automake and other Autotools. There
2444 is some re-arrangement of the files and adjustment of comments consequent
2445 on this.
2446
2447 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2448 for recursive directory scanning broke on some systems because the files
2449 are not scanned in any specific order and on different systems the order
2450 was different. A call to "sort" has been inserted into RunGrepTest for the
2451 approprate test as a short-term fix. In the longer term there may be an
2452 alternative.
2453
2454 3. I had an email from Eric Raymond about problems translating some of PCRE's
2455 man pages to HTML (despite the fact that I distribute HTML pages, some
2456 people do their own conversions for various reasons). The problems
2457 concerned the use of low-level troff macros .br and .in. I have therefore
2458 removed all such uses from the man pages (some were redundant, some could
2459 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2460 HTML has been updated to handle .nf/.fi and to complain if it encounters
2461 .br or .in.
2462
2463 4. Updated comments in configure.ac that get placed in config.h.in and also
2464 arranged for config.h to be included in the distribution, with the name
2465 config.h.generic, for the benefit of those who have to compile without
2466 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2467
2468 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2469 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2470 makevp.bat for the current PCRE, using the additional files
2471 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2472
2473 6. A Windows user reported a minor discrepancy with test 2, which turned out
2474 to be caused by a trailing space on an input line that had got lost in his
2475 copy. The trailing space was an accident, so I've just removed it.
2476
2477 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2478 that is needed.
2479
2480 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2481 as "const" (a) because they are and (b) because it helps the PHP
2482 maintainers who have recently made a script to detect big data structures
2483 in the php code that should be moved to the .rodata section. I remembered
2484 to update Builducptable as well, so it won't revert if ucptable.h is ever
2485 re-created.
2486
2487 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2488 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2489 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2490 support is not required. This saves 1.5-2K of code, which is important in
2491 some applications.
2492
2493 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2494 so as not to refer to the tables, even though these functions will never be
2495 called when UTF-8 support is disabled. Otherwise there are problems with a
2496 shared library.
2497
2498 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2499
2500 (a) It was defining its arguments as char * instead of void *.
2501
2502 (b) It was assuming that all moves were upwards in memory; this was true
2503 a long time ago when I wrote it, but is no longer the case.
2504
2505 The emulated memove() is provided for those environments that have neither
2506 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2507 is clearly not the case, as these two bugs were recently reported.
2508
2509 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2510 and Detrail to create the HTML documentation, the .txt form of the man
2511 pages, and it removes trailing spaces from listed files. It also creates
2512 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2513 case, it wraps all the #defines with #ifndefs. This script should be run
2514 before "make dist".
2515
2516 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2517 with Unicode property support.
2518
2519 (a) For a maximizing quantifier, if the two different cases of the
2520 character were of different lengths in their UTF-8 codings (there are
2521 some cases like this - I found 11), and the matching function had to
2522 back up over a mixture of the two cases, it incorrectly assumed they
2523 were both the same length.
2524
2525 (b) When PCRE was configured to use the heap rather than the stack for
2526 recursion during matching, it was not correctly preserving the data for
2527 the other case of a UTF-8 character when checking ahead for a match
2528 while processing a minimizing repeat. If the check also involved
2529 matching a wide character, but failed, corruption could cause an
2530 erroneous result when trying to check for a repeat of the original
2531 character.
2532
2533 13. Some tidying changes to the testing mechanism:
2534
2535 (a) The RunTest script now detects the internal link size and whether there
2536 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2537 values substituted by "configure". (The RunGrepTest script already did
2538 this for UTF-8.) The configure.ac script no longer substitutes the
2539 relevant variables.
2540
2541 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2542 with length and offset values. This means that the output is different
2543 for different internal link sizes. Test 2 is skipped for link sizes
2544 other than 2 because of this, bypassing the problem. Unfortunately,
2545 there was also a test in test 3 (the locale tests) that used /B and
2546 failed for link sizes other than 2. Rather than cut the whole test out,
2547 I have added a new /Z option to pcretest that replaces the length and
2548 offset values with spaces. This is now used to make test 3 independent
2549 of link size. (Test 2 will be tidied up later.)
2550
2551 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2552 segmentation fault instead of returning the appropriate error message.
2553
2554 15. In multiline mode when the newline sequence was set to "any", the pattern
2555 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2556 This doesn't seem right; it now treats the CRLF combination as the line
2557 ending, and so does not match in that case. It's only a pattern such as ^$
2558 that would hit this one: something like ^ABC$ would have failed after \r
2559 and then tried again after \r\n.
2560
2561 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2562 in an attempt to make files that differ only in their line terminators
2563 compare equal. This works on Linux.
2564
2565 17. Under certain error circumstances pcregrep might try to free random memory
2566 as it exited. This is now fixed, thanks to valgrind.
2567
2568 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2569 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2570 was because its rules for how to advance for /g after matching an empty
2571 string at the end of a line did not allow for this case. They now check for
2572 it specially.
2573
2574 20. pcretest is supposed to handle patterns and data of any length, by
2575 extending its buffers when necessary. It was getting this wrong when the
2576 buffer for a data line had to be extended.
2577
2578 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2579 CRLF as a newline sequence.
2580
2581 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2582 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2583 I have nevertheless tidied it up.
2584
2585 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2586
2587 24. Added a man page for pcre-config.
2588
2589
2590 Version 7.0 19-Dec-06
2591 ---------------------
2592
2593 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2594 moving to gcc 4.1.1.
2595
2596 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2597 sys/time.h, which is documented as needed for this function. It doesn't
2598 seem to matter on Linux, but it showed up on some releases of OS X.
2599
2600 3. It seems that there are systems where bytes whose values are greater than
2601 127 match isprint() in the "C" locale. The "C" locale should be the
2602 default when a C program starts up. In most systems, only ASCII printing
2603 characters match isprint(). This difference caused the output from pcretest
2604 to vary, making some of the tests fail. I have changed pcretest so that:
2605
2606 (a) When it is outputting text in the compiled version of a pattern, bytes
2607 other than 32-126 are always shown as hex escapes.
2608
2609 (b) When it is outputting text that is a matched part of a subject string,
2610 it does the same, unless a different locale has been set for the match
2611 (using the /L modifier). In this case, it uses isprint() to decide.
2612
2613 4. Fixed a major bug that caused incorrect computation of the amount of memory
2614 required for a compiled pattern when options that changed within the
2615 pattern affected the logic of the preliminary scan that determines the
2616 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2617 that the computed length was too small. The symptoms of this bug were
2618 either the PCRE error "internal error: code overflow" from pcre_compile(),
2619 or a glibc crash with a message such as "pcretest: free(): invalid next
2620 size (fast)". Examples of patterns that provoked this bug (shown in
2621 pcretest format) are:
2622
2623 /(?-x: )/x
2624 /(?x)(?-x: \s*#\s*)/
2625 /((?i)[\x{c0}])/8
2626 /(?i:[\x{c0}])/8
2627
2628 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2629 is now done differently.
2630
2631 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2632 wrapper classes; (b) implement a new function in the C++ scanner that is
2633 more efficient than the old way of doing things because it avoids levels of
2634 recursion in the regex matching; (c) add a paragraph to the documentation
2635 for the FullMatch() function.
2636
2637 6. The escape sequence \n was being treated as whatever was defined as
2638 "newline". Not only was this contrary to the documentation, which states
2639 that \n is character 10 (hex 0A), but it also went horribly wrong when
2640 "newline" was defined as CRLF. This has been fixed.
2641
2642 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2643 was being set to -1 for the "end of line" case (supposedly a value that no
2644 character can have). Though this value is never used (the check for end of
2645 line is "zero bytes in current character"), it caused compiler complaints.
2646 I've changed it to 0xffffffff.
2647
2648 8. In pcre_version.c, the version string was being built by a sequence of
2649 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2650 string (as it is for production releases) called a macro with an empty
2651 argument. The C standard says the result of this is undefined. The gcc
2652 compiler treats it as an empty string (which was what was wanted) but it is
2653 reported that Visual C gives an error. The source has been hacked around to
2654 avoid this problem.
2655
2656 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2657 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2658 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2659 of them did).
2660
2661 10. Originally, pcretest opened its input and output without "b"; then I was
2662 told that "b" was needed in some environments, so it was added for release
2663 5.0 to both the input and output. (It makes no difference on Unix-like
2664 systems.) Later I was told that it is wrong for the input on Windows. I've
2665 now abstracted the modes into two macros, to make it easier to fiddle with
2666 them, and removed "b" from the input mode under Windows.
2667
2668 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2669
2670 12. Added -help and --help to pcretest as an official way of being reminded
2671 of the options.
2672
2673 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2674 and pcrecpp.cc because they annoy compilers at high warning levels.
2675
2676 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2677
2678 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2679 scripts are not C programs :-) and which was not noticed because it works
2680 on Linux.
2681
2682 16. pcretest is supposed to handle any length of pattern and data line (as one
2683 line or as a continued sequence of lines) by extending its input buffer if
2684 necessary. This feature was broken for very long pattern lines, leading to
2685 a string of junk being passed to pcre_compile() if the pattern was longer
2686 than about 50K.
2687
2688 17. I have done a major re-factoring of the way pcre_compile() computes the
2689 amount of memory needed for a compiled pattern. Previously, there was code
2690 that made a preliminary scan of the pattern in order to do this. That was
2691 OK when PCRE was new, but as the facilities have expanded, it has become
2692 harder and harder to keep it in step with the real compile phase, and there
2693 have been a number of bugs (see for example, 4 above). I have now found a
2694 cunning way of running the real compile function in a "fake" mode that
2695 enables it to compute how much memory it would need, while actually only
2696 ever using a few hundred bytes of working memory and without too many
2697 tests of the mode. This should make future maintenance and development
2698 easier. A side effect of this work is that the limit of 200 on the nesting
2699 depth of parentheses has been removed (though this was never a serious
2700 limitation, I suspect). However, there is a downside: pcre_compile() now
2701 runs more slowly than before (30% or more, depending on the pattern). I
2702 hope this isn't a big issue. There is no effect on runtime performance.
2703
2704 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2705 newline (only possible for the last line of a file) and it was a
2706 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2707
2708 19. Added additional timing features to pcretest. (1) The -tm option now times
2709 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2710 separate command line item, by a number that specifies the number of
2711 repeats to use when timing. The default is 50000; this gives better
2712 precision, but takes uncomfortably long for very large patterns.
2713
2714 20. Extended pcre_study() to be more clever in cases where a branch of a
2715 subpattern has no definite first character. For example, (a*|b*)[cd] would
2716 previously give no result from pcre_study(). Now it recognizes that the
2717 first character must be a, b, c, or d.
2718
2719 21. There was an incorrect error "recursive call could loop indefinitely" if
2720 a subpattern (or the entire pattern) that was being tested for matching an
2721 empty string contained only one non-empty item after a nested subpattern.
2722 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2723 incorrectly, because the \d was being skipped in the check.
2724
2725 22. The pcretest program now has a new pattern option /B and a command line
2726 option -b, which is equivalent to adding /B to every pattern. This causes
2727 it to show the compiled bytecode, without the additional information that
2728 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2729 is the same as /B/I).
2730
2731 23. A new optimization is now able automatically to treat some sequences such
2732 as a*b as a*+b. More specifically, if something simple (such as a character
2733 or a simple class like \d) has an unlimited quantifier, and is followed by
2734 something that cannot possibly match the quantified thing, the quantifier
2735 is automatically "possessified".
2736
2737 24. A recursive reference to a subpattern whose number was greater than 39
2738 went wrong under certain circumstances in UTF-8 mode. This bug could also
2739 have affected the operation of pcre_study().
2740
2741 25. Realized that a little bit of performance could be had by replacing
2742 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2743
2744 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2745
2746 27. Possessive quantifiers such as a++ were previously implemented by turning
2747 them into atomic groups such as ($>a+). Now they have their own opcodes,
2748 which improves performance. This includes the automatically created ones
2749 from 23 above.
2750
2751 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2752 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2753 the first matched character to be a colon. This applied both to named and
2754 numbered groups.
2755
2756 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2757
2758 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2759 building PCRE on Windows easier, so I have included it in the distribution.
2760
2761 31. There is now a check in pcretest against a ridiculously large number being
2762 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2763 loop, the loop is abandoned.
2764
2765 32. Forward references to subpatterns in conditions such as (?(2)...) where
2766 subpattern 2 is defined later cause pcre_compile() to search forwards in
2767 the pattern for the relevant set of parentheses. This search went wrong
2768 when there were unescaped parentheses in a character class, parentheses
2769 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2770
2771 33. "Subroutine" calls and backreferences were previously restricted to
2772 referencing subpatterns earlier in the regex. This restriction has now
2773 been removed.
2774
2775 34. Added a number of extra features that are going to be in Perl 5.10. On the
2776 whole, these are just syntactic alternatives for features that PCRE had
2777 previously implemented using the Python syntax or my own invention. The
2778 other formats are all retained for compatibility.
2779
2780 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2781 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2782 also .NET compatible.
2783
2784 (b) A recursion or subroutine call to a named group can now be defined as
2785 (?&name) as well as (?P>name).
2786
2787 (c) A backreference to a named group can now be defined as \k<name> or
2788 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2789 5.10, are also .NET compatible.
2790
2791 (d) A conditional reference to a named group can now use the syntax
2792 (?(<name>) or (?('name') as well as (?(name).
2793
2794 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2795 groups (named and numbered) that are never evaluated inline, but can be
2796 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2797 is always false. There may be only one alternative in such a group.
2798
2799 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2800 as the simple (?(R). The condition is true only if the most recent
2801 recursion is that of the given number or name. It does not search out
2802 through the entire recursion stack.
2803
2804 (g) The escape \gN or \g{N} has been added, where N is a positive or
2805 negative number, specifying an absolute or relative reference.
2806
2807 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2808 some "unreachable code" warnings.
2809
2810 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2811 things, this adds five new scripts.
2812
2813 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2814 There were also incompatibilities regarding the handling of \Q..\E inside
2815 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2816 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2817
2818 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2819 matches an empty string, and forcibly breaks the loop. There were bugs in
2820 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2821 against aaaa the result was just "a" rather than "aaaa", for example. Two
2822 separate and independent bugs (that affected different cases) have been
2823 fixed.
2824
2825 39. Refactored the code to abolish the use of different opcodes for small
2826 capturing bracket numbers. This is a tidy that I avoided doing when I
2827 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2828 The new approach is not only tidier, it makes it possible to reduce the
2829 memory needed to fix the previous bug (38).
2830
2831 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2832 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2833 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2834 mode.
2835
2836 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2837 report.
2838
2839 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2840 copy construction and assignment in the C++ wrapper.
2841
2842 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2843 couple of bugs that could have given wrong results in the "--newline=crlf"
2844 case.
2845
2846 44. Added a number of casts and did some reorganization of signed/unsigned int
2847 variables following suggestions from Dair Grant. Also renamed the variable
2848 "this" as "item" because it is a C++ keyword.
2849
2850 45. Arranged for dftables to add
2851
2852 #include "pcre_internal.h"
2853
2854 to pcre_chartables.c because without it, gcc 4.x may remove the array
2855 definition from the final binary if PCRE is built into a static library and
2856 dead code stripping is activated.
2857
2858 46. For an unanchored pattern, if a match attempt fails at the start of a
2859 newline sequence, and the newline setting is CRLF or ANY, and the next two
2860 characters are CRLF, advance by two characters instead of one.
2861
2862
2863 Version 6.7 04-Jul-06
2864 ---------------------
2865
2866 1. In order to handle tests when input lines are enormously long, pcretest has
2867 been re-factored so that it automatically extends its buffers when
2868 necessary. The code is crude, but this _is_ just a test program. The
2869 default size has been increased from 32K to 50K.
2870
2871 2. The code in pcre_study() was using the value of the re argument before
2872 testing it for NULL. (Of course, in any sensible call of the function, it
2873 won't be NULL.)
2874
2875 3. The memmove() emulation function in pcre_internal.h, which is used on
2876 systems that lack both memmove() and bcopy() - that is, hardly ever -
2877 was missing a "static" storage class specifier.
2878
2879 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2880 containing an extended class (one that cannot be represented by a bitmap
2881 because it contains high-valued characters or Unicode property items, e.g.
2882 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2883 pattern, but PCRE should not loop if you do not (it no longer does).
2884 [Detail: two cases were found: (a) a repeated subpattern containing an
2885 extended class; (b) a recursive reference to a subpattern that followed a
2886 previous extended class. It wasn't skipping over the extended class
2887 correctly when UTF-8 mode was not set.]
2888
2889 5. A negated single-character class was not being recognized as fixed-length
2890 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2891 compile error "lookbehind assertion is not fixed length".
2892
2893 6. The RunPerlTest auxiliary script was showing an unexpected difference
2894 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2895 write a Perl script that can interpret lines of an input file either as
2896 byte characters or as UTF-8, which is what "perltest" was being required to
2897 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2898 can't do is switch easily at run time between having the "use utf8;" pragma
2899 or not. In the end, I fudged it by using the RunPerlTest script to insert
2900 "use utf8;" explicitly for the UTF-8 tests.
2901
2902 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2903 the end of the subject string, contrary to the documentation and to what
2904 Perl does. This was true of both matching functions. Now it matches only at
2905 the start of the subject and immediately after *internal* newlines.
2906
2907 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2908 a pointer to an int instead of a pointer to an unsigned long int. This
2909 caused problems on 64-bit systems.
2910
2911 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2912 instance of the 'standard' template library not being so standard".
2913
2914 10. There was no check on the number of named subpatterns nor the maximum
2915 length of a subpattern name. The product of these values is used to compute
2916 the size of the memory block for a compiled pattern. By supplying a very
2917 long subpattern name and a large number of named subpatterns, the size
2918 computation could be caused to overflow. This is now prevented by limiting
2919 the length of names to 32 characters, and the number of named subpatterns
2920 to 10,000.
2921
2922 11. Subpatterns that are repeated with specific counts have to be replicated in
2923 the compiled pattern. The size of memory for this was computed from the
2924 length of the subpattern and the repeat count. The latter is limited to
2925 65535, but there was no limit on the former, meaning that integer overflow
2926 could in principle occur. The compiled length of a repeated subpattern is
2927 now limited to 30,000 bytes in order to prevent this.
2928
2929 12. Added the optional facility to have named substrings with the same name.
2930
2931 13. Added the ability to use a named substring as a condition, using the
2932 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2933 are numbers (not recommended). Forward references are permitted.
2934
2935 14. Added forward references in named backreferences (if you see what I mean).
2936
2937 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2938 pattern could run off the end of the subject. For example, the pattern
2939 "(?s)(.{1,5})"8 did this with the subject "ab".
2940
2941 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2942 PCRE_CASELESS was set when matching characters that were quantified with ?
2943 or *.
2944
2945 17. A character class other than a single negated character that had a minimum
2946 but no maximum quantifier - for example [ab]{6,} - was not handled
2947 correctly by pce_dfa_exec(). It would match only one character.
2948
2949 18. A valid (though odd) pattern that looked like a POSIX character
2950 class but used an invalid character after [ (for example [[,abc,]]) caused
2951 pcre_compile() to give the error "Failed: internal error: code overflow" or
2952 in some cases to crash with a glibc free() error. This could even happen if
2953 the pattern terminated after [[ but there just happened to be a sequence of
2954 letters, a binary zero, and a closing ] in the memory that followed.
2955
2956 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2957 over the years. Originally (before any Unicode support), just the bottom 8
2958 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2959 output from "man perlunicode" includes this:
2960
2961 The regular expression compiler produces polymorphic opcodes. That
2962 is, the pattern adapts to the data and automatically switches to
2963 the Unicode character scheme when presented with Unicode data--or
2964 instead uses a traditional byte scheme when presented with byte
2965 data.
2966
2967 Sadly, a wide octal escape does not cause a switch, and in a string with
2968 no other multibyte characters, these octal escapes are treated as before.
2969 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2970 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2971 Unicode string.
2972
2973 I have not perpetrated such confusion in PCRE. Up till now, it took just
2974 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2975 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2976 translate to the appropriate multibyte character.
2977
2978 29. Applied some refactoring to reduce the number of warnings from Microsoft
2979 and Borland compilers. This has included removing the fudge introduced
2980 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2981 a warning about an unused variable.
2982
2983 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2984 characters since release 4.0, because Perl (from release 5.004) does not.
2985 [Or at least, is documented not to: some releases seem to be in conflict
2986 with the documentation.] However, when a pattern was studied with
2987 pcre_study() and all its branches started with \s, PCRE still included VT
2988 as a possible starting character. Of course, this did no harm; it just
2989 caused an unnecessary match attempt.
2990
2991 22. Removed a now-redundant internal flag bit that recorded the fact that case
2992 dependency changed within the pattern. This was once needed for "required
2993 byte" processing, but is no longer used. This recovers a now-scarce options
2994 bit. Also moved the least significant internal flag bit to the most-
2995 significant bit of the word, which was not previously used (hangover from
2996 the days when it was an int rather than a uint) to free up another bit for
2997 the future.
2998
2999 23. Added support for CRLF line endings as well as CR and LF. As well as the
3000 default being selectable at build time, it can now be changed at runtime
3001 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3002 specify that it is scanning data with non-default line endings.
3003
3004 24. Changed the definition of CXXLINK to make it agree with the definition of
3005 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3006
3007 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3008 recursions. This makes a big different to stack usage for some patterns.
3009
3010 26. If a subpattern containing a named recursion or subroutine reference such
3011 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3012 the space required for the compiled pattern went wrong and gave too small a
3013 value. Depending on the environment, this could lead to "Failed: internal
3014 error: code overflow at offset 49" or "glibc detected double free or
3015 corruption" errors.
3016
3017 27. Applied patches from Google (a) to support the new newline modes and (b) to
3018 advance over multibyte UTF-8 characters in GlobalReplace.
3019
3020 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3021 difference for some implementation of PCRE in some Windows version.
3022
3023 29. Added some extra testing facilities to pcretest:
3024
3025 \q<number> in a data line sets the "match limit" value
3026 \Q<number> in a data line sets the "match recursion limt" value
3027 -S <number> sets the stack size, where <number> is in megabytes
3028
3029 The -S option isn't available for Windows.
3030
3031
3032 Version 6.6 06-Feb-06
3033 ---------------------
3034
3035 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3036 in pcreposix.h. I have copied the definition from pcre.h.
3037
3038 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3039 because pcre.h is no longer a built file.
3040
3041 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3042 not normally included in the compiled code.
3043
3044
3045 Version 6.5 01-Feb-06
3046 ---------------------
3047
3048 1. When using the partial match feature with pcre_dfa_exec(), it was not
3049 anchoring the second and subsequent partial matches at the new starting
3050 point. This could lead to incorrect results. For example, with the pattern
3051 /1234/, partially matching against "123" and then "a4" gave a match.
3052
3053 2. Changes to pcregrep:
3054
3055 (a) All non-match returns from pcre_exec() were being treated as failures
3056 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3057 error message is output. Some extra information is given for the
3058 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3059 probably the only errors that are likely to be caused by users (by
3060 specifying a regex that has nested indefinite repeats, for instance).
3061 If there are more than 20 of these errors, pcregrep is abandoned.
3062
3063 (b) A binary zero was treated as data while matching, but terminated the
3064 output line if it was written out. This has been fixed: binary zeroes
3065 are now no different to any other data bytes.
3066
3067 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3068 used to set a locale for matching. The --locale=xxxx long option has
3069 been added (no short equivalent) to specify a locale explicitly on the
3070 pcregrep command, overriding the environment variables.
3071
3072 (d) When -B was used with -n, some line numbers in the output were one less
3073 than they should have been.
3074
3075 (e) Added the -o (--only-matching) option.
3076
3077 (f) If -A or -C was used with -c (count only), some lines of context were
3078 accidentally printed for the final match.
3079
3080 (g) Added the -H (--with-filename) option.
3081
3082 (h) The combination of options -rh failed to suppress file names for files
3083 that were found from directory arguments.
3084
3085 (i) Added the -D (--devices) and -d (--directories) options.
3086
3087 (j) Added the -F (--fixed-strings) option.
3088
3089 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3090
3091 (l) Added the --colo(u)r option.
3092
3093 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3094 is not present by default.
3095
3096 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3097 items such as (?R) or (?1), when the recursion could match a number of
3098 alternatives. If it matched one of the alternatives, but subsequently,
3099 outside the recursion, there was a failure, the code tried to back up into
3100 the recursion. However, because of the way PCRE is implemented, this is not
3101 possible, and the result was an incorrect result from the match.
3102
3103 In order to prevent this happening, the specification of recursion has
3104 been changed so that all such subpatterns are automatically treated as
3105 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3106
3107 4. I had overlooked the fact that, in some locales, there are characters for
3108 which isalpha() is true but neither isupper() nor islower() are true. In
3109 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3110 and ordfeminine) are like this. This affected the treatment of \w and \W
3111 when they appeared in character classes, but not when they appeared outside
3112 a character class. The bit map for "word" characters is now created
3113 separately from the results of isalnum() instead of just taking it from the
3114 upper, lower, and digit maps. (Plus the underscore character, of course.)
3115
3116 5. The above bug also affected the handling of POSIX character classes such as
3117 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3118 permanent tables. Instead, the bit maps for such a class were previously
3119 created as the appropriate unions of the upper, lower, and digit bitmaps.
3120 Now they are created by subtraction from the [[:word:]] class, which has
3121 its own bitmap.
3122
3123 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3124 It is created by subtracting the vertical space characters (\x09, \x0a,
3125 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3126 subtraction was done in the overall bitmap for a character class, meaning
3127 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3128 be recognized. This bug has been fixed.
3129
3130 7. Patches from the folks at Google:
3131
3132 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3133 real life, but is still worth protecting against".
3134
3135 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3136 regular expressions".
3137
3138 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3139 have it.
3140
3141 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3142 "configure" and the latter not, in order to fix a problem somebody had
3143 with compiling the Arg class on HP-UX.
3144
3145 (e) Improve the error-handling of the C++ wrapper a little bit.
3146
3147 (f) New tests for checking recursion limiting.
3148
3149 8. The pcre_memmove() function, which is used only if the environment does not
3150 have a standard memmove() function (and is therefore rarely compiled),
3151 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3152 returning a result (though PCRE never actually uses the result).
3153
3154 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3155 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3156 returned instead of calling malloc() with an overflowing number that would
3157 most likely cause subsequent chaos.
3158
3159 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3160
3161 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3162 with this option is matched, the nmatch and pmatch options of regexec() are
3163 ignored.
3164
3165 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3166 provided in case anyone wants to the the POSIX interface with UTF-8
3167 strings.
3168
3169 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3170 C++ linking (needed for some HP-UX environments).
3171
3172 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3173 (unused parameter) and in the pcre_printint() function (omitted "default"
3174 switch label when the default is to do nothing).
3175
3176 15. Added some code to make it possible, when PCRE is compiled as a C++
3177 library, to replace subject pointers for pcre_exec() with a smart pointer
3178 class, thus making it possible to process discontinuous strings.
3179
3180 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3181 much the same function. They were added by different people who were trying
3182 to make PCRE easy to compile on non-Unix systems. It has been suggested
3183 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3184 for compiling on Windows systems. I have therefore replaced it with
3185 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3186 defaults to "extern" for C or "extern C" for C++, which works fine on
3187 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3188 SCOPE with something explicit in config.h. In addition:
3189
3190 (a) pcreposix.h still had just "extern" instead of either of these macros;
3191 I have replaced it with PCRE_DATA_SCOPE.
3192
3193 (b) Functions such as _pcre_xclass(), which are internal to the library,
3194 but external in the C sense, all had PCRE_EXPORT in their definitions.
3195 This is apparently wrong for the Windows case, so I have removed it.
3196 (It makes no difference on Unix-like systems.)
3197
3198 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3199 of recursive calls to match(). This is different to MATCH_LIMIT because
3200 that limits the total number of calls to match(), not all of which increase
3201 the depth of recursion. Limiting the recursion depth limits the amount of
3202 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3203 when PCRE is compiled, and changed at run time. A patch from Google adds
3204 this functionality to the C++ interface.
3205
3206 18. Changes to the handling of Unicode character properties:
3207
3208 (a) Updated the table to Unicode 4.1.0.
3209
3210 (b) Recognize characters that are not in the table as "Cn" (undefined).
3211
3212 (c) I revised the way the table is implemented to a much improved format
3213 which includes recognition of ranges. It now supports the ranges that
3214 are defined in UnicodeData.txt, and it also amalgamates other
3215 characters into ranges. This has reduced the number of entries in the
3216 table from around 16,000 to around 3,000, thus reducing its size
3217 considerably. I realized I did not need to use a tree structure after
3218 all - a binary chop search is just as efficient. Having reduced the
3219 number of entries, I extended their size from 6 bytes to 8 bytes to
3220 allow for more data.
3221
3222 (d) Added support for Unicode script names via properties such as \p{Han}.
3223
3224 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3225 matching that character.
3226
3227 20. When matching a repeated Unicode property with a minimum greater than zero,
3228 (for example \pL{2,}), PCRE could look past the end of the subject if it
3229 reached it while seeking the minimum number of characters. This could
3230 happen only if some of the characters were more than one byte long, because
3231 there is a check for at least the minimum number of bytes.
3232
3233 21. Refactored the implementation of \p and \P so as to be more general, to
3234 allow for more different types of property in future. This has changed the
3235 compiled form incompatibly. Anybody with saved compiled patterns that use
3236 \p or \P will have to recompile them.
3237
3238 22. Added "Any" and "L&" to the supported property types.
3239
3240 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3241 but give a compile time error if the value is greater than 0xff.
3242
3243 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3244 accidentally not being installed or uninstalled.
3245
3246 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3247 made were to insert the current release number. This seemed silly, because
3248 it made things harder for people building PCRE on systems that don't run
3249 "configure". I have turned pcre.h into a distributed file, no longer built
3250 by "configure", with the version identification directly included. There is
3251 no longer a pcre.h.in file.
3252
3253 However, this change necessitated a change to the pcre-config script as
3254 well. It is built from pcre-config.in, and one of the substitutions was the
3255 release number. I have updated configure.ac so that ./configure now finds
3256 the release number by grepping pcre.h.
3257
3258 26. Added the ability to run the tests under valgrind.
3259
3260
3261 Version 6.4 05-Sep-05
3262 ---------------------
3263
3264 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3265 "--" to be printed when multiple files were scanned, even when none of the
3266 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3267 consider it to be a bug, and have restored the previous behaviour.
3268
3269 2. A couple of code tidies to get rid of compiler warnings.
3270
3271 3. The pcretest program used to cheat by referring to symbols in the library
3272 whose names begin with _pcre_. These are internal symbols that are not
3273 really supposed to be visible externally, and in some environments it is
3274 possible to suppress them. The cheating is now confined to including
3275 certain files from the library's source, which is a bit cleaner.
3276
3277 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3278 file's purpose clearer.
3279
3280 5. Reorganized pcre_ucp_findchar().
3281
3282
3283 Version 6.3 15-Aug-05
3284 ---------------------
3285
3286 1. The file libpcre.pc.in did not have general read permission in the tarball.
3287
3288 2. There were some problems when building without C++ support:
3289
3290 (a) If C++ support was not built, "make install" and "make test" still
3291 tried to test it.
3292
3293 (b) There were problems when the value of CXX was explicitly set. Some
3294 changes have been made to try to fix these, and ...
3295
3296 (c) --disable-cpp can now be used to explicitly disable C++ support.
3297
3298 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3299 backslash in a target when C++ was disabled. This confuses some
3300 versions of "make", apparently. Using an intermediate variable solves
3301 this. (Same for CPP_LOBJ.)
3302
3303 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3304 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3305 necessary on certain architectures.
3306
3307 4. Added a setting of -export-symbols-regex to the link command to remove
3308 those symbols that are exported in the C sense, but actually are local
3309 within the library, and not documented. Their names all begin with
3310 "_pcre_". This is not a perfect job, because (a) we have to except some
3311 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3312 available (and never for static libraries). I have made a note to try to
3313 find a way round (a) in the future.
3314
3315
3316 Version 6.2 01-Aug-05
3317 ---------------------
3318
3319 1. There was no test for integer overflow of quantifier values. A construction
3320 such as {1111111111111111} would give undefined results. What is worse, if
3321 a minimum quantifier for a parenthesized subpattern overflowed and became
3322 negative, the calculation of the memory size went wrong. This could have
3323 led to memory overwriting.
3324
3325 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3326
3327 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3328 operating environments where this matters.
3329
3330 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3331 PCRE options from within the C++ wrapper.
3332
3333 5. Named capturing subpatterns were not being correctly counted when a pattern
3334 was compiled. This caused two problems: (a) If there were more than 100
3335 such subpatterns, the calculation of the memory needed for the whole
3336 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3337 back references of the form \12, where the number was greater than 9, were
3338 not recognized as back references, even though there were sufficient
3339 previous subpatterns.
3340
3341 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3342 versions of gcc, e.g. 2.95.4.
3343
3344
3345 Version 6.1 21-Jun-05
3346 ---------------------
3347
3348 1. There was one reference to the variable "posix" in pcretest.c that was not
3349 surrounded by "#if !defined NOPOSIX".
3350
3351 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3352 the cross-check on the old pcre_info() function, for the benefit of the
3353 cut-down version of PCRE that is currently imported into Exim.
3354
3355 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3356 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3357 patterns that start (?i) but I don't think that matters. The use of (?i) is
3358 just an example; this all applies to the other options as well.
3359
3360 4. Since libtool seems to echo the compile commands it is issuing, the output
3361 from "make" can be reduced a bit by putting "@" in front of each libtool
3362 compile command.
3363
3364 5. Patch from the folks at Google for configure.in to be a bit more thorough
3365 in checking for a suitable C++ installation before trying to compile the
3366 C++ stuff. This should fix a reported problem when a compiler was present,
3367 but no suitable headers.
3368
3369 6. The man pages all had just "PCRE" as their title. I have changed them to
3370 be the relevant file name. I have also arranged that these names are
3371 retained in the file doc/pcre.txt, which is a concatenation in text format
3372 of all the man pages except the little individual ones for each function.
3373
3374 7. The NON-UNIX-USE file had not been updated for the different set of source
3375 files that come with release 6. I also added a few comments about the C++
3376 wrapper.
3377
3378
3379 Version 6.0 07-Jun-05
3380 ---------------------
3381
3382 1. Some minor internal re-organization to help with my DFA experiments.
3383
3384 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3385 didn't matter for the library itself when fully configured, but did matter
3386 when compiling without UCP support, or within Exim, where the ucp files are
3387 not imported.
3388
3389 3. Refactoring of the library code to split up the various functions into
3390 different source modules. The addition of the new DFA matching code (see
3391 below) to a single monolithic source would have made it really too
3392 unwieldy, quite apart from causing all the code to be include in a
3393 statically linked application, when only some functions are used. This is
3394 relevant even without the DFA addition now that patterns can be compiled in
3395 one application and matched in another.
3396
3397 The downside of splitting up is that there have to be some external
3398 functions and data tables that are used internally in different modules of
3399 the library but which are not part of the API. These have all had their
3400 names changed to start with "_pcre_" so that they are unlikely to clash
3401 with other external names.
3402
3403 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3404 a different (DFA) algorithm. Although it is slower than the original
3405 function, it does have some advantages for certain types of matching
3406 problem.
3407
3408 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3409 including restarting after a partial match.
3410
3411 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3412 defined when compiling for Windows was sent to me. I have put it into the
3413 code, though I have no means of testing or verifying it.
3414
3415 7. Added the pcre_refcount() auxiliary function.
3416
3417 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3418 match before or at the first newline in the subject string. In pcretest,
3419 the /f option on a pattern can be used to set this.
3420
3421 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3422 would behave wrongly. This has been present in PCRE since release 4.0.
3423
3424 10. A number of changes to the pcregrep command:
3425
3426 (a) Refactored how -x works; insert ^(...)$ instead of setting
3427 PCRE_ANCHORED and checking the length, in preparation for adding
3428 something similar for -w.
3429
3430 (b) Added the -w (match as a word) option.
3431
3432 (c) Refactored the way lines are read and buffered so as to have more
3433 than one at a time available.
3434
3435 (d) Implemented a pcregrep test script.
3436
3437 (e) Added the -M (multiline match) option. This allows patterns to match
3438 over several lines of the subject. The buffering ensures that at least
3439 8K, or the rest of the document (whichever is the shorter) is available
3440 for matching (and similarly the previous 8K for lookbehind assertions).
3441
3442 (f) Changed the --help output so that it now says
3443
3444 -w, --word-regex(p)
3445
3446 instead of two lines, one with "regex" and the other with "regexp"
3447 because that confused at least one person since the short forms are the
3448 same. (This required a bit of code, as the output is generated
3449 automatically from a table. It wasn't just a text change.)
3450
3451 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3452 option but starts with a hyphen. Could be a pattern or a path name
3453 starting with a hyphen, for instance.
3454
3455 (h) "-" can be given as a file name to represent stdin.
3456
3457 (i) When file names are being printed, "(standard input)" is used for
3458 the standard input, for compatibility with GNU grep. Previously
3459 "<stdin>" was used.
3460
3461 (j) The option --label=xxx can be used to supply a name to be used for
3462 stdin when file names are being printed. There is no short form.
3463
3464 (k) Re-factored the options decoding logic because we are going to add
3465 two more options that take data. Such options can now be given in four
3466 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3467
3468 (l) Added the -A, -B, and -C options for requesting that lines of context
3469 around matches be printed.
3470
3471 (m) Added the -L option to print the names of files that do not contain
3472 any matching lines, that is, the complement of -l.
3473
3474 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3475 continue to scan other files.
3476
3477 (o) The -s option was incorrectly implemented. For compatibility with other
3478 greps, it now suppresses the error message for a non-existent or non-
3479 accessible file (but not the return code). There is a new option called
3480 -q that suppresses the output of matching lines, which was what -s was
3481 previously doing.
3482
3483 (p) Added --include and --exclude options to specify files for inclusion
3484 and exclusion when recursing.
3485
3486 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3487 Hopefully, it now does.
3488
3489 12. Missing cast in pcre_study().
3490
3491 13. Added an "uninstall" target to the makefile.
3492
3493 14. Replaced "extern" in the function prototypes in Makefile.in with
3494 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3495 world, but is set differently for Windows.
3496
3497 15. Added a second compiling function called pcre_compile2(). The only
3498 difference is that it has an extra argument, which is a pointer to an
3499 integer error code. When there is a compile-time failure, this is set
3500 non-zero, in addition to the error test pointer being set to point to an
3501 error message. The new argument may be NULL if no error number is required
3502 (but then you may as well call pcre_compile(), which is now just a
3503 wrapper). This facility is provided because some applications need a
3504 numeric error indication, but it has also enabled me to tidy up the way
3505 compile-time errors are handled in the POSIX wrapper.
3506
3507 16. Added VPATH=.libs to the makefile; this should help when building with one
3508 prefix path and installing with another. (Or so I'm told by someone who
3509 knows more about this stuff than I do.)
3510
3511 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3512 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3513 match everything, including newlines. This is not POSIX-compatible, but
3514 somebody wanted the feature. From pcretest it can be activated by using
3515 both the P and the s flags.
3516
3517 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3518
3519 19. libpcre.pc was being incorrectly installed as executable.
3520
3521 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3522 it now also looks for '\r' so that it will work unmodified on Windows.
3523
3524 21. Added Google's contributed C++ wrapper to the distribution.
3525
3526 22. Added some untidy missing memory free() calls in pcretest, to keep
3527 Electric Fence happy when testing.
3528
3529
3530
3531 Version 5.0 13-Sep-04
3532 ---------------------
3533
3534 1. Internal change: literal characters are no longer packed up into items
3535 containing multiple characters in a single byte-string. Each character
3536 is now matched using a separate opcode. However, there may be more than one
3537 byte in the character in UTF-8 mode.
3538
3539 2. The pcre_callout_block structure has two new fields: pattern_position and
3540 next_item_length. These contain the offset in the pattern to the next match
3541 item, and its length, respectively.
3542
3543 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3544 insertion of callouts before each pattern item. Added the /C option to
3545 pcretest to make use of this.
3546
3547 4. On the advice of a Windows user, the lines
3548
3549 #if defined(_WIN32) || defined(WIN32)
3550 _setmode( _fileno( stdout ), 0x8000 );
3551 #endif /* defined(_WIN32) || defined(WIN32) */
3552
3553 have been added to the source of pcretest. This apparently does useful
3554 magic in relation to line terminators.
3555
3556 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3557 for the benefit of those environments where the "b" makes a difference.
3558
3559 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3560 to know about it. I have put a hack into configure.in that adds in code
3561 to set GCC=yes if CC=icc. This seems to end up at a point in the
3562 generated configure script that is early enough to affect the setting of
3563 compiler options, which is what is needed, but I have no means of testing
3564 whether it really works. (The user who reported this had patched the
3565 generated configure script, which of course I cannot do.)
3566
3567 LATER: After change 22 below (new libtool files), the configure script
3568 seems to know about icc (and also ecc). Therefore, I have commented out
3569 this hack in configure.in.
3570
3571 7. Added support for pkg-config (2 patches were sent in).
3572
3573 8. Negated POSIX character classes that used a combination of internal tables
3574 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3575 [[:^ascii]]. Typically, they would match almost any characters. The other
3576 POSIX classes were not broken in this way.
3577
3578 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3579 to find the match, as PCRE was deluded into thinking that the match had to
3580 start at the start point or following a newline. The same bug applied to
3581 patterns with negative forward assertions or any backward assertions
3582 preceding ".*" at the start, unless the pattern required a fixed first
3583 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3584
3585 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3586 starting at the last subject character, bytes beyond the end of the subject
3587 string were read.
3588
3589 11. Renamed the variable "class" as "classbits" to make life easier for C++
3590 users. (Previously there was a macro definition, but it apparently wasn't
3591 enough.)
3592
3593 12. Added the new field "tables" to the extra data so that tables can be passed
3594 in at exec time, or the internal tables can be re-selected. This allows
3595 a compiled regex to be saved and re-used at a later time by a different
3596 program that might have everything at different addresses.
3597
3598 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3599 -R library as well as a -L library.
3600
3601 14. The debugging options of pcretest (-d on the command line or D on a
3602 pattern) showed incorrect output for anything following an extended class
3603 that contained multibyte characters and which was followed by a quantifier.
3604
3605 15. Added optional support for general category Unicode character properties
3606 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3607 support. It adds about 90K to the size of the library. The meanings of the
3608 inbuilt class escapes such as \d and \s have NOT been changed.
3609
3610 16. Updated pcredemo.c to include calls to free() to release the memory for the
3611 compiled pattern.
3612
3613 17. The generated file chartables.c was being created in the source directory
3614 instead of in the building directory. This caused the build to fail if the
3615 source directory was different from the building directory, and was
3616 read-only.
3617
3618 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3619 file. No doubt somebody will tell me if they don't make sense... Also added
3620 Dan Mooney's comments about building on OpenVMS.
3621
3622 19. Added support for partial matching via the PCRE_PARTIAL option for
3623 pcre_exec() and the \P data escape in pcretest.
3624
3625 20. Extended pcretest with 3 new pattern features:
3626
3627 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3628 write the compiled pattern to the file whose name is "rest-of-line".
3629 This is a straight binary dump of the data, with the saved pointer to
3630 the character tables forced to be NULL. The study data, if any, is
3631 written too. After writing, pcretest reads a new pattern.
3632
3633 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3634 compiled pattern from the given file. There must not be any
3635 occurrences of "<" in the file name (pretty unlikely); if there are,
3636 pcretest will instead treat the initial "<" as a pattern delimiter.
3637 After reading in the pattern, pcretest goes on to read data lines as
3638 usual.
3639
3640 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3641 and 16-bit fields in a compiled pattern, to simulate a pattern that
3642 was compiled on a host of opposite endianness.
3643
3644 21. The pcre-exec() function can now cope with patterns that were compiled on
3645 hosts of opposite endianness, with this restriction:
3646
3647 As for any compiled expression that is saved and used later, the tables
3648 pointer field cannot be preserved; the extra_data field in the arguments
3649 to pcre_exec() should be used to pass in a tables address if a value
3650 other than the default internal tables were used at compile time.
3651
3652 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3653 now diagnosed as an error. Previously, most of the time, a negative number
3654 would have been treated as zero, but if in addition "ovector" was passed as
3655 NULL, a crash could occur.
3656
3657 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3658 new versions from the libtool 1.5 distribution (the last one is a copy of
3659 a file called libtool.m4). This seems to have fixed the need to patch
3660 "configure" to support Darwin 1.3 (which I used to do). However, I still
3661 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3662 workstation).
3663
3664 24. Changed the PCRE licence to be the more standard "BSD" licence.
3665
3666
3667 Version 4.5 01-Dec-03
3668 ---------------------
3669
3670 1. There has been some re-arrangement of the code for the match() function so
3671 that it can be compiled in a version that does not call itself recursively.
3672 Instead, it keeps those local variables that need separate instances for
3673 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3674 needs to "recurse". Keeping track of where control must go is done by means
3675 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3676 hide most of the details from the main code, and operates only if
3677 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3678 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3679 operating.
3680
3681 To make it easier for callers to provide specially tailored get/free
3682 functions for this usage, two new functions, pcre_stack_malloc, and
3683 pcre_stack_free, are used. They are always called in strict stacking order,
3684 and the size of block requested is always the same.
3685
3686 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3687 PCRE has been compiled to use the stack or the heap for recursion. The
3688 -C option of pcretest uses this to show which version is compiled.
3689
3690 A new data escape \S, is added to pcretest; it causes the amounts of store
3691 obtained and freed by both kinds of malloc/free at match time to be added
3692 to the output.
3693
3694 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3695 what's available on my current Linux desktop machine.
3696
3697 3. When matching a UTF-8 string, the test for a valid string at the start has
3698 been extended. If start_offset is not zero, PCRE now checks that it points
3699 to a byte that is the start of a UTF-8 character. If not, it returns
3700 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3701 this is necessary because there may be backward assertions in the pattern.
3702 When matching the same subject several times, it may save resources to use
3703 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3704
3705 4. The code for checking the validity of UTF-8 strings has been tightened so
3706 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3707 containing "overlong sequences".
3708
3709 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3710 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3711 should have been "&", but it just so happened that all the cases this let
3712 through by mistake were picked up later in the function.
3713
3714 6. I had used a variable called "isblank" - this is a C99 function, causing
3715 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3716
3717 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3718 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3719 so the version is shown; (c) stop "make test" echoing "./RunTest".
3720
3721 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3722
3723 9. The prototype for memmove() for systems that don't have it was using
3724 size_t, but the inclusion of the header that defines size_t was later. I've
3725 moved the #includes for the C headers earlier to avoid this.
3726
3727 10. Added some adjustments to the code to make it easier to compiler on certain
3728 special systems:
3729
3730 (a) Some "const" qualifiers were missing.
3731 (b) Added the macro EXPORT before all exported functions; by default this
3732 is defined to be empty.
3733 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3734 that it reads its output file name as an argument instead of writing
3735 to the standard output and assuming this can be redirected.
3736
3737 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3738 class containing characters with values greater than 255, PCRE compilation
3739 went into a loop.
3740
3741 12. A recursive reference to a subpattern that was within another subpattern
3742 that had a minimum quantifier of zero caused PCRE to crash. For example,
3743 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3744 recursion. If the recursively-called subpattern itself had a zero repeat,
3745 that was OK.
3746
3747 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3748 buffer into which it was copied (for escape processing) was still set at
3749 1024, so long lines caused crashes.
3750
3751 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3752 "internal error: code overflow...". This applied to any character class
3753 that was followed by a possessive quantifier.
3754
3755 15. Modified the Makefile to add libpcre.la as a prerequisite for
3756 libpcreposix.la because I was told this is needed for a parallel build to
3757 work.
3758
3759 16. If a pattern that contained .* following optional items at the start was
3760 studied, the wrong optimizing data was generated, leading to matching
3761 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3762 matching string must start with a or b or c. The correct conclusion for
3763 this pattern is that a match can start with any character.
3764
3765
3766 Version 4.4 13-Aug-03
3767 ---------------------
3768
3769 1. In UTF-8 mode, a character class containing characters with values between
3770 127 and 255 was not handled correctly if the compiled pattern was studied.
3771 In fixing this, I have also improved the studying algorithm for such
3772 classes (slightly).
3773
3774 2. Three internal functions had redundant arguments passed to them. Removal
3775 might give a very teeny performance improvement.
3776
3777 3. Documentation bug: the value of the capture_top field in a callout is *one
3778 more than* the number of the hightest numbered captured substring.
3779
3780 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3781 in incorrectly linking with a previously installed version. They now link
3782 explicitly with libpcre.la.
3783
3784 5. configure.in no longer needs to recognize Cygwin specially.
3785
3786 6. A problem in pcre.in for Windows platforms is fixed.
3787
3788 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3789 pcretest, it used to include the size of the study block as part of its
3790 output. Unfortunately, the structure contains a field that has a different
3791 size on different hardware architectures. This meant that the tests that
3792 showed this size failed. As the block is currently always of a fixed size,
3793 this information isn't actually particularly useful in pcretest output, so
3794 I have just removed it.
3795
3796 8. Three pre-processor statements accidentally did not start in column 1.
3797 Sadly, there are *still* compilers around that complain, even though
3798 standard C has not required this for well over a decade. Sigh.
3799
3800 9. In pcretest, the code for checking callouts passed small integers in the
3801 callout_data field, which is a void * field. However, some picky compilers
3802 complained about the casts involved for this on 64-bit systems. Now
3803 pcretest passes the address of the small integer instead, which should get
3804 rid of the warnings.
3805
3806 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3807 both compile and run time, and gives an error if an invalid UTF-8 sequence
3808 is found. There is a option for disabling this check in cases where the
3809 string is known to be correct and/or the maximum performance is wanted.
3810
3811 11. In response to a bug report, I changed one line in Makefile.in from
3812
3813 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3814 to
3815 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3816
3817 to look similar to other lines, but I have no way of telling whether this
3818 is the right thing to do, as I do not use Windows. No doubt I'll get told
3819 if it's wrong...
3820
3821
3822 Version 4.3 21-May-03
3823 ---------------------
3824
3825 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3826 Makefile.
3827
3828 2. Some refactoring to improve the quality of the code:
3829
3830 (i) The utf8_table... variables are now declared "const".
3831
3832 (ii) The code for \cx, which used the "case flipping" table to upper case
3833 lower case letters, now just substracts 32. This is ASCII-specific,
3834 but the whole concept of \cx is ASCII-specific, so it seems
3835 reasonable.
3836
3837 (iii) PCRE was using its character types table to recognize decimal and
3838 hexadecimal digits in the pattern. This is silly, because it handles
3839 only 0-9, a-f, and A-F, but the character types table is locale-
3840 specific, which means strange things might happen. A private
3841 table is now used for this - though it costs 256 bytes, a table is
3842 much faster than multiple explicit tests. Of course, the standard
3843 character types table is still used for matching digits in subject
3844 strings against \d.
3845
3846 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3847 ending in _t are). So I've renamed it as ESC_tee.
3848
3849 3. The first argument for regexec() in the POSIX wrapper should have been
3850 defined as "const".
3851
3852 4. Changed pcretest to use malloc() for its buffers so that they can be
3853 Electric Fenced for debugging.
3854
3855 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3856 to read one or more bytes before the start of the subject string. Often this
3857 had no effect on PCRE's behaviour, but in some circumstances it could
3858 provoke a segmentation fault.
3859
3860 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3861 to try to read one or more bytes before the start of the subject string.
3862
3863 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3864 UTF-8 support could misbehave in various ways if the subject string
3865 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3866 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3867 back over UTF-8 characters.)
3868
3869
3870 Version 4.2 14-Apr-03
3871 ---------------------
3872
3873 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3874
3875 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3876 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3877 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3878 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3879 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3880 and BUILD_EXEEXT
3881 Note: automatic setting of the BUILD variables is not yet working
3882 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3883 compile-time but not at link-time
3884 [LINK]: use for linking executables only
3885 make different versions for Windows and non-Windows
3886 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3887 libraries
3888 [LINK_FOR_BUILD]: new variable
3889 [OBJEXT]: use throughout
3890 [EXEEXT]: use throughout
3891 <winshared>: new target
3892 <wininstall>: new target
3893 <dftables.o>: use native compiler
3894 <dftables>: use native linker
3895 <install>: handle Windows platform correctly
3896 <clean>: ditto
3897 <check>: ditto
3898 copy DLL to top builddir before testing
3899
3900 As part of these changes, -no-undefined was removed again. This was reported
3901 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3902 in any case.
3903
3904 3. Some tidies to get rid of compiler warnings:
3905
3906 . In the match_data structure, match_limit was an unsigned long int, whereas
3907 match_call_count was an int. I've made them both unsigned long ints.
3908
3909 . In pcretest the fact that a const uschar * doesn't automatically cast to
3910 a void * provoked a warning.
3911
3912 . Turning on some more compiler warnings threw up some "shadow" variables
3913 and a few more missing casts.
3914
3915 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3916 option, a class that contained a single character with a value between 128
3917 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3918
3919 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3920 option, a class that contained several characters, but with at least one
3921 whose value was between 128 and 255 caused PCRE to crash.
3922
3923
3924 Version 4.1 12-Mar-03
3925 ---------------------
3926
3927 1. Compiling with gcc -pedantic found a couple of places where casts were
3928 needed, and a string in dftables.c that was longer than standard compilers are
3929 required to support.
3930
3931 2. Compiling with Sun's compiler found a few more places where the code could
3932 be tidied up in order to avoid warnings.
3933
3934 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3935 first of these names is deprecated in the latest Autoconf in favour of the name
3936 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3937 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3938 analogy I have changed it to CFLAGS_FOR_BUILD.
3939
3940 4. Added -no-undefined to the linking command in the Makefile, because this is
3941 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3942 linking step for the pcreposix library.
3943
3944 5. PCRE was failing to diagnose the case of two named groups with the same
3945 name.
3946
3947 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3948 literal character that is needed in the subject for a match, and scans along to
3949 ensure that it is present before embarking on the full matching process. This
3950 saves time in cases of nested unlimited repeats that are never going to match.
3951 Problem: the scan can take a lot of time if the subject is very long (e.g.
3952 megabytes), thus penalizing straightforward matches. It is now done only if the
3953 amount of subject to be scanned is less than 1000 bytes.
3954
3955 7. A lesser problem with the same optimization is that it was recording the
3956 first character of an anchored pattern as "needed", thus provoking a search
3957 right along the subject, even when the first match of the pattern was going to
3958 fail. The "needed" character is now not set for anchored patterns, unless it
3959 follows something in the pattern that is of non-fixed length. Thus, it still
3960 fulfils its original purpose of finding quick non-matches in cases of nested
3961 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3962
3963
3964 Version 4.0 17-Feb-03
3965 ---------------------
3966
3967 1. If a comment in an extended regex that started immediately after a meta-item
3968 extended to the end of string, PCRE compiled incorrect data. This could lead to
3969 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3970
3971 2. Moved to autoconf 2.53 and libtool 1.4.2.
3972
3973 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3974 the special perltest8 script is no longer needed - all the tests can be run
3975 from a single perltest script.
3976
3977 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3978 by \s. It has now been removed in PCRE. This means it isn't recognized as
3979 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3980 class [:space:] *does* include VT, thereby creating a mess.
3981
3982 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3983 space and tab.
3984
3985 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3986 its new features into the main test script, reducing the number of scripts.
3987
3988 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3989 were backward compatible, and made the (?i) apply to the whole pattern, as if
3990 /i were given. Now it behaves more logically, and applies the option setting
3991 only to what follows. PCRE has been changed to follow suit. However, if it
3992 finds options settings right at the start of the pattern, it extracts them into
3993 the global options, as before. Thus, they show up in the info data.
3994
3995 8. Added support for the \Q...\E escape sequence. Characters in between are
3996 treated as literals. This is slightly different from Perl in that $ and @ are
3997 also handled as literals inside the quotes. In Perl, they will cause variable
3998 interpolation. Note the following examples:
3999
4000 Pattern PCRE matches Perl matches
4001
4002 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4003 \Qabc\$xyz\E abc\$xyz abc\$xyz
4004 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4005
4006 For compatibility with Perl, \Q...\E sequences are recognized inside character
4007 classes as well as outside them.
4008
4009 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4010 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4011 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4012 signed/unsigned warnings.
4013
4014 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4015 option for pcretest, so I've replaced it by a simple function that does just
4016 that job.
4017
4018 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4019 "pcregrep -".
4020
4021 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4022 Java package. This provides some syntactic sugar for simple cases of what my
4023 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4024 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4025 item, you can use this simplified notation. Note that only makes sense with
4026 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4027 greediness, whatever the setting of the PCRE_UNGREEDY option.
4028
4029 13. A change of greediness default within a pattern was not taking effect at
4030 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4031 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4032 was abstracted outside.
4033
4034 14. PCRE now supports the \G assertion. It is true when the current matching
4035 position is at the start point of the match. This differs from \A when the
4036 starting offset is non-zero. Used with the /g option of pcretest (or similar
4037 code), it works in the same way as it does for Perl's /g option. If all
4038 alternatives of a regex begin with \G, the expression is anchored to the start
4039 match position, and the "anchored" flag is set in the compiled expression.
4040
4041 15. Some bugs concerning the handling of certain option changes within patterns
4042 have been fixed. These applied to options other than (?ims). For example,
4043 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4044 been the other way round. Some of this was related to change 7 above.
4045
4046 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4047 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4048 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4049 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4050
4051 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4052 mode. Unlike ".", it always matches newline, whatever the setting of
4053 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4054 assertions. Perl allows it, but it doesn't (in general) work because it can't
4055 calculate the length of the lookbehind. At least, that's the case for Perl
4056 5.8.0 - I've been told they are going to document that it doesn't work in
4057 future.
4058
4059 18. Added an error diagnosis for escapes that PCRE does not support: these are
4060 \L, \l, \N, \P, \p, \U, \u, and \X.
4061
4062 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4063 reading past the end of the pattern in cases such as /[abcd/.
4064
4065 20. PCRE was getting more memory than necessary for patterns with classes that
4066 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4067
4068 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4069 compiling PCRE for use with Virtual Pascal.
4070
4071 22. Small fix to the Makefile to make it work properly if the build is done
4072 outside the source tree.
4073
4074 23. Added a new extension: a condition to go with recursion. If a conditional
4075 subpattern starts with (?(R) the "true" branch is used if recursion has
4076 happened, whereas the "false" branch is used only at the top level.
4077
4078 24. When there was a very long string of literal characters (over 255 bytes
4079 without UTF support, over 250 bytes with UTF support), the computation of how
4080 much memory was required could be incorrect, leading to segfaults or other
4081 strange effects.
4082
4083 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4084 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4085 there was a subsequent back reference to those brackets. This meant that, for
4086 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4087 possible to check for precisely this case. All we can do is abandon the
4088 optimization if .* occurs inside capturing brackets when there are any back
4089 references whatsoever. (See below for a better fix that came later.)
4090
4091 26. The handling of the optimization for finding the first character of a
4092 non-anchored pattern, and for finding a character that is required later in the
4093 match were failing in some cases. This didn't break the matching; it just
4094 failed to optimize when it could. The way this is done has been re-implemented.
4095
4096 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4097
4098 28. Added a new feature that provides some of the functionality that Perl
4099 provides with (?{...}). The facility is termed a "callout". The way it is done
4100 in PCRE is for the caller to provide an optional function, by setting
4101 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4102 global variable. By default it is unset, which disables all calling out. To get
4103 the function called, the regex must include (?C) at appropriate points. This
4104 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4105 This provides a means of identifying different callout points. When PCRE
4106 reaches such a point in the regex, if pcre_callout has been set, the external
4107 function is called. It is provided with data in a structure called
4108 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4109 matching continues; if it returns a non-zero value, the match at the current
4110 point fails. However, backtracking will occur if possible. [This was changed
4111 later and other features added - see item 49 below.]
4112
4113 29. pcretest is upgraded to test the callout functionality. It provides a
4114 callout function that displays information. By default, it shows the start of
4115 the match and the current position in the text. There are some new data escapes
4116 to vary what happens:
4117
4118 \C+ in addition, show current contents of captured substrings
4119 \C- do not supply a callout function
4120 \C!n return 1 when callout number n is reached
4121 \C!n!m return 1 when callout number n is reached for the mth time
4122
4123 30. If pcregrep was called with the -l option and just a single file name, it
4124 output "<stdin>" if a match was found, instead of the file name.
4125
4126 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4127 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4128 pcre_exec(). This saves a malloc/free per call. The default value of
4129 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4130 when configuring.
4131
4132 32. The default maximum size of a compiled pattern is 64K. There have been a
4133 few cases of people hitting this limit. The code now uses macros to handle the
4134 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4135 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4136 configuring. Tests 2 and 5 work only with 2-byte links because they output
4137 debugging information about compiled patterns.
4138
4139 33. Internal code re-arrangements:
4140
4141 (a) Moved the debugging function for printing out a compiled regex into
4142 its own source file (printint.c) and used #include to pull it into
4143 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4144 separate copies.
4145
4146 (b) Defined the list of op-code names for debugging as a macro in
4147 internal.h so that it is next to the definition of the opcodes.
4148
4149 (c) Defined a table of op-code lengths for simpler skipping along compiled
4150 code. This is again a macro in internal.h so that it is next to the
4151 definition of the opcodes.
4152
4153 34. Added support for recursive calls to individual subpatterns, along the
4154 lines of Robin Houston's patch (but implemented somewhat differently).
4155
4156 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4157 allow it to read and process whole directories in Win32. This code was
4158 contributed by Lionel Fourquaux; it has not been tested by me.
4159
4160 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4161 used to name a group. Names consist of alphanumerics and underscores, and must
4162 be unique. Back references use the syntax (?P=name) and recursive calls use
4163 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4164 numbers. The function pcre_fullinfo() can be used after compilation to extract
4165 a name/number map. There are three relevant calls:
4166
4167 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4168 PCRE_INFO_NAMECOUNT yields the number of entries
4169 PCRE_INFO_NAMETABLE yields a pointer to the map.
4170
4171 The map is a vector of fixed-size entries. The size of each entry depends on
4172 the length of the longest name used. The first two bytes of each entry are the
4173 group number, most significant byte first. There follows the corresponding
4174 name, zero terminated. The names are in alphabetical order.
4175
4176 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4177 case instead of 255. Making it the same both with and without UTF-8 support
4178 means that the same test output works with both.
4179
4180 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4181 calling malloc() with a zero argument.
4182
4183 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4184 optimization. I've improved things by keeping a bitmap of backreferences with
4185 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4186 fact referenced, the optimization can be applied. It is unlikely that a
4187 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4188 the match to follow \n) will appear inside brackets with a number greater than
4189 31, but if it does, any back reference > 31 suppresses the optimization.
4190
4191 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4192 of disabling numbered capturing parentheses. Any opening parenthesis that is
4193 not followed by ? behaves as if it were followed by ?: but named parentheses
4194 can still be used for capturing (and they will acquire numbers in the usual
4195 way).
4196
4197 41. Redesigned the return codes from the match() function into yes/no/error so
4198 that errors can be passed back from deep inside the nested calls. A malloc
4199 failure while inside a recursive subpattern call now causes the
4200 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4201
4202 42. It is now possible to set a limit on the number of times the match()
4203 function is called in a call to pcre_exec(). This facility makes it possible to
4204 limit the amount of recursion and backtracking, though not in a directly
4205 obvious way, because the match() function is used in a number of different
4206 circumstances. The count starts from zero for each position in the subject
4207 string (for non-anchored patterns). The default limit is, for compatibility, a
4208 large number, namely 10 000 000. You can change this in two ways:
4209
4210 (a) When configuring PCRE before making, you can use --with-match-limit=n
4211 to set a default value for the compiled library.
4212
4213 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4214 a different value is set. See 45 below.
4215
4216 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4217
4218 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4219 of things that can be changed at compile time. The first argument specifies
4220 what is wanted and the second points to where the information is to be placed.
4221 The current list of available information is:
4222
4223 PCRE_CONFIG_UTF8
4224
4225 The output is an integer that is set to one if UTF-8 support is available;
4226 otherwise it is set to zero.
4227
4228 PCRE_CONFIG_NEWLINE
4229
4230 The output is an integer that it set to the value of the code that is used for
4231 newline. It is either LF (10) or CR (13).
4232
4233 PCRE_CONFIG_LINK_SIZE
4234
4235 The output is an integer that contains the number of bytes used for internal
4236 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4237
4238 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4239
4240 The output is an integer that contains the threshold above which the POSIX
4241 interface uses malloc() for output vectors. See item 31 above.
4242
4243 PCRE_CONFIG_MATCH_LIMIT
4244
4245 The output is an unsigned integer that contains the default limit of the number
4246 of match() calls in a pcre_exec() execution. See 42 above.
4247
4248 44. pcretest has been upgraded by the addition of the -C option. This causes it
4249 to extract all the available output from the new pcre_config() function, and to
4250 output it. The program then exits immediately.
4251
4252 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4253 order to support additional features. One way would have been to define
4254 pcre_exec2() (for example) with extra arguments, but this would not have been
4255 extensible, and would also have required all calls to the original function to
4256 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4257 is used for passing in "extra" data from pcre_study().
4258
4259 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4260 contains the following fields:
4261
4262 flags a bitmap indicating which of the following fields are set
4263 study_data opaque data from pcre_study()
4264 match_limit a way of specifying a limit on match() calls for a specific
4265 call to pcre_exec()
4266 callout_data data for callouts (see 49 below)
4267
4268 The flag bits are also defined in pcre.h, and are
4269
4270 PCRE_EXTRA_STUDY_DATA
4271 PCRE_EXTRA_MATCH_LIMIT
4272 PCRE_EXTRA_CALLOUT_DATA
4273
4274 The pcre_study() function now returns one of these new pcre_extra blocks, with
4275 the actual study data pointed to by the study_data field, and the
4276 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4277 before. That is, this change is entirely upwards-compatible and requires no
4278 change to existing code.
4279
4280 If you want to pass in additional data to pcre_exec(), you can either place it
4281 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4282 block.
4283
4284 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4285 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4286 times with different match limits, until it finds the minimum value needed for
4287 pcre_exec() to complete. The value is then output. This can be instructive; for
4288 most simple matches the number is quite small, but for pathological cases it
4289 gets very large very quickly.
4290
4291 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4292 returns the size of the data block pointed to by the study_data field in a
4293 pcre_extra block, that is, the value that was passed as the argument to
4294 pcre_malloc() when PCRE was getting memory in which to place the information
4295 created by pcre_study(). The fourth argument should point to a size_t variable.
4296 pcretest has been extended so that this information is shown after a successful
4297 pcre_study() call when information about the compiled regex is being displayed.
4298
4299 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4300 because what follows is always an absolute path. (Later: it turns out that this
4301 is more than cosmetic for MinGW, because it doesn't like empty path
4302 components.)
4303
4304 49. Some changes have been made to the callout feature (see 28 above):
4305
4306 (i) A callout function now has three choices for what it returns:
4307
4308 0 => success, carry on matching
4309 > 0 => failure at this point, but backtrack if possible
4310 < 0 => serious error, return this value from pcre_exec()
4311
4312 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4313 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4314 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4315 use by callout functions. It will never be used by PCRE itself.
4316
4317 (ii) The pcre_extra structure (see 45 above) has a void * field called
4318 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4319 pcre_callout_block structure has a field of the same name. The contents of
4320 the field passed in the pcre_extra structure are passed to the callout
4321 function in the corresponding field in the callout block. This makes it
4322 easier to use the same callout-containing regex from multiple threads. For
4323 testing, the pcretest program has a new data escape
4324
4325 \C*n pass the number n (may be negative) as callout_data
4326
4327 If the callout function in pcretest receives a non-zero value as
4328 callout_data, it returns that value.
4329
4330 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4331 there were some redundant $(CFLAGS) in commands that are now specified as
4332 $(LINK), which already includes $(CFLAGS).
4333
4334 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4335 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4336 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4337 one-byte characters throughout. Note that case-insensitive matching applies
4338 only to characters whose values are less than 256. PCRE doesn't support the
4339 notion of cases for higher-valued characters.
4340
4341 (i) A character class whose characters are all within 0-255 is handled as
4342 a bit map, and the map is inverted for negative classes. Previously, a
4343 character > 255 always failed to match such a class; however it should
4344 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4345
4346 (ii) A negated character class with a single character < 255 is coded as
4347 "not this character" (OP_NOT). This wasn't working properly when the test
4348 character was multibyte, either singly or repeated.
4349
4350 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4351 mode, for example: \x{100}{2,3}.
4352
4353 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4354 singly or repeated) now correctly test multibyte characters. However,
4355 PCRE doesn't recognize any characters with values greater than 255 as
4356 digits, spaces, or word characters. Such characters always match \D, \S,
4357 and \W, and never match \d, \s, or \w.
4358
4359 (v) Classes may now contain characters and character ranges with values
4360 greater than 255. For example: [ab\x{100}-\x{400}].
4361
4362 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4363 PCRE in UTF-8 mode.
4364
4365 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4366 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4367 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4368 value.)
4369
4370 53. The single man page has become too large. I have therefore split it up into
4371 a number of separate man pages. These also give rise to individual HTML pages;
4372 these are now put in a separate directory, and there is an index.html page that
4373 lists them all. Some hyperlinking between the pages has been installed.
4374
4375 54. Added convenience functions for handling named capturing parentheses.
4376
4377 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4378 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4379 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4380 are faulted.
4381
4382 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4383 calling configure. These values are used when compiling the dftables.c program
4384 which is run to generate the source of the default character tables. They
4385 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4386 you will need to set these values.
4387
4388 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4389
4390
4391 Version 3.9 02-Jan-02
4392 ---------------------
4393
4394 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4395
4396 2. If --disable-static was given, the building process failed when trying to
4397 build pcretest and pcregrep. (For some reason it was using libtool to compile
4398 them, which is not right, as they aren't part of the library.)
4399
4400
4401 Version 3.8 18-Dec-01
4402 ---------------------
4403
4404 1. The experimental UTF-8 code was completely screwed up. It was packing the
4405 bytes in the wrong order. How dumb can you get?
4406
4407
4408 Version 3.7 29-Oct-01
4409 ---------------------
4410
4411 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4412 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4413 this didn't happen under Solaris 8, where I normally test things.
4414
4415 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4416 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4417 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4418 this fix an hour or so after the initial 3.7 release.)
4419
4420
4421 Version 3.6 23-Oct-01
4422 ---------------------
4423
4424 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4425 offsets passed as NULL with zero offset count.
4426
4427 2. The config.guess and config.sub files had not been updated when I moved to
4428 the latest autoconf.
4429
4430
4431 Version 3.5 15-Aug-01
4432 ---------------------
4433
4434 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4435 had been forgotten.
4436
4437 2. By using declared but undefined structures, we can avoid using "void"
4438 definitions in pcre.h while keeping the internal definitions of the structures
4439 private.
4440
4441 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4442 user point of view, this means that both static and shared libraries are built
4443 by default, but this can be individually controlled. More of the work of
4444 handling this static/shared cases is now inside libtool instead of PCRE's make
4445 file.
4446
4447 4. The pcretest utility is now installed along with pcregrep because it is
4448 useful for users (to test regexs) and by doing this, it automatically gets
4449 relinked by libtool. The documentation has been turned into a man page, so
4450 there are now .1, .txt, and .html versions in /doc.
4451
4452 5. Upgrades to pcregrep:
4453 (i) Added long-form option names like gnu grep.
4454 (ii) Added --help to list all options with an explanatory phrase.
4455 (iii) Added -r, --recursive to recurse into sub-directories.
4456 (iv) Added -f, --file to read patterns from a file.
4457
4458 6. pcre_exec() was referring to its "code" argument before testing that
4459 argument for NULL (and giving an error if it was NULL).
4460
4461 7. Upgraded Makefile.in to allow for compiling in a different directory from
4462 the source directory.
4463
4464 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4465 options bits, the pointer it was passed was to an int instead of to an unsigned
4466 long int. This mattered only on 64-bit systems.
4467
4468 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4469 generated) instead of pcre.in, which it its source. Also made the same change
4470 in several of the .c files.
4471
4472 10. A new release of gcc defines printf() as a macro, which broke pcretest
4473 because it had an ifdef in the middle of a string argument for printf(). Fixed
4474 by using separate calls to printf().
4475
4476 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4477 script, to force use of CR or LF instead of \n in the source. On non-Unix
4478 systems, the value can be set in config.h.
4479
4480 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4481 absolute limit. Changed the text of the error message to make this clear, and
4482 likewise updated the man page.
4483
4484 13. The limit of 99 on the number of capturing subpatterns has been removed.
4485 The new limit is 65535, which I hope will not be a "real" limit.
4486
4487
4488 Version 3.4 22-Aug-00
4489 ---------------------
4490
4491 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4492
4493 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4494
4495
4496 Version 3.3 01-Aug-00
4497 ---------------------
4498
4499 1. If an octal character was given, but the value was greater than \377, it
4500 was not getting masked to the least significant bits, as documented. This could
4501 lead to crashes in some systems.
4502
4503 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4504 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4505
4506 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4507 These just pass their arguments on to (pcre_free)(), but they are provided
4508 because some uses of PCRE bind it to non-C systems that can call its functions,
4509 but cannot call free() or pcre_free() directly.
4510
4511 4. Add "make test" as a synonym for "make check". Corrected some comments in
4512 the Makefile.
4513
4514 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4515 Makefile.
4516
4517 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4518 command called pgrep for grepping around the active processes.
4519
4520 7. Added the beginnings of support for UTF-8 character strings.
4521
4522 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4523 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4524 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4525 out for the ar command.)
4526
4527
4528 Version 3.2 12-May-00
4529 ---------------------
4530
4531 This is purely a bug fixing release.
4532
4533 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4534 of ZA. This was just one example of several cases that could provoke this bug,
4535 which was introduced by change 9 of version 2.00. The code for breaking
4536 infinite loops after an iteration that matches an empty string was't working
4537 correctly.
4538
4539 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4540 when matched against abbab (for example). After matching an empty string, it
4541 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4542 caused it to match further down the string than it should.
4543
4544 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4545 was there because it doesn't seem to be needed, and it causes trouble on some
4546 systems, as it is not a Standard C header. It has been removed.
4547
4548 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4549 were reported on the Macintosh. The changes were from
4550
4551 while ((c = *(++ptr)) != 0 && c != '\n');
4552 to
4553 while ((c = *(++ptr)) != 0 && c != '\n') ;
4554
4555 Totally extraordinary, but if that's what it takes...
4556
4557 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4558 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4559 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4560 assumes the way PCRE uses memmove() (always moving upwards).
4561
4562 6. PCRE is being used in one environment where strchr() is not available. There
4563 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4564 faster code anyway.
4565
4566
4567 Version 3.1 09-Feb-00
4568 ---------------------
4569
4570 The only change in this release is the fixing of some bugs in Makefile.in for
4571 the "install" target:
4572
4573 (1) It was failing to install pcreposix.h.
4574
4575 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4576
4577
4578 Version 3.0 01-Feb-00
4579 ---------------------
4580
4581 1. Add support for the /+ modifier to perltest (to output $` like it does in
4582 pcretest).
4583
4584 2. Add support for the /g modifier to perltest.
4585
4586 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4587 matches null strings.
4588
4589 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4590 pattern. Perl treats empty patterns specially - it reuses the most recent
4591 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4592 effect.
4593
4594 5. The POSIX interface was broken in that it was just handing over the POSIX
4595 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4596 required a bigger vector, with some working space on the end. This means that
4597 the POSIX wrapper now has to get and free some memory, and copy the results.
4598
4599 6. Added some simple autoconf support, placing the test data and the
4600 documentation in separate directories, re-organizing some of the
4601 information files, and making it build pcre-config (a GNU standard). Also added
4602 libtool support for building PCRE as a shared library, which is now the
4603 default.
4604
4605 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4606 09 are not valid octal constants. Single digits will be used for minor values
4607 less than 10.
4608
4609 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4610 existing programs that set these in the POSIX interface can use PCRE without
4611 modification.
4612
4613 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4614 return all that pcre_info() returns, plus additional data. The pcre_info()
4615 function is retained for compatibility, but is considered to be obsolete.
4616
4617 10. Added experimental recursion feature (?R) to handle one common case that
4618 Perl 5.6 will be able to do with (?p{...}).
4619
4620 11. Added support for POSIX character classes like [:alpha:], which Perl is
4621 adopting.
4622
4623
4624 Version 2.08 31-Aug-99
4625 ----------------------
4626
4627 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4628 trying to match at the startoffset position, but instead was moving forward to
4629 the next newline as if a previous match had failed.
4630
4631 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4632 and could get into a loop if a null string was matched other than at the start
4633 of the subject.
4634
4635 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4636 be distinguished at compile time, and for completeness also added PCRE_DATE.
4637
4638 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4639 in GnuWin32 environments.
4640
4641
4642 Version 2.07 29-Jul-99
4643 ----------------------
4644
4645 1. The documentation is now supplied in plain text form and HTML as well as in
4646 the form of man page sources.
4647
4648 2. C++ compilers don't like assigning (void *) values to other pointer types.
4649 In particular this affects malloc(). Although there is no problem in Standard
4650 C, I've put in casts to keep C++ compilers happy.
4651
4652 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4653 should be (const char *).
4654
4655 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4656 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4657 However, I haven't made this a standard facility. The documentation doesn't
4658 mention it, and the Makefile doesn't support it.
4659
4660 5. The Makefile now contains an "install" target, with editable destinations at
4661 the top of the file. The pcretest program is not installed.
4662
4663 6. pgrep -V now gives the PCRE version number and date.
4664
4665 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4666 causing the entire string to be ignored, instead of just the last character.
4667
4668 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4669 non-matching string, it can take a very, very long time, even for strings of
4670 quite modest length, because of the nested recursion. PCRE now does better in
4671 some of these cases. It does this by remembering the last required literal
4672 character in the pattern, and pre-searching the subject to ensure it is present
4673 before running the real match. In other words, it applies a heuristic to detect
4674 some types of certain failure quickly, and in the above example, if presented
4675 with a string that has no trailing " it gives "no match" very quickly.
4676
4677 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4678 other alternatives are tried instead.
4679
4680
4681 Version 2.06 09-Jun-99
4682 ----------------------
4683
4684 1. Change pcretest's output for amount of store used to show just the code
4685 space, because the remainder (the data block) varies in size between 32-bit and
4686 64-bit systems.
4687
4688 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4689 start matching at. This allows lookbehinds to work when searching for multiple
4690 occurrences in a string.
4691
4692 3. Added additional options to pcretest for testing multiple occurrences:
4693
4694 /+ outputs the rest of the string that follows a match
4695 /g loops for multiple occurrences, using the new startoffset argument
4696 /G loops for multiple occurrences by passing an incremented pointer
4697
4698 4. PCRE wasn't doing the "first character" optimization for patterns starting
4699 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4700 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4701 the letter 'x'. On long subject strings, this gives a significant speed-up.
4702
4703
4704 Version 2.05 21-Apr-99
4705 ----------------------
4706
4707 1. Changed the type of magic_number from int to long int so that it works
4708 properly on 16-bit systems.
4709
4710 2. Fixed a bug which caused patterns starting with .* not to work correctly
4711 when the subject string contained newline characters. PCRE was assuming
4712 anchoring for such patterns in all cases, which is not correct because .* will
4713 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4714 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4715 must be retried after every newline in the subject.
4716
4717
4718 Version 2.04 18-Feb-99
4719 ----------------------
4720
4721 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4722 computation of the store needed to hold the pattern was incorrect (too large).
4723 If such patterns were nested a few deep, this could multiply and become a real
4724 problem.
4725
4726 2. Added /M option to pcretest to show the memory requirement of a specific
4727 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4728
4729 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4730 compiled in such a way that the backtracking after subsequent failure was
4731 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4732 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4733
4734
4735 Version 2.03 02-Feb-99
4736 ----------------------
4737
4738 1. Fixed typo and small mistake in man page.
4739
4740 2. Added 4th condition (GPL supersedes if conflict) and created separate
4741 LICENCE file containing the conditions.
4742
4743 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4744 Perl, that is the internal \ allows the delimiter to be included in the
4745 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4746 the final delimiter, add \ to the end of the pattern (to test the error).
4747
4748 4. Added the convenience functions for extracting substrings after a successful
4749 match. Updated pcretest to make it able to test these functions.
4750
4751
4752 Version 2.02 14-Jan-99
4753 ----------------------
4754
4755 1. Initialized the working variables associated with each extraction so that
4756 their saving and restoring doesn't refer to uninitialized store.
4757
4758 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4759 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4760 fix the problem.
4761
4762 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4763 calls, and wasn't printing the correct value for compiling calls. Increased the
4764 default value of LOOPREPEAT, and the number of significant figures in the
4765 times.
4766
4767 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4768
4769 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4770 a building problem on Windows NT with a FAT file system.
4771
4772
4773 Version 2.01 21-Oct-98
4774 ----------------------
4775
4776 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4777 to character tables built by pcre_maketables() in the current locale. If NULL
4778 is passed, the default tables are used.
4779
4780
4781 Version 2.00 24-Sep-98
4782 ----------------------
4783
4784 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4785 it any more.
4786
4787 2. Allow quantification of (?>) groups, and make it work correctly.
4788
4789 3. The first character computation wasn't working for (?>) groups.
4790
4791 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4792 end of the subject) and add 5.005's \z, which really does match only at the
4793 very end of the subject.
4794
4795 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4796
4797 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4798 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4799 localized options. All options to pcre_study() were also removed.
4800
4801 7. Add other new features from 5.005:
4802
4803 $(?<= positive lookbehind
4804 $(?<! negative lookbehind
4805 (?imsx-imsx) added the unsetting capability
4806 such a setting is global if at outer level; local otherwise
4807 (?imsx-imsx:) non-capturing groups with option setting
4808 (?(cond)re|re) conditional pattern matching
4809
4810 A backreference to itself in a repeated group matches the previous
4811 captured string.
4812
4813 8. General tidying up of studying (both automatic and via "study")
4814 consequential on the addition of new assertions.
4815
4816 9. As in 5.005, unlimited repeated groups that could match an empty substring
4817 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4818 runtime if any iteration does actually match an empty substring.
4819
4820 10. Include the RunTest script in the distribution.
4821
4822 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4823 discrepancies, some of which were old and were also with respect to 5.004. They
4824 have now been fixed.
4825
4826
4827 Version 1.09 28-Apr-98
4828 ----------------------
4829
4830 1. A negated single character class followed by a quantifier with a minimum
4831 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4832 program crashes, or just wrong answers. This did not apply to negated classes
4833 containing more than one character, or to minima other than one.
4834
4835
4836 Version 1.08 27-Mar-98
4837 ----------------------
4838
4839 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4840
4841 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4842 latter must appear before anything that relies on it in the pattern.
4843
4844
4845 Version 1.07 16-Feb-98
4846 ----------------------
4847
4848 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4849 repeat of a potentially empty string).
4850
4851
4852 Version 1.06 23-Jan-98
4853 ----------------------
4854
4855 1. Added Markus Oberhumer's little patches for C++.
4856
4857 2. Literal strings longer than 255 characters were broken.
4858
4859
4860 Version 1.05 23-Dec-97
4861 ----------------------
4862
4863 1. Negated character classes containing more than one character were failing if
4864 PCRE_CASELESS was set at run time.
4865
4866
4867 Version 1.04 19-Dec-97
4868 ----------------------
4869
4870 1. Corrected the man page, where some "const" qualifiers had been omitted.
4871
4872 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4873 input syntax.
4874
4875 3. Fixed memory leak which occurred when a regex with back references was
4876 matched with an offsets vector that wasn't big enough. The temporary memory
4877 that is used in this case wasn't being freed if the match failed.
4878
4879 4. Tidied pcretest to ensure it frees memory that it gets.
4880
4881 5. Temporary memory was being obtained in the case where the passed offsets
4882 vector was exactly big enough.
4883
4884 6. Corrected definition of offsetof() from change 5 below.
4885
4886 7. I had screwed up change 6 below and broken the rules for the use of
4887 setjmp(). Now fixed.
4888
4889
4890 Version 1.03 18-Dec-97
4891 ----------------------
4892
4893 1. A erroneous regex with a missing opening parenthesis was correctly
4894 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4895 on some systems.
4896
4897 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4898 it was reported that one broken compiler failed on the former because "code" is
4899 also an independent variable.
4900
4901 3. The erroneous regex a[]b caused an array overrun reference.
4902
4903 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4904 fail on data ending with that character. (It was going on too far, and checking
4905 the next character, typically a binary zero.) This was specific to the
4906 optimized code for single-character negative classes.
4907
4908 5. Added a contributed patch from the TIN world which does the following:
4909
4910 + Add an undef for memmove, in case the the system defines a macro for it.
4911
4912 + Add a definition of offsetof(), in case there isn't one. (I don't know
4913 the reason behind this - offsetof() is part of the ANSI standard - but
4914 it does no harm).
4915
4916 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4917 most of the places where whitespace preceded '#'. I have given up and
4918 allowed the remaining 2 cases to be at the margin.
4919
4920 + Rename some variables in pcre to eliminate shadowing. This seems very
4921 pedantic, but does no harm, of course.
4922
4923 6. Moved the call to setjmp() into its own function, to get rid of warnings
4924 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4925
4926 7. Constructs such as \d{8,} were compiling into the equivalent of
4927 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4928 outcome, but in this particular case used more store than had been allocated,
4929 which caused the bug to be discovered because it threw up an internal error.
4930
4931 8. The debugging code in both pcre and pcretest for outputting the compiled
4932 form of a regex was going wrong in the case of back references followed by
4933 curly-bracketed repeats.
4934
4935
4936 Version 1.02 12-Dec-97
4937 ----------------------
4938
4939 1. Typos in pcre.3 and comments in the source fixed.
4940
4941 2. Applied a contributed patch to get rid of places where it used to remove
4942 'const' from variables, and fixed some signed/unsigned and uninitialized
4943 variable warnings.
4944
4945 3. Added the "runtest" target to Makefile.
4946
4947 4. Set default compiler flag to -O2 rather than just -O.
4948
4949
4950 Version 1.01 19-Nov-97
4951 ----------------------
4952
4953 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4954 like /([ab]*)*/, that is, for classes with more than one character in them.
4955
4956 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4957 as /((?>a*))*/ (a PCRE_EXTRA facility).
4958
4959
4960 Version 1.00 18-Nov-97
4961 ----------------------
4962
4963 1. Added compile-time macros to support systems such as SunOS4 which don't have
4964 memmove() or strerror() but have other things that can be used instead.
4965
4966 2. Arranged that "make clean" removes the executables.
4967
4968
4969 Version 0.99 27-Oct-97
4970 ----------------------
4971
4972 1. Fixed bug in code for optimizing classes with only one character. It was
4973 initializing a 32-byte map regardless, which could cause it to run off the end
4974 of the memory it had got.
4975
4976 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4977
4978
4979 Version 0.98 22-Oct-97
4980 ----------------------
4981
4982 1. Fixed bug in code for handling temporary memory usage when there are more
4983 back references than supplied space in the ovector. This could cause segfaults.
4984
4985
4986 Version 0.97 21-Oct-97
4987 ----------------------
4988
4989 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4990
4991 2. Optimized negated single characters not to use a bit map.
4992
4993 3. Brought error texts together as macro definitions; clarified some of them;
4994 fixed one that was wrong - it said "range out of order" when it meant "invalid
4995 escape sequence".
4996
4997 4. Changed some char * arguments to const char *.
4998
4999 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5000
5001 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5002 pcretest.
5003
5004
5005 Version 0.96 16-Oct-97
5006 ----------------------
5007
5008 1. Added a simple "pgrep" utility to the distribution.
5009
5010 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5011 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5012 where "ddd" means "one or more decimal digits".
5013
5014 3. Fixed serious bug. If a pattern had a back reference, but the call to
5015 pcre_exec() didn't supply a large enough ovector to record the related
5016 identifying subpattern, the match always failed. PCRE now remembers the number
5017 of the largest back reference, and gets some temporary memory in which to save
5018 the offsets during matching if necessary, in order to ensure that
5019 backreferences always work.
5020
5021 4. Increased the compatibility with Perl in a number of ways:
5022
5023 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5024 to request this handling. The option can be set at compile or exec time.
5025
5026 (b) $ matches before a terminating newline by default; an option
5027 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5028 mode). The option can be set at compile or exec time.
5029
5030 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5031 the same as Perl's. If the decimal number it represents is less than 10
5032 or there aren't that many previous left capturing parentheses, an octal
5033 escape is read. Inside a character class, it's always an octal escape,
5034 even if it is a single digit.
5035
5036 (d) An escaped but undefined alphabetic character is taken as a literal,
5037 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5038 escapes.
5039
5040 (e) {0} is now permitted. (The previous item is removed from the compiled
5041 pattern).
5042
5043 5. Changed all the names of code files so that the basic parts are no longer
5044 than 10 characters, and abolished the teeny "globals.c" file.
5045
5046 6. Changed the handling of character classes; they are now done with a 32-byte
5047 bit map always.
5048
5049 7. Added the -d and /D options to pcretest to make it possible to look at the
5050 internals of compilation without having to recompile pcre.
5051
5052
5053 Version 0.95 23-Sep-97
5054 ----------------------
5055
5056 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5057 \x20 at the start of a run of normal characters. These were being treated as
5058 real characters, instead of the source characters being re-checked.
5059
5060
5061 Version 0.94 18-Sep-97
5062 ----------------------
5063
5064 1. The functions are now thread-safe, with the caveat that the global variables
5065 containing pointers to malloc() and free() or alternative functions are the
5066 same for all threads.
5067
5068 2. Get pcre_study() to generate a bitmap of initial characters for non-
5069 anchored patterns when this is possible, and use it if passed to pcre_exec().
5070
5071
5072 Version 0.93 15-Sep-97
5073 ----------------------
5074
5075 1. /(b)|(:+)/ was computing an incorrect first character.
5076
5077 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5078 but not actually doing anything yet.
5079
5080 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5081 as Perl does (e.g. [-az] or [az-]).
5082
5083 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5084 all possible positions.
5085
5086 5. Split up into different modules to avoid including unneeded functions in a
5087 compiled binary. However, compile and exec are still in one module. The "study"
5088 function is split off.
5089
5090 6. The character tables are now in a separate module whose source is generated
5091 by an auxiliary program - but can then be edited by hand if required. There are
5092 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5093 toupper() in the code.
5094
5095 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5096 make them global. Abolish the function for setting them, as the caller can now
5097 set them directly.
5098
5099
5100 Version 0.92 11-Sep-97
5101 ----------------------
5102
5103 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5104 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5105
5106 2. Caseless matching was not working in character classes if the characters in
5107 the pattern were in upper case.
5108
5109 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5110
5111 4. Make PCRE_ANCHORED public and accept as a compile option.
5112
5113 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5114 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5115 pass them.
5116
5117 6. Give an error if bad option bits passed at compile or run time.
5118
5119 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5120 pcretest to cause it to pass that flag.
5121
5122 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5123 options, and the first character, if set.
5124
5125 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5126
5127
5128 Version 0.91 10-Sep-97
5129 ----------------------
5130
5131 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5132 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5133
5134 2. PCRE was looping on encountering an indefinitely repeated back reference to
5135 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5136 Perl does - treats the match as successful.
5137
5138 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5