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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1251 - (show annotations)
Wed Feb 20 17:42:03 2013 UTC (6 years, 9 months ago) by ph10
File size: 242593 byte(s)
Pass back the bumpalong value for partial matches.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.33 xx-xxxx-201x
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
68 Version 8.32 30-November-2012
69 -----------------------------
70
71 1. Improved JIT compiler optimizations for first character search and single
72 character iterators.
73
74 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
75 Patch by Daniel Richard G.
76
77 3. Single character iterator optimizations in the JIT compiler.
78
79 4. Improved JIT compiler optimizations for character ranges.
80
81 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
82 Reported by Giuseppe D'Angelo.
83
84 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
85 of a line, was being set incorrectly in cases where .* appeared inside
86 atomic brackets at the start of a pattern, or where there was a subsequent
87 *PRUNE or *SKIP.
88
89 7. Improved instruction cache flush for POWER/PowerPC.
90 Patch by Daniel Richard G.
91
92 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
93 grep:
94
95 (a) There is now no limit to the number of patterns to be matched.
96
97 (b) An error is given if a pattern is too long.
98
99 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
100 are now supported.
101
102 (d) --exclude-from and --include-from (multiple use) have been added.
103
104 (e) Exclusions and inclusions now apply to all files and directories, not
105 just to those obtained from scanning a directory recursively.
106
107 (f) Multiple uses of -f and --file-list are now supported.
108
109 (g) In a Windows environment, the default for -d has been changed from
110 "read" (the GNU grep default) to "skip", because otherwise the presence
111 of a directory in the file list provokes an error.
112
113 (h) The documentation has been revised and clarified in places.
114
115 9. Improve the matching speed of capturing brackets.
116
117 10. Changed the meaning of \X so that it now matches a Unicode extended
118 grapheme cluster.
119
120 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
121 out POSIX threads when JIT support is configured.
122
123 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
124
125 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
126 structure could go wrong in environments where size_t is not the same size
127 as int.
128
129 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
130 set.
131
132 15. The EBCDIC support had decayed; later updates to the code had included
133 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
134 general tidy up of EBCDIC-related issues, and the documentation was also
135 not quite right. There is now a test that can be run on ASCII systems to
136 check some of the EBCDIC-related things (but is it not a full test).
137
138 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
139 in a small tidy to the code.
140
141 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
142
143 18. If the --only-matching (-o) option in pcregrep is specified multiple
144 times, each one causes appropriate output. For example, -o1 -o2 outputs the
145 substrings matched by the 1st and 2nd capturing parentheses. A separating
146 string can be specified by --om-separator (default empty).
147
148 19. Improving the first n character searches.
149
150 20. Turn case lists for horizontal and vertical white space into macros so that
151 they are defined only once.
152
153 21. This set of changes together give more compatible Unicode case-folding
154 behaviour for characters that have more than one other case when UCP
155 support is available.
156
157 (a) The Unicode property table now has offsets into a new table of sets of
158 three or more characters that are case-equivalent. The MultiStage2.py
159 script that generates these tables (the pcre_ucd.c file) now scans
160 CaseFolding.txt instead of UnicodeData.txt for character case
161 information.
162
163 (b) The code for adding characters or ranges of characters to a character
164 class has been abstracted into a generalized function that also handles
165 case-independence. In UTF-mode with UCP support, this uses the new data
166 to handle characters with more than one other case.
167
168 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
169 whose other case is greater than 256 are now correctly matched
170 caselessly. Previously, the high codepoint matched the low one, but not
171 vice versa.
172
173 (d) The processing of \h, \H, \v, and \ in character classes now makes use
174 of the new class addition function, using character lists defined as
175 macros alongside the case definitions of 20 above.
176
177 (e) Caseless back references now work with characters that have more than
178 one other case.
179
180 (f) General caseless matching of characters with more than one other case
181 is supported.
182
183 22. Unicode character properties were updated from Unicode 6.2.0
184
185 23. Improved CMake support under Windows. Patch by Daniel Richard G.
186
187 24. Add support for 32-bit character strings, and UTF-32
188
189 25. Major JIT compiler update (code refactoring and bugfixing).
190 Experimental Sparc 32 support is added.
191
192 26. Applied a modified version of Daniel Richard G's patch to create
193 pcre.h.generic and config.h.generic by "make" instead of in the
194 PrepareRelease script.
195
196 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
197 pcre_compile.c when checking for a zero character.
198
199 28. Introducing a native interface for JIT. Through this interface, the compiled
200 machine code can be directly executed. The purpose of this interface is to
201 provide fast pattern matching, so several sanity checks are not performed.
202 However, feature tests are still performed. The new interface provides
203 1.4x speedup compared to the old one.
204
205 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
206 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
207 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
208
209 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
210 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
211 the "old" RFC 2279). Instead, it ended up passing a negative length to
212 pcre_exec().
213
214 31. Add support for GCC's visibility feature to hide internal functions.
215
216 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
217 "unknown -C option" after outputting 0 or 1.
218
219 33. There is now support for generating a code coverage report for the test
220 suite in environments where gcc is the compiler and lcov is installed. This
221 is mainly for the benefit of the developers.
222
223 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
224 unaddressable using valgrind annotations, allowing valgrind to detect
225 invalid memory accesses. This is mainly for the benefit of the developers.
226
227 25. (*UTF) can now be used to start a pattern in any of the three libraries.
228
229 26. Give configure error if --enable-cpp but no C++ compiler found.
230
231
232 Version 8.31 06-July-2012
233 -------------------------
234
235 1. Fixing a wrong JIT test case and some compiler warnings.
236
237 2. Removed a bashism from the RunTest script.
238
239 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
240 to unsigned type, result still unsigned" that was given by an MS compiler
241 on encountering the code "-sizeof(xxx)".
242
243 4. Partial matching support is added to the JIT compiler.
244
245 5. Fixed several bugs concerned with partial matching of items that consist
246 of more than one character:
247
248 (a) /^(..)\1/ did not partially match "aba" because checking references was
249 done on an "all or nothing" basis. This also applied to repeated
250 references.
251
252 (b) \R did not give a hard partial match if \r was found at the end of the
253 subject.
254
255 (c) \X did not give a hard partial match after matching one or more
256 characters at the end of the subject.
257
258 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
259 a partial match for the string "\r".
260
261 (e) When newline was set to CRLF, the metacharacter "." did not recognize
262 a partial match for a CR character at the end of the subject string.
263
264 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
265 running pcretest, the text "(JIT)" added to the output whenever JIT is
266 actually used to run the match.
267
268 7. Individual JIT compile options can be set in pcretest by following -s+[+]
269 or /S+[+] with a digit between 1 and 7.
270
271 8. OP_NOT now supports any UTF character not just single-byte ones.
272
273 9. (*MARK) control verb is now supported by the JIT compiler.
274
275 10. The command "./RunTest list" lists the available tests without actually
276 running any of them. (Because I keep forgetting what they all are.)
277
278 11. Add PCRE_INFO_MAXLOOKBEHIND.
279
280 12. Applied a (slightly modified) user-supplied patch that improves performance
281 when the heap is used for recursion (compiled with --disable-stack-for-
282 recursion). Instead of malloc and free for each heap frame each time a
283 logical recursion happens, frames are retained on a chain and re-used where
284 possible. This sometimes gives as much as 30% improvement.
285
286 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
287 call.
288
289 14. As documented, (*COMMIT) is now confined to within a positive assertion.
290
291 15. It is now possible to link pcretest with libedit as an alternative to
292 libreadline.
293
294 16. (*COMMIT) control verb is now supported by the JIT compiler.
295
296 17. The Unicode data tables have been updated to Unicode 6.1.0.
297
298 18. Added --file-list option to pcregrep.
299
300 19. Added binary file support to pcregrep, including the -a, --binary-files,
301 -I, and --text options.
302
303 20. The madvise function is renamed for posix_madvise for QNX compatibility
304 reasons. Fixed by Giuseppe D'Angelo.
305
306 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
307
308 22. Changed the option for creating symbolic links for 16-bit man pages from
309 -s to -sf so that re-installing does not cause issues.
310
311 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
312
313 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
314 matches in certain environments (the workspace was not being correctly
315 retained). Also added to pcre_dfa_exec() a simple plausibility check on
316 some of the workspace data at the beginning of a restart.
317
318 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
319 was not doing so when it should - probably a typo introduced by SVN 528
320 (change 8.10/14).
321
322 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
323 \w+ when the character tables indicated that \x{c4} was a word character.
324 There were several related cases, all because the tests for doing a table
325 lookup were testing for characters less than 127 instead of 255.
326
327 27. If a pattern contains capturing parentheses that are not used in a match,
328 their slots in the ovector are set to -1. For those that are higher than
329 any matched groups, this happens at the end of processing. In the case when
330 there were back references that the ovector was too small to contain
331 (causing temporary malloc'd memory to be used during matching), and the
332 highest capturing number was not used, memory off the end of the ovector
333 was incorrectly being set to -1. (It was using the size of the temporary
334 memory instead of the true size.)
335
336 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
337 ovector size, it uses memory at the end of the block that it has got.
338
339 29. Check for an overlong MARK name and give an error at compile time. The
340 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
341
342 30. JIT compiler update.
343
344 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
345 Rill for the patch.
346
347 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
348
349 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
350
351 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
352 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
353 was enabled.
354
355 35. Improve JIT code generation for greedy plus quantifier.
356
357 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
358 1 to "aa" instead of to an empty string. The bug affected repeated groups
359 that could potentially match an empty string.
360
361 37. Optimizing single character iterators in JIT.
362
363 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
364 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
365 codepoints that are too big for the mode are faulted, and in a UTF mode,
366 disallowed codepoints are also faulted.
367
368 39. If PCRE was compiled with UTF support, in three places in the DFA
369 matcher there was code that should only have been obeyed in UTF mode, but
370 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
371 processing when bytes with values greater than 127 were present. In 16-bit
372 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
373 both cases the values are those that cannot be the first data item in a UTF
374 character. The three items that might have provoked this were recursions,
375 possessively repeated groups, and atomic groups.
376
377 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
378 and pcregrep, because some OS require shared objects to be explicitly
379 passed to ld, causing the link step to fail if they are not.
380
381 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
382 mode, patterns that started with \h* or \R* might be incorrectly matched.
383
384
385 Version 8.30 04-February-2012
386 -----------------------------
387
388 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
389 name is defined in ctype.h.
390
391 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
392 only in quite long subpatterns.
393
394 3. Removed the function pcre_info(), which has been obsolete and deprecated
395 since it was replaced by pcre_fullinfo() in February 2000.
396
397 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
398 match a (*MARK), and the match failed at the start of the subject, a
399 reference to memory before the start of the subject could occur. This bug
400 was introduced by fix 17 of release 8.21.
401
402 5. A reference to an unset group with zero minimum repetition was giving
403 totally wrong answers (in non-JavaScript-compatibility mode). For example,
404 /(another)?(\1?)test/ matched against "hello world test". This bug was
405 introduced in release 8.13.
406
407 6. Add support for 16-bit character strings (a large amount of work involving
408 many changes and refactorings).
409
410 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
411 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
412 from a file.
413
414 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
415 rounding is not applied in this particular case).
416
417 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
418 if they appear, or are escaped, in patterns.
419
420 10. Get rid of a number of -Wunused-but-set-variable warnings.
421
422 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
423 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
424 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
425 also returns the mark "x". This bug applied to capturing parentheses,
426 non-capturing parentheses, and atomic parentheses. It also applied to some
427 assertions.
428
429 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
430 information out of configure.ac instead of relying on pcre.h.generic, which
431 is not stored in the repository.
432
433 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
434 -lreadline.
435
436 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
437
438 15. Applied Graycode's patch to put the top-level frame on the stack rather
439 than the heap when not using the stack for recursion. This gives a
440 performance improvement in many cases when recursion is not deep.
441
442 16. Experimental code added to "pcretest -C" to output the stack frame size.
443
444
445 Version 8.21 12-Dec-2011
446 ------------------------
447
448 1. Updating the JIT compiler.
449
450 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
451 are added as well.
452
453 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
454 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
455 calling _pcre_jit_exec. Some extra comments are added.
456
457 4. (*MARK) settings inside atomic groups that do not contain any capturing
458 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
459 was introduced by change 18 for 8.20.
460
461 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
462 ECMA-262 standard.
463
464 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
465 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
466 This bug was probably introduced by change 9 of 8.13.
467
468 7. While fixing 6 above, I noticed that a number of other items were being
469 incorrectly rejected as "not fixed length". This arose partly because newer
470 opcodes had not been added to the fixed-length checking code. I have (a)
471 corrected the bug and added tests for these items, and (b) arranged for an
472 error to occur if an unknown opcode is encountered while checking for fixed
473 length instead of just assuming "not fixed length". The items that were
474 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
475 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
476 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
477
478 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
479 being incorrectly compiled and would have given unpredicatble results.
480
481 9. A possessively repeated subpattern with minimum repeat count greater than
482 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
483 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
484 the first (A) could occur when it should not.
485
486 10. Add a cast and remove a redundant test from the code.
487
488 11. JIT should use pcre_malloc/pcre_free for allocation.
489
490 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
491 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
492 is anything other than /usr, -L is still shown).
493
494 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
495
496 14. Perl does not support \N without a following name in a [] class; PCRE now
497 also gives an error.
498
499 15. If a forward reference was repeated with an upper limit of around 2000,
500 it caused the error "internal error: overran compiling workspace". The
501 maximum number of forward references (including repeats) was limited by the
502 internal workspace, and dependent on the LINK_SIZE. The code has been
503 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
504 the default depends on LINK_SIZE. There is a new upper limit (for safety)
505 of around 200,000 forward references. While doing this, I also speeded up
506 the filling in of repeated forward references.
507
508 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
509 incorrectly expecting the subject to contain another "a" after the start.
510
511 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
512 in the match, the SKIP should be ignored. This was not happening; instead
513 the SKIP was being treated as NOMATCH. For patterns such as
514 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
515 tested.
516
517 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
518 now much more compatible with Perl, in particular in cases where the result
519 is a non-match for a non-anchored pattern. For example, if
520 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
521 "m", where previously it did not return a name. A side effect of this
522 change is that for partial matches, the last encountered mark name is
523 returned, as for non matches. A number of tests that were previously not
524 Perl-compatible have been moved into the Perl-compatible test files. The
525 refactoring has had the pleasing side effect of removing one argument from
526 the match() function, thus reducing its stack requirements.
527
528 19. If the /S+ option was used in pcretest to study a pattern using JIT,
529 subsequent uses of /S (without +) incorrectly behaved like /S+.
530
531 21. Retrieve executable code size support for the JIT compiler and fixing
532 some warnings.
533
534 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
535 not work when the shorter character appeared right at the end of the
536 subject string.
537
538 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
539 systems.
540
541 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
542 output it when the /M option is used in pcretest.
543
544 25. The CheckMan script was not being included in the distribution. Also, added
545 an explicit "perl" to run Perl scripts from the PrepareRelease script
546 because this is reportedly needed in Windows.
547
548 26. If study data was being save in a file and studying had not found a set of
549 "starts with" bytes for the pattern, the data written to the file (though
550 never used) was taken from uninitialized memory and so caused valgrind to
551 complain.
552
553 27. Updated RunTest.bat as provided by Sheri Pierce.
554
555 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
556
557 29. Computation of memory usage for the table of capturing group names was
558 giving an unnecessarily large value.
559
560
561 Version 8.20 21-Oct-2011
562 ------------------------
563
564 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
565 a POSIX class. After further experiments with Perl, which convinced me that
566 Perl has bugs and confusions, a closing square bracket is no longer allowed
567 in a POSIX name. This bug also affected patterns with classes that started
568 with full stops.
569
570 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
571 captured substring, but while checking the failing first alternative,
572 substring 1 is temporarily captured. If the output vector supplied to
573 pcre_exec() was not big enough for this capture, the yield of the function
574 was still zero ("insufficient space for captured substrings"). This cannot
575 be totally fixed without adding another stack variable, which seems a lot
576 of expense for a edge case. However, I have improved the situation in cases
577 such as /(a)(b)x|abc/ matched against "abc", where the return code
578 indicates that fewer than the maximum number of slots in the ovector have
579 been set.
580
581 3. Related to (2) above: when there are more back references in a pattern than
582 slots in the output vector, pcre_exec() uses temporary memory during
583 matching, and copies in the captures as far as possible afterwards. It was
584 using the entire output vector, but this conflicts with the specification
585 that only 2/3 is used for passing back captured substrings. Now it uses
586 only the first 2/3, for compatibility. This is, of course, another edge
587 case.
588
589 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
590 main code base, and can be used by building with --enable-jit. When this is
591 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
592 runtime --no-jit option is given.
593
594 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
595 ovector, the return from the function was zero, implying that there were
596 other matches that did not fit. The correct "exactly full" value is now
597 returned.
598
599 6. If a subpattern that was called recursively or as a subroutine contained
600 (*PRUNE) or any other control that caused it to give a non-standard return,
601 invalid errors such as "Error -26 (nested recursion at the same subject
602 position)" or even infinite loops could occur.
603
604 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
605 computing the minimum length on reaching *ACCEPT, and so ended up with the
606 wrong value of 1 rather than 0. Further investigation indicates that
607 computing a minimum subject length in the presence of *ACCEPT is difficult
608 (think back references, subroutine calls), and so I have changed the code
609 so that no minimum is registered for a pattern that contains *ACCEPT.
610
611 8. If (*THEN) was present in the first (true) branch of a conditional group,
612 it was not handled as intended. [But see 16 below.]
613
614 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
615 Sheri Pierce.
616
617 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
618 the first byte in a match must be "a".
619
620 11. Change 17 for 8.13 increased the recursion depth for patterns like
621 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
622 pattern contains any instances of (*THEN). If it does not, the old
623 optimizations are restored. It would be nice to do this on a per-group
624 basis, but at the moment that is not feasible.
625
626 12. In some environments, the output of pcretest -C is CRLF terminated. This
627 broke RunTest's code that checks for the link size. A single white space
628 character after the value is now allowed for.
629
630 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
631 For "fr", it uses the Windows-specific input and output files.
632
633 14. If (*THEN) appeared in a group that was called recursively or as a
634 subroutine, it did not work as intended. [But see next item.]
635
636 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
637 pattern fragments (but not containing any | characters). If A and B are
638 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
639 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
640 D. In other words, Perl considers parentheses that do not contain any |
641 characters to be part of a surrounding alternative, whereas PCRE was
642 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
643 differently. PCRE now behaves in the same way as Perl, except in the case
644 of subroutine/recursion calls such as (?1) which have in any case always
645 been different (but PCRE had them first :-).
646
647 16. Related to 15 above: Perl does not treat the | in a conditional group as
648 creating alternatives. Such a group is treated in the same way as an
649 ordinary group without any | characters when processing (*THEN). PCRE has
650 been changed to match Perl's behaviour.
651
652 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
653 RunGrepTest script failed.
654
655 18. Change 22 for version 13 caused atomic groups to use more stack. This is
656 inevitable for groups that contain captures, but it can lead to a lot of
657 stack use in large patterns. The old behaviour has been restored for atomic
658 groups that do not contain any capturing parentheses.
659
660 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
661 suppress the check for a minimum subject length at run time. (If it was
662 given to pcre_exec() or pcre_dfa_exec() it did work.)
663
664 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
665 fail to work when decoding hex characters in data strings in EBCDIC
666 environments.
667
668 21. It appears that in at least one Mac OS environment, the isxdigit() function
669 is implemented as a macro that evaluates to its argument more than once,
670 contravening the C 90 Standard (I haven't checked a later standard). There
671 was an instance in pcretest which caused it to go wrong when processing
672 \x{...} escapes in subject strings. The has been rewritten to avoid using
673 things like p++ in the argument of isxdigit().
674
675
676 Version 8.13 16-Aug-2011
677 ------------------------
678
679 1. The Unicode data tables have been updated to Unicode 6.0.0.
680
681 2. Two minor typos in pcre_internal.h have been fixed.
682
683 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
684 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
685 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
686
687 4. There were a number of related bugs in the code for matching backrefences
688 caselessly in UTF-8 mode when codes for the characters concerned were
689 different numbers of bytes. For example, U+023A and U+2C65 are an upper
690 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
691 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
692 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
693 2-byte code at the end of the subject (it thought there wasn't enough data
694 left).
695
696 5. Comprehensive information about what went wrong is now returned by
697 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
698 as the output vector has at least 2 elements. The offset of the start of
699 the failing character and a reason code are placed in the vector.
700
701 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
702 now returned is for the first byte of the failing character, instead of the
703 last byte inspected. This is an incompatible change, but I hope it is small
704 enough not to be a problem. It makes the returned offset consistent with
705 pcre_exec() and pcre_dfa_exec().
706
707 7. pcretest now gives a text phrase as well as the error number when
708 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
709 failure, the offset and reason code are output.
710
711 8. When \R was used with a maximizing quantifier it failed to skip backwards
712 over a \r\n pair if the subsequent match failed. Instead, it just skipped
713 back over a single character (\n). This seems wrong (because it treated the
714 two characters as a single entity when going forwards), conflicts with the
715 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
716 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
717 has been changed.
718
719 9. Some internal refactoring has changed the processing so that the handling
720 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
721 time (the PCRE_DOTALL option was changed this way some time ago: version
722 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
723 which was always a bit of a fudge. It also means that there is one less
724 argument for the match() function, which reduces its stack requirements
725 slightly. This change also fixes an incompatibility with Perl: the pattern
726 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
727
728 10. More internal refactoring has drastically reduced the number of recursive
729 calls to match() for possessively repeated groups such as (abc)++ when
730 using pcre_exec().
731
732 11. While implementing 10, a number of bugs in the handling of groups were
733 discovered and fixed:
734
735 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
736 (a|)*(?1) gave a compile-time internal error.
737 ((a|)+)+ did not notice that the outer group could match an empty string.
738 (^a|^)+ was not marked as anchored.
739 (.*a|.*)+ was not marked as matching at start or after a newline.
740
741 12. Yet more internal refactoring has removed another argument from the match()
742 function. Special calls to this function are now indicated by setting a
743 value in a variable in the "match data" data block.
744
745 13. Be more explicit in pcre_study() instead of relying on "default" for
746 opcodes that mean there is no starting character; this means that when new
747 ones are added and accidentally left out of pcre_study(), testing should
748 pick them up.
749
750 14. The -s option of pcretest has been documented for ages as being an old
751 synonym of -m (show memory usage). I have changed it to mean "force study
752 for every regex", that is, assume /S for every regex. This is similar to -i
753 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
754 using it. It makes it easier to run collections of tests with and without
755 study enabled, and thereby test pcre_study() more easily. All the standard
756 tests are now run with and without -s (but some patterns can be marked as
757 "never study" - see 20 below).
758
759 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
760 restoration of the capturing data to the outer values was not happening
761 correctly.
762
763 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
764 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
765 pattern had matched an empty string, and so incorrectly returned a no
766 match.
767
768 17. There was optimizing code for the last branch of non-capturing parentheses,
769 and also for the obeyed branch of a conditional subexpression, which used
770 tail recursion to cut down on stack usage. Unfortunately, now that there is
771 the possibility of (*THEN) occurring in these branches, tail recursion is
772 no longer possible because the return has to be checked for (*THEN). These
773 two optimizations have therefore been removed. [But see 8.20/11 above.]
774
775 18. If a pattern containing \R was studied, it was assumed that \R always
776 matched two bytes, thus causing the minimum subject length to be
777 incorrectly computed because \R can also match just one byte.
778
779 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
780 was incorrectly computed.
781
782 20. If /S is present twice on a test pattern in pcretest input, it now
783 *disables* studying, thereby overriding the use of -s on the command line
784 (see 14 above). This is necessary for one or two tests to keep the output
785 identical in both cases.
786
787 21. When (*ACCEPT) was used in an assertion that matched an empty string and
788 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
789
790 22. When an atomic group that contained a capturing parenthesis was
791 successfully matched, but the branch in which it appeared failed, the
792 capturing was not being forgotten if a higher numbered group was later
793 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
794 group 1 to "a", when in fact it should be unset. This applied to multi-
795 branched capturing and non-capturing groups, repeated or not, and also to
796 positive assertions (capturing in negative assertions does not happen
797 in PCRE) and also to nested atomic groups.
798
799 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
800 subject after a captured substring, to make it easier to tell which of a
801 number of identical substrings has been captured.
802
803 24. The way atomic groups are processed by pcre_exec() has been changed so that
804 if they are repeated, backtracking one repetition now resets captured
805 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
806 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
807 "aaa". Previously, it would have been "a". As part of this code
808 refactoring, the way recursive calls are handled has also been changed.
809
810 25. If an assertion condition captured any substrings, they were not passed
811 back unless some other capturing happened later. For example, if
812 (?(?=(a))a) was matched against "a", no capturing was returned.
813
814 26. When studying a pattern that contained subroutine calls or assertions,
815 the code for finding the minimum length of a possible match was handling
816 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
817 group 1 called group 2 while simultaneously a separate group 2 called group
818 1). A stack overflow occurred in this case. I have fixed this by limiting
819 the recursion depth to 10.
820
821 27. Updated RunTest.bat in the distribution to the version supplied by Tom
822 Fortmann. This supports explicit test numbers on the command line, and has
823 argument validation and error reporting.
824
825 28. An instance of \X with an unlimited repeat could fail if at any point the
826 first character it looked at was a mark character.
827
828 29. Some minor code refactoring concerning Unicode properties and scripts
829 should reduce the stack requirement of match() slightly.
830
831 30. Added the '=' option to pcretest to check the setting of unused capturing
832 slots at the end of the pattern, which are documented as being -1, but are
833 not included in the return count.
834
835 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
836 compiled something random. Now it gives a compile-time error (as does
837 Perl).
838
839 32. A *MARK encountered during the processing of a positive assertion is now
840 recorded and passed back (compatible with Perl).
841
842 33. If --only-matching or --colour was set on a pcregrep call whose pattern
843 had alternative anchored branches, the search for a second match in a line
844 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
845 matched the line "0102" twice. The same bug affected patterns that started
846 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
847 twice.
848
849 34. Previously, PCRE did not allow quantification of assertions. However, Perl
850 does, and because of capturing effects, quantifying parenthesized
851 assertions may at times be useful. Quantifiers are now allowed for
852 parenthesized assertions.
853
854 35. A minor code tidy in pcre_compile() when checking options for \R usage.
855
856 36. \g was being checked for fancy things in a character class, when it should
857 just be a literal "g".
858
859 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
860 appearance of a nested POSIX class supersedes an apparent external class.
861 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
862 unescaped square brackets may also appear as part of class names. For
863 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
864 more like Perl. (But see 8.20/1 above.)
865
866 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
867 was because it thought it was \N{name}, which is not supported).
868
869 39. Add minix to OS list not supporting the -S option in pcretest.
870
871 40. PCRE tries to detect cases of infinite recursion at compile time, but it
872 cannot analyze patterns in sufficient detail to catch mutual recursions
873 such as ((?1))((?2)). There is now a runtime test that gives an error if a
874 subgroup is called recursively as a subpattern for a second time at the
875 same position in the subject string. In previous releases this might have
876 been caught by the recursion limit, or it might have run out of stack.
877
878 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
879 happen only once. PCRE was, however incorrectly giving a compile time error
880 "recursive call could loop indefinitely" because it cannot analyze the
881 pattern in sufficient detail. The compile time test no longer happens when
882 PCRE is compiling a conditional subpattern, but actual runaway loops are
883 now caught at runtime (see 40 above).
884
885 42. It seems that Perl allows any characters other than a closing parenthesis
886 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
887 has been changed to be the same.
888
889 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
890 as not to get warnings when autogen.sh is called. Also changed
891 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
892
893 44. To help people who use pcregrep to scan files containing exceedingly long
894 lines, the following changes have been made:
895
896 (a) The default value of the buffer size parameter has been increased from
897 8K to 20K. (The actual buffer used is three times this size.)
898
899 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
900 PCRE is built.
901
902 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
903 to be set at run time.
904
905 (d) Numerical values in pcregrep options can be followed by K or M, for
906 example --buffer-size=50K.
907
908 (e) If a line being scanned overflows pcregrep's buffer, an error is now
909 given and the return code is set to 2.
910
911 45. Add a pointer to the latest mark to the callout data block.
912
913 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
914 partial match of an empty string instead of no match. This was specific to
915 the use of ".".
916
917 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
918 complete match instead of a partial match. This bug was dependent on both
919 the PCRE_UTF8 and PCRE_DOTALL options being set.
920
921 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
922 starting byte set, because \b was not being ignored.
923
924
925 Version 8.12 15-Jan-2011
926 ------------------------
927
928 1. Fixed some typos in the markup of the man pages, and wrote a script that
929 checks for such things as part of the documentation building process.
930
931 2. On a big-endian 64-bit system, pcregrep did not correctly process the
932 --match-limit and --recursion-limit options (added for 8.11). In
933 particular, this made one of the standard tests fail. (The integer value
934 went into the wrong half of a long int.)
935
936 3. If the --colour option was given to pcregrep with -v (invert match), it
937 did strange things, either producing crazy output, or crashing. It should,
938 of course, ignore a request for colour when reporting lines that do not
939 match.
940
941 4. Another pcregrep bug caused similar problems if --colour was specified with
942 -M (multiline) and the pattern match finished with a line ending.
943
944 5. In pcregrep, when a pattern that ended with a literal newline sequence was
945 matched in multiline mode, the following line was shown as part of the
946 match. This seems wrong, so I have changed it.
947
948 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
949 the check for further matches in the same line (so they could be coloured)
950 to overrun the end of the current line. If another match was found, it was
951 incorrectly shown (and then shown again when found in the next line).
952
953 7. If pcregrep was compiled under Windows, there was a reference to the
954 function pcregrep_exit() before it was defined. I am assuming this was
955 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
956 reported by a user. I've moved the definition above the reference.
957
958
959 Version 8.11 10-Dec-2010
960 ------------------------
961
962 1. (*THEN) was not working properly if there were untried alternatives prior
963 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
964 backtracked to try for "b" instead of moving to the next alternative branch
965 at the same level (in this case, to look for "c"). The Perl documentation
966 is clear that when (*THEN) is backtracked onto, it goes to the "next
967 alternative in the innermost enclosing group".
968
969 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
970 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
971 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
972 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
973 (*THEN).
974
975 3. If \s appeared in a character class, it removed the VT character from
976 the class, even if it had been included by some previous item, for example
977 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
978 of \s, but is part of the POSIX "space" class.)
979
980 4. A partial match never returns an empty string (because you can always
981 match an empty string at the end of the subject); however the checking for
982 an empty string was starting at the "start of match" point. This has been
983 changed to the "earliest inspected character" point, because the returned
984 data for a partial match starts at this character. This means that, for
985 example, /(?<=abc)def/ gives a partial match for the subject "abc"
986 (previously it gave "no match").
987
988 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
989 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
990 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
991 has an implication that the given string is incomplete (because a partial
992 match is preferred over a full match). For this reason, these items now
993 give a partial match in this situation. [Aside: previously, the one case
994 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
995 match rather than a full match, which was wrong by the old rules, but is
996 now correct.]
997
998 6. There was a bug in the handling of #-introduced comments, recognized when
999 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1000 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1001 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1002 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1003 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1004 places in pcre_compile().
1005
1006 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1007 comments when looking ahead for named forward references to subpatterns,
1008 the only newline sequence it recognized was NL. It now handles newlines
1009 according to the set newline convention.
1010
1011 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1012 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1013 cater for a lack of strerror(). These oversights have been fixed.
1014
1015 9. Added --match-limit and --recursion-limit to pcregrep.
1016
1017 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1018
1019 11. When the -o option was used, pcregrep was setting a return code of 1, even
1020 when matches were found, and --line-buffered was not being honoured.
1021
1022 12. Added an optional parentheses number to the -o and --only-matching options
1023 of pcregrep.
1024
1025 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1026 can match an empty string. The code to do it in pcretest and pcredemo
1027 needed fixing:
1028
1029 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1030 only one byte after an empty string match just before CRLF (this case
1031 just got forgotten; "any" and "anycrlf" were OK).
1032
1033 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1034 mode when an empty string match preceded an ASCII character followed by
1035 a non-ASCII character. (The code for advancing by one character rather
1036 than one byte was nonsense.)
1037
1038 (c) The pcredemo.c sample program did not have any code at all to handle
1039 the cases when CRLF is a valid newline sequence.
1040
1041 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1042 as a starting offset was within the subject string. There is now a new
1043 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1044 negative or greater than the length of the string. In order to test this,
1045 pcretest is extended to allow the setting of negative starting offsets.
1046
1047 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1048 starting offset points to the beginning of a UTF-8 character was
1049 unnecessarily clumsy. I tidied it up.
1050
1051 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1052 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1053
1054 17. Nobody had reported that the --include_dir option, which was added in
1055 release 7.7 should have been called --include-dir (hyphen, not underscore)
1056 for compatibility with GNU grep. I have changed it to --include-dir, but
1057 left --include_dir as an undocumented synonym, and the same for
1058 --exclude-dir, though that is not available in GNU grep, at least as of
1059 release 2.5.4.
1060
1061 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1062 characters from a string of bytes) have been redefined so as not to use
1063 loops, in order to improve performance in some environments. At the same
1064 time, I abstracted some of the common code into auxiliary macros to save
1065 repetition (this should not affect the compiled code).
1066
1067 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1068 compile-time error is now given if \c is not followed by an ASCII
1069 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1070 different, and any byte value is allowed.)
1071
1072 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1073 START_OPTIMIZE option, which is now allowed at compile time - but just
1074 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1075 to pcregrep and other applications that have no direct access to PCRE
1076 options. The new /Y option in pcretest sets this option when calling
1077 pcre_compile().
1078
1079 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1080 back references. Groups containing recursive back references were forced to
1081 be atomic by that change, but in the case of named groups, the amount of
1082 memory required was incorrectly computed, leading to "Failed: internal
1083 error: code overflow". This has been fixed.
1084
1085 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1086 pcretest.c, to avoid build problems in some Borland environments.
1087
1088
1089 Version 8.10 25-Jun-2010
1090 ------------------------
1091
1092 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1093 THEN.
1094
1095 2. (*ACCEPT) was not working when inside an atomic group.
1096
1097 3. Inside a character class, \B is treated as a literal by default, but
1098 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1099 causes the error). The code is unchanged, but I tidied the documentation.
1100
1101 4. Inside a character class, PCRE always treated \R and \X as literals,
1102 whereas Perl faults them if its -w option is set. I have changed PCRE so
1103 that it faults them when PCRE_EXTRA is set.
1104
1105 5. Added support for \N, which always matches any character other than
1106 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1107
1108 6. When compiling pcregrep with newer versions of gcc which may have
1109 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1110 declared with attribute warn_unused_result" were given. Just casting the
1111 result to (void) does not stop the warnings; a more elaborate fudge is
1112 needed. I've used a macro to implement this.
1113
1114 7. Minor change to pcretest.c to avoid a compiler warning.
1115
1116 8. Added four artifical Unicode properties to help with an option to make
1117 \s etc use properties (see next item). The new properties are: Xan
1118 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1119
1120 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1121 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1122 this option. Modified pcretest to add /W to test this facility. Added
1123 REG_UCP to make it available via the POSIX interface.
1124
1125 10. Added --line-buffered to pcregrep.
1126
1127 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1128 studied, and the match started with a letter with a code point greater than
1129 127 whose first byte was different to the first byte of the other case of
1130 the letter, the other case of this starting letter was not recognized
1131 (#976).
1132
1133 12. If a pattern that was studied started with a repeated Unicode property
1134 test, for example, \p{Nd}+, there was the theoretical possibility of
1135 setting up an incorrect bitmap of starting bytes, but fortunately it could
1136 not have actually happened in practice until change 8 above was made (it
1137 added property types that matched character-matching opcodes).
1138
1139 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1140 possible starting bytes for non-anchored patterns.
1141
1142 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1143 \R, and also a number of cases that involve Unicode properties, both
1144 explicit and implicit when PCRE_UCP is set.
1145
1146 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1147 input, it could crash or give wrong results if characters with values
1148 greater than 0xc0 were present in the subject string. (Detail: it assumed
1149 UTF-8 input when processing these items.)
1150
1151 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1152 size_t is 64-bit (#991).
1153
1154 17. Added a check for running out of memory when PCRE is compiled with
1155 --disable-stack-for-recursion (#990).
1156
1157 18. If the last data line in a file for pcretest does not have a newline on
1158 the end, a newline was missing in the output.
1159
1160 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1161 less than 128) in its various bitmaps. However, there is a facility for
1162 generating tables according to the current locale when PCRE is compiled. It
1163 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1164 characters, are recognized by isspace() and therefore were getting set in
1165 these tables, and indeed these tables seem to approximate to ISO 8859. This
1166 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1167 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1168 which of course cannot start UTF-8 characters. I have changed the code so
1169 that only real ASCII characters (less than 128) and the correct starting
1170 bytes for UTF-8 encodings are set for characters greater than 127 when in
1171 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1172 altogether.)
1173
1174 20. Added the /T option to pcretest so as to be able to run tests with non-
1175 standard character tables, thus making it possible to include the tests
1176 used for 19 above in the standard set of tests.
1177
1178 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1179 reference to a subpattern the other side of a comment that contains an
1180 opening parenthesis caused either an internal compiling error, or a
1181 reference to the wrong subpattern.
1182
1183
1184 Version 8.02 19-Mar-2010
1185 ------------------------
1186
1187 1. The Unicode data tables have been updated to Unicode 5.2.0.
1188
1189 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1190 configured.
1191
1192 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1193 original author of that file, following a query about its status.
1194
1195 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1196 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1197
1198 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1199 quantifier applied to a forward-referencing subroutine call, could compile
1200 incorrect code or give the error "internal error: previously-checked
1201 referenced subpattern not found".
1202
1203 6. Both MS Visual Studio and Symbian OS have problems with initializing
1204 variables to point to external functions. For these systems, therefore,
1205 pcre_malloc etc. are now initialized to local functions that call the
1206 relevant global functions.
1207
1208 7. There were two entries missing in the vectors called coptable and poptable
1209 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1210 I've fixed the data, and added a kludgy way of testing at compile time that
1211 the lengths are correct (equal to the number of opcodes).
1212
1213 8. Following on from 7, I added a similar kludge to check the length of the
1214 eint vector in pcreposix.c.
1215
1216 9. Error texts for pcre_compile() are held as one long string to avoid too
1217 much relocation at load time. To find a text, the string is searched,
1218 counting zeros. There was no check for running off the end of the string,
1219 which could happen if a new error number was added without updating the
1220 string.
1221
1222 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1223
1224 11. \K was not working if it appeared in an atomic group or in a group that
1225 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1226 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1227 the assertion is positive, but not if it is negative.
1228
1229 12. Change 11 fortuitously reduced the size of the stack frame used in the
1230 "match()" function of pcre_exec.c by one pointer. Forthcoming
1231 implementation of support for (*MARK) will need an extra pointer on the
1232 stack; I have reserved it now, so that the stack frame size does not
1233 decrease.
1234
1235 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1236 item in branch that calls a recursion is a subroutine call - as in the
1237 second branch in the above example - was incorrectly given the compile-
1238 time error "recursive call could loop indefinitely" because pcre_compile()
1239 was not correctly checking the subroutine for matching a non-empty string.
1240
1241 14. The checks for overrunning compiling workspace could trigger after an
1242 overrun had occurred. This is a "should never occur" error, but it can be
1243 triggered by pathological patterns such as hundreds of nested parentheses.
1244 The checks now trigger 100 bytes before the end of the workspace.
1245
1246 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1247
1248
1249 Version 8.01 19-Jan-2010
1250 ------------------------
1251
1252 1. If a pattern contained a conditional subpattern with only one branch (in
1253 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1254 computed the wrong minimum data length (which is of course zero for such
1255 subpatterns). This could cause incorrect "no match" results.
1256
1257 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1258 the pattern is reset in the first branch, pcre_compile() failed with
1259 "internal error: code overflow at offset...". This happened only when
1260 the reset was to the original external option setting. (An optimization
1261 abstracts leading options settings into an external setting, which was the
1262 cause of this.)
1263
1264 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1265 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1266 assertion pattern did not match (meaning that the assertion was true), it
1267 was incorrectly treated as false if the SKIP had been reached during the
1268 matching. This also applied to assertions used as conditions.
1269
1270 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1271 assertion subpattern, including such a pattern used as a condition,
1272 unpredictable results occurred, instead of the error return
1273 PCRE_ERROR_DFA_UITEM.
1274
1275 5. The C++ GlobalReplace function was not working like Perl for the special
1276 situation when an empty string is matched. It now does the fancy magic
1277 stuff that is necessary.
1278
1279 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1280 removed. (These were left over from very, very early versions of PCRE.)
1281
1282 7. Some cosmetic changes to the code to make life easier when compiling it
1283 as part of something else:
1284
1285 (a) Change DEBUG to PCRE_DEBUG.
1286
1287 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1288 called "current" as "current_branch", to prevent a collision with the
1289 Linux macro when compiled as a kernel module.
1290
1291 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1292 prevent a collision with the Linux macro when compiled as a kernel
1293 module.
1294
1295 8. In pcre_compile() there are some checks for integer overflows that used to
1296 cast potentially large values to (double). This has been changed to that
1297 when building, a check for int64_t is made, and if it is found, it is used
1298 instead, thus avoiding the use of floating point arithmetic. (There is no
1299 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1300 double.
1301
1302 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1303 2005 (difference between two addresses compared to an unsigned value).
1304
1305 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1306 custom one, because of the following reported problem in Windows:
1307
1308 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1309 under Win32.
1310 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1311 therefore missing the function definition.
1312 - The compiler thus generates a "C" signature for the test function.
1313 - The linker fails to find the "C" function.
1314 - PCRE fails to configure if asked to do so against libbz2.
1315
1316 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1317 messages were output:
1318
1319 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1320 rerunning libtoolize, to keep the correct libtool macros in-tree.
1321 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1322
1323 I have done both of these things.
1324
1325 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1326 most of the time, it *can* run out if it is given a pattern that contains a
1327 runaway infinite recursion. I updated the discussion in the pcrestack man
1328 page.
1329
1330 13. Now that we have gone to the x.xx style of version numbers, the minor
1331 version may start with zero. Using 08 or 09 is a bad idea because users
1332 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1333 interpreted as invalid octal numbers. I've updated the previous comment in
1334 configure.ac, and also added a check that gives an error if 08 or 09 are
1335 used.
1336
1337 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1338 causing partial matching to fail when the end of the subject matched \W
1339 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1340
1341 15. There were some discrepancies between the declarations in pcre_internal.h
1342 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1343 their definitions. The declarations used "const uschar *" and the
1344 definitions used USPTR. Even though USPTR is normally defined as "const
1345 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1346 reported that: "This difference in casting confuses some C++ compilers, for
1347 example, SunCC recognizes above declarations as different functions and
1348 generates broken code for hbpcre." I have changed the declarations to use
1349 USPTR.
1350
1351 16. GNU libtool is named differently on some systems. The autogen.sh script now
1352 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1353 (FreeBSD).
1354
1355 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1356 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1357 comment: "Figure out how to create a longlong from a string: strtoll and
1358 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1359 instance, but it only takes 2 args instead of 3!"
1360
1361 18. A subtle bug concerned with back references has been fixed by a change of
1362 specification, with a corresponding code fix. A pattern such as
1363 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1364 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1365 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1366 same bug. Such groups have to be quantified to be useful, or contained
1367 inside another quantified group. (If there's no repetition, the reference
1368 can never match.) The problem arises because, having left the group and
1369 moved on to the rest of the pattern, a later failure that backtracks into
1370 the group uses the captured value from the final iteration of the group
1371 rather than the correct earlier one. I have fixed this in PCRE by forcing
1372 any group that contains a reference to itself to be an atomic group; that
1373 is, there cannot be any backtracking into it once it has completed. This is
1374 similar to recursive and subroutine calls.
1375
1376
1377 Version 8.00 19-Oct-09
1378 ----------------------
1379
1380 1. The table for translating pcre_compile() error codes into POSIX error codes
1381 was out-of-date, and there was no check on the pcre_compile() error code
1382 being within the table. This could lead to an OK return being given in
1383 error.
1384
1385 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1386 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1387 in a Windows environment.
1388
1389 3. The pcregrep --count option prints the count for each file even when it is
1390 zero, as does GNU grep. However, pcregrep was also printing all files when
1391 --files-with-matches was added. Now, when both options are given, it prints
1392 counts only for those files that have at least one match. (GNU grep just
1393 prints the file name in this circumstance, but including the count seems
1394 more useful - otherwise, why use --count?) Also ensured that the
1395 combination -clh just lists non-zero counts, with no names.
1396
1397 4. The long form of the pcregrep -F option was incorrectly implemented as
1398 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1399 but it seems right to fix it, and I didn't think it was worth preserving
1400 the old behaviour.
1401
1402 5. The command line items --regex=pattern and --regexp=pattern were not
1403 recognized by pcregrep, which required --regex pattern or --regexp pattern
1404 (with a space rather than an '='). The man page documented the '=' forms,
1405 which are compatible with GNU grep; these now work.
1406
1407 6. No libpcreposix.pc file was created for pkg-config; there was just
1408 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1409
1410 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1411 when UCP support is not needed, by modifying the Python script that
1412 generates it from Unicode data files. This should not matter if the module
1413 is correctly used as a library, but I received one complaint about 50K of
1414 unwanted data. My guess is that the person linked everything into his
1415 program rather than using a library. Anyway, it does no harm.
1416
1417 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1418 was a minimum greater than 1 for a wide character in a possessive
1419 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1420 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1421 character. Chaos in the form of incorrect output or a compiling loop could
1422 result.
1423
1424 9. The restrictions on what a pattern can contain when partial matching is
1425 requested for pcre_exec() have been removed. All patterns can now be
1426 partially matched by this function. In addition, if there are at least two
1427 slots in the offset vector, the offset of the earliest inspected character
1428 for the match and the offset of the end of the subject are set in them when
1429 PCRE_ERROR_PARTIAL is returned.
1430
1431 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1432 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1433 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1434 and may be more useful for multi-segment matching.
1435
1436 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1437 used to be given if ever the end of the subject was reached; now it is
1438 given only if matching could not proceed because another character was
1439 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1440 string "Z", which now yields "no match" instead of "partial match". In the
1441 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1442 final character ended with (*FAIL).
1443
1444 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1445 if the pattern had a "must contain" character that was already found in the
1446 earlier partial match, unless partial matching was again requested. For
1447 example, with the pattern /dog.(body)?/, the "must contain" character is
1448 "g". If the first part-match was for the string "dog", restarting with
1449 "sbody" failed. This bug has been fixed.
1450
1451 13. The string returned by pcre_dfa_exec() after a partial match has been
1452 changed so that it starts at the first inspected character rather than the
1453 first character of the match. This makes a difference only if the pattern
1454 starts with a lookbehind assertion or \b or \B (\K is not supported by
1455 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1456 matching functions compatible, and I think it's the right thing to do.
1457
1458 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1459 so that the demonstration program is easily available in environments where
1460 PCRE has not been installed from source.
1461
1462 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1463 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1464 library.
1465
1466 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1467 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1468 is not the first non-POSIX option to be added. Clearly some people find
1469 these options useful.
1470
1471 17. If a caller to the POSIX matching function regexec() passes a non-zero
1472 value for nmatch with a NULL value for pmatch, the value of
1473 nmatch is forced to zero.
1474
1475 18. RunGrepTest did not have a test for the availability of the -u option of
1476 the diff command, as RunTest does. It now checks in the same way as
1477 RunTest, and also checks for the -b option.
1478
1479 19. If an odd number of negated classes containing just a single character
1480 interposed, within parentheses, between a forward reference to a named
1481 subpattern and the definition of the subpattern, compilation crashed with
1482 an internal error, complaining that it could not find the referenced
1483 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1484 [The bug was that it was starting one character too far in when skipping
1485 over the character class, thus treating the ] as data rather than
1486 terminating the class. This meant it could skip too much.]
1487
1488 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1489 /g option in pcretest when the pattern contains \K, which makes it possible
1490 to have an empty string match not at the start, even when the pattern is
1491 anchored. Updated pcretest and pcredemo to use this option.
1492
1493 21. If the maximum number of capturing subpatterns in a recursion was greater
1494 than the maximum at the outer level, the higher number was returned, but
1495 with unset values at the outer level. The correct (outer level) value is
1496 now given.
1497
1498 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1499 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1500 make it do so. The string so far is captured, making this feature
1501 compatible with Perl.
1502
1503 23. The tests have been re-organized, adding tests 11 and 12, to make it
1504 possible to check the Perl 5.10 features against Perl 5.10.
1505
1506 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1507 pattern matches a fixed length string. PCRE did not allow this; now it
1508 does. Neither allows recursion.
1509
1510 25. I finally figured out how to implement a request to provide the minimum
1511 length of subject string that was needed in order to match a given pattern.
1512 (It was back references and recursion that I had previously got hung up
1513 on.) This code has now been added to pcre_study(); it finds a lower bound
1514 to the length of subject needed. It is not necessarily the greatest lower
1515 bound, but using it to avoid searching strings that are too short does give
1516 some useful speed-ups. The value is available to calling programs via
1517 pcre_fullinfo().
1518
1519 26. While implementing 25, I discovered to my embarrassment that pcretest had
1520 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1521 study optimizations had never been tested with that matching function.
1522 Oops. What is worse, even when it was passed study data, there was a bug in
1523 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1524 were also very few tests of studied patterns with pcre_dfa_exec().
1525
1526 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1527 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1528 on the other side of the coin, they are no longer allowed to have different
1529 names, because these cannot be distinguished in PCRE, and this has caused
1530 confusion. (This is a difference from Perl.)
1531
1532 28. When duplicate subpattern names are present (necessarily with different
1533 numbers, as required by 27 above), and a test is made by name in a
1534 conditional pattern, either for a subpattern having been matched, or for
1535 recursion in such a pattern, all the associated numbered subpatterns are
1536 tested, and the overall condition is true if the condition is true for any
1537 one of them. This is the way Perl works, and is also more like the way
1538 testing by number works.
1539
1540
1541 Version 7.9 11-Apr-09
1542 ---------------------
1543
1544 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1545 (pcretest), all targets were linked against these libraries. This included
1546 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1547 libraries. This caused unwanted dependencies to be created. This problem
1548 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1549 pcretest is linked with readline.
1550
1551 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1552 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1553 moved outside it again, because FALSE and TRUE are already defined in AIX,
1554 but BOOL is not.
1555
1556 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1557 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1558
1559 4. The pcregrep documentation said spaces were inserted as well as colons (or
1560 hyphens) following file names and line numbers when outputting matching
1561 lines. This is not true; no spaces are inserted. I have also clarified the
1562 wording for the --colour (or --color) option.
1563
1564 5. In pcregrep, when --colour was used with -o, the list of matching strings
1565 was not coloured; this is different to GNU grep, so I have changed it to be
1566 the same.
1567
1568 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1569 each matching line was coloured. Now it goes on to look for further matches
1570 of any of the test patterns, which is the same behaviour as GNU grep.
1571
1572 7. A pattern that could match an empty string could cause pcregrep to loop; it
1573 doesn't make sense to accept an empty string match in pcregrep, so I have
1574 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1575 seems to be how GNU grep behaves.
1576
1577 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1578 start or after a newline", because the conditional assertion was not being
1579 correctly handled. The rule now is that both the assertion and what follows
1580 in the first alternative must satisfy the test.
1581
1582 9. If auto-callout was enabled in a pattern with a conditional group whose
1583 condition was an assertion, PCRE could crash during matching, both with
1584 pcre_exec() and pcre_dfa_exec().
1585
1586 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1587 used for matching.
1588
1589 11. Unicode property support in character classes was not working for
1590 characters (bytes) greater than 127 when not in UTF-8 mode.
1591
1592 12. Added the -M command line option to pcretest.
1593
1594 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1595
1596 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1597
1598 16. Added comments and documentation about mis-use of no_arg in the C++
1599 wrapper.
1600
1601 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1602 from Martin Jerabek that uses macro names for all relevant character and
1603 string constants.
1604
1605 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1606 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1607 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1608 these, but not everybody uses configure.
1609
1610 19. A conditional group that had only one branch was not being correctly
1611 recognized as an item that could match an empty string. This meant that an
1612 enclosing group might also not be so recognized, causing infinite looping
1613 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1614 with the subject "ab", where knowledge that the repeated group can match
1615 nothing is needed in order to break the loop.
1616
1617 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1618 exec(), but without supplying a callout function, matching went wrong.
1619
1620 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1621 leak if the size of the offset vector was greater than 30. When the vector
1622 is smaller, the saved offsets during recursion go onto a local stack
1623 vector, but for larger vectors malloc() is used. It was failing to free
1624 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1625 error, in fact).
1626
1627 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1628 heapframe that is used only when UTF-8 support is enabled. This caused no
1629 problem, but was untidy.
1630
1631 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1632 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1633 included within another project.
1634
1635 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1636 slightly modified by me:
1637
1638 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1639 not building pcregrep.
1640
1641 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1642 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1643
1644 25. Forward references, both numeric and by name, in patterns that made use of
1645 duplicate group numbers, could behave incorrectly or give incorrect errors,
1646 because when scanning forward to find the reference group, PCRE was not
1647 taking into account the duplicate group numbers. A pattern such as
1648 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1649
1650 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1651 the feature of a custom pointer more persuasive (as requested by a user).
1652
1653 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1654 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1655 pre-defined.
1656
1657 28. Added support for (*UTF8) at the start of a pattern.
1658
1659 29. Arrange for flags added by the "release type" setting in CMake to be shown
1660 in the configuration summary.
1661
1662
1663 Version 7.8 05-Sep-08
1664 ---------------------
1665
1666 1. Replaced UCP searching code with optimized version as implemented for Ad
1667 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1668 stage table and inline lookup instead of a function, giving speed ups of 2
1669 to 5 times on some simple patterns that I tested. Permission was given to
1670 distribute the MultiStage2.py script that generates the tables (it's not in
1671 the tarball, but is in the Subversion repository).
1672
1673 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1674 scripts.
1675
1676 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1677 a group with a zero qualifier. The result of the study could be incorrect,
1678 or the function might crash, depending on the pattern.
1679
1680 4. Caseless matching was not working for non-ASCII characters in back
1681 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1682 It now works when Unicode Property Support is available.
1683
1684 5. In pcretest, an escape such as \x{de} in the data was always generating
1685 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1686 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1687 truncation.
1688
1689 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1690
1691 7. Added two (int) casts to pcregrep when printing the difference of two
1692 pointers, in case they are 64-bit values.
1693
1694 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1695 test 2 if it fails.
1696
1697 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1698 and a #define of that name to empty if it is not externally set. This is to
1699 allow users of MSVC to set it if necessary.
1700
1701 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1702 the convenience functions in the pcre_get.c source file.
1703
1704 11. An option change at the start of a pattern that had top-level alternatives
1705 could cause overwriting and/or a crash. This command provoked a crash in
1706 some environments:
1707
1708 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1709
1710 This potential security problem was recorded as CVE-2008-2371.
1711
1712 12. For a pattern where the match had to start at the beginning or immediately
1713 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1714 pcre_dfa_exec() could read past the end of the passed subject if there was
1715 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1716 pcretest so that it places the subject at the end of its malloc-ed buffer.
1717
1718 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1719 exec() might read past the end of the data buffer in UTF-8 mode.
1720
1721 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1722 the data contained the byte 0x85 as part of a UTF-8 character within its
1723 first line. This applied both to normal and DFA matching.
1724
1725 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1726 /^[^d]*?$/8 failed to match "abc".
1727
1728 16. Added a missing copyright notice to pcrecpp_internal.h.
1729
1730 17. Make it more clear in the documentation that values returned from
1731 pcre_exec() in ovector are byte offsets, not character counts.
1732
1733 18. Tidied a few places to stop certain compilers from issuing warnings.
1734
1735 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1736 supplied by Stefan Weber. I made a further small update for 7.8 because
1737 there is a change of source arrangements: the pcre_searchfuncs.c module is
1738 replaced by pcre_ucd.c.
1739
1740
1741 Version 7.7 07-May-08
1742 ---------------------
1743
1744 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1745 a string to a long long, pretend we don't even have a long long." This is
1746 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1747
1748 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1749 pre-7.6 versions, which defined a global no_arg variable instead of putting
1750 it in the RE class. (See also #8 below.)
1751
1752 3. Remove a line of dead code, identified by coverity and reported by Nuno
1753 Lopes.
1754
1755 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1756
1757 (1) The include/exclude patterns were being applied to the whole pathnames
1758 of files, instead of just to the final components.
1759
1760 (2) If there was more than one level of directory, the subdirectories were
1761 skipped unless they satisfied the include/exclude conditions. This is
1762 inconsistent with GNU grep (and could even be seen as contrary to the
1763 pcregrep specification - which I improved to make it absolutely clear).
1764 The action now is always to scan all levels of directory, and just
1765 apply the include/exclude patterns to regular files.
1766
1767 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1768 --exclude_dir in the tests to avoid scanning .svn directories.
1769
1770 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1771 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1772 doesn't support NULs in patterns.
1773
1774 7. Added some missing "const"s to declarations of static tables in
1775 pcre_compile.c and pcre_dfa_exec.c.
1776
1777 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1778 caused by fix #2 above. (Subsequently also a second patch to fix the
1779 first patch. And a third patch - this was a messy problem.)
1780
1781 9. Applied Craig's patch to remove the use of push_back().
1782
1783 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1784 matching function regexec().
1785
1786 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1787 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1788 references. PCRE supports relative numbers with this syntax (I don't think
1789 Oniguruma does).
1790
1791 12. Previously, a group with a zero repeat such as (...){0} was completely
1792 omitted from the compiled regex. However, this means that if the group
1793 was called as a subroutine from elsewhere in the pattern, things went wrong
1794 (an internal error was given). Such groups are now left in the compiled
1795 pattern, with a new opcode that causes them to be skipped at execution
1796 time.
1797
1798 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1799 to the way PCRE behaves:
1800
1801 (a) A lone ] character is dis-allowed (Perl treats it as data).
1802
1803 (b) A back reference to an unmatched subpattern matches an empty string
1804 (Perl fails the current match path).
1805
1806 (c) A data ] in a character class must be notated as \] because if the
1807 first data character in a class is ], it defines an empty class. (In
1808 Perl it is not possible to have an empty class.) The empty class []
1809 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1810 The negative empty class [^] matches any one character, independently
1811 of the DOTALL setting.
1812
1813 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1814 non-existent subpattern following a character class starting with ']' and
1815 containing () gave an internal compiling error instead of "reference to
1816 non-existent subpattern". Fortunately, when the pattern did exist, the
1817 compiled code was correct. (When scanning forwards to check for the
1818 existencd of the subpattern, it was treating the data ']' as terminating
1819 the class, so got the count wrong. When actually compiling, the reference
1820 was subsequently set up correctly.)
1821
1822 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1823 it was being rejected as not supported by pcre_dfa_exec(), even though
1824 other assertions are supported. I have made pcre_dfa_exec() support
1825 (*FAIL).
1826
1827 16. The implementation of 13c above involved the invention of a new opcode,
1828 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1829 cannot be changed at match time, I realized I could make a small
1830 improvement to matching performance by compiling OP_ALLANY instead of
1831 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1832 on the OP_ANY path.
1833
1834 17. Compiling pcretest on Windows with readline support failed without the
1835 following two fixes: (1) Make the unistd.h include conditional on
1836 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1837
1838 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1839 ncurses library to be included for pcretest when ReadLine support is
1840 requested, but also to allow for it to be overridden. This patch came from
1841 Daniel Bergström.
1842
1843 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1844 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1845 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1846 spotting this.
1847
1848
1849 Version 7.6 28-Jan-08
1850 ---------------------
1851
1852 1. A character class containing a very large number of characters with
1853 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1854 overflow.
1855
1856 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1857 HAVE_LONG_LONG is not defined.
1858
1859 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1860 bring them up to date and include new features. This patch includes:
1861
1862 - Fixed PH's badly added libz and libbz2 support.
1863 - Fixed a problem with static linking.
1864 - Added pcredemo. [But later removed - see 7 below.]
1865 - Fixed dftables problem and added an option.
1866 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1867 HAVE_LONG_LONG.
1868 - Added readline support for pcretest.
1869 - Added an listing of the option settings after cmake has run.
1870
1871 4. A user submitted a patch to Makefile that makes it easy to create
1872 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1873 Makefile.am that cause it to include this special target, without
1874 affecting anything else. Note that the same mingw target plus all
1875 the other distribution libraries and programs are now supported
1876 when configuring with CMake (see 6 below) instead of with
1877 Configure/Make.
1878
1879 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1880 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1881 exported in the Windows port". It has not yet been confirmed that the patch
1882 solves the problem, but it does no harm.
1883
1884 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1885 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1886 with CMake, and also correct the comment about stack recursion.
1887
1888 7. Remove the automatic building of pcredemo from the ./configure system and
1889 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1890 of a program that users should build themselves after PCRE is installed, so
1891 building it automatically is not really right. What is more, it gave
1892 trouble in some build environments.
1893
1894 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1895
1896
1897 Version 7.5 10-Jan-08
1898 ---------------------
1899
1900 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1901 values in parens when parsing an RE using the C++ wrapper."
1902
1903 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1904 Characters greater than 255 were excluded from the class instead of being
1905 included.
1906
1907 3. The same bug as (2) above applied to negated POSIX classes such as
1908 [:^space:].
1909
1910 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1911 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1912 I have changed it.
1913
1914 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1915 first named subpattern) and a construct such as (?&a) would reference the
1916 first named subpattern whose name started with "a" (in other words, the
1917 length check was missing). Both these problems are fixed. "Subpattern name
1918 expected" is now given for (?&) (a zero-length name), and this patch also
1919 makes it give the same error for \k'' (previously it complained that that
1920 was a reference to a non-existent subpattern).
1921
1922 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1923 this is right because (?- can be followed by option settings as well as by
1924 digits. I have, however, made the messages clearer.
1925
1926 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1927 than the number used in the conditional) now cause a compile-time error.
1928 This is actually not compatible with Perl, which accepts such patterns, but
1929 treats the conditional as always being FALSE (as PCRE used to), but it
1930 seems to me that giving a diagnostic is better.
1931
1932 8. Change "alphameric" to the more common word "alphanumeric" in comments
1933 and messages.
1934
1935 9. Fix two occurrences of "backslash" in comments that should have been
1936 "backspace".
1937
1938 10. Remove two redundant lines of code that can never be obeyed (their function
1939 was moved elsewhere).
1940
1941 11. The program that makes PCRE's Unicode character property table had a bug
1942 which caused it to generate incorrect table entries for sequences of
1943 characters that have the same character type, but are in different scripts.
1944 It amalgamated them into a single range, with the script of the first of
1945 them. In other words, some characters were in the wrong script. There were
1946 thirteen such cases, affecting characters in the following ranges:
1947
1948 U+002b0 - U+002c1
1949 U+0060c - U+0060d
1950 U+0061e - U+00612
1951 U+0064b - U+0065e
1952 U+0074d - U+0076d
1953 U+01800 - U+01805
1954 U+01d00 - U+01d77
1955 U+01d9b - U+01dbf
1956 U+0200b - U+0200f
1957 U+030fc - U+030fe
1958 U+03260 - U+0327f
1959 U+0fb46 - U+0fbb1
1960 U+10450 - U+1049d
1961
1962 12. The -o option (show only the matching part of a line) for pcregrep was not
1963 compatible with GNU grep in that, if there was more than one match in a
1964 line, it showed only the first of them. It now behaves in the same way as
1965 GNU grep.
1966
1967 13. If the -o and -v options were combined for pcregrep, it printed a blank
1968 line for every non-matching line. GNU grep prints nothing, and pcregrep now
1969 does the same. The return code can be used to tell if there were any
1970 non-matching lines.
1971
1972 14. Added --file-offsets and --line-offsets to pcregrep.
1973
1974 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
1975 infinitely looping recursion. The bug was that positive lookaheads were not
1976 being skipped when checking for a possible empty match (negative lookaheads
1977 and both kinds of lookbehind were skipped).
1978
1979 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
1980 inclusion of <windows.h> to before rather than after the definition of
1981 INVALID_FILE_ATTRIBUTES (patch from David Byron).
1982
1983 17. Specifying a possessive quantifier with a specific limit for a Unicode
1984 character property caused pcre_compile() to compile bad code, which led at
1985 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
1986 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
1987 caused the error; without that there was no problem.
1988
1989 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
1990
1991 19. Added --enable-pcretest-libreadline.
1992
1993 20. In pcrecpp.cc, the variable 'count' was incremented twice in
1994 RE::GlobalReplace(). As a result, the number of replacements returned was
1995 double what it should be. I removed one of the increments, but Craig sent a
1996 later patch that removed the other one (the right fix) and added unit tests
1997 that check the return values (which was not done before).
1998
1999 21. Several CMake things:
2000
2001 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2002 the names libpcre and libpcreposix, not just pcre and pcreposix.
2003
2004 (2) The above change means that pcretest and pcregrep are now correctly
2005 linked with the newly-built libraries, not previously installed ones.
2006
2007 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2008
2009 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2010 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2011 UTF-8 newline character). The key issue is that the pattern starts .*;
2012 this means that the match must be either at the beginning, or after a
2013 newline. The bug was in the code for advancing after a failed match and
2014 checking that the new position followed a newline. It was not taking
2015 account of UTF-8 characters correctly.
2016
2017 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2018 character classes. PCRE was not treating the sequence [:...:] as a
2019 character class unless the ... were all letters. Perl, however, seems to
2020 allow any characters between [: and :], though of course it rejects as
2021 unknown any "names" that contain non-letters, because all the known class
2022 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2023 for example, whereas PCRE did not - it did not recognize a POSIX character
2024 class. This seemed a bit dangerous, so the code has been changed to be
2025 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2026 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2027 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2028 Perl does, and where it didn't before.
2029
2030 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2031 Windows environments %n is disabled by default.
2032
2033
2034 Version 7.4 21-Sep-07
2035 ---------------------
2036
2037 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2038 means that a class such as [\s] counted as "explicit reference to CR or
2039 LF". That isn't really right - the whole point of the change was to try to
2040 help when there was an actual mention of one of the two characters. So now
2041 the change happens only if \r or \n (or a literal CR or LF) character is
2042 encountered.
2043
2044 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2045 of both had grown to the point where there were only 3 bits left.
2046 Fortunately, there was spare space in the data structure, and so I have
2047 moved the internal flags into a new 16-bit field to free up more option
2048 bits.
2049
2050 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2051 but did not set the internal JCHANGED flag - either of these is enough to
2052 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2053 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2054 start sets both bits.
2055
2056 4. Added options (at build time, compile time, exec time) to change \R from
2057 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2058
2059 5. doc/pcresyntax.html was missing from the distribution.
2060
2061 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2062 compatibility, even though it is no longer used.
2063
2064 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2065 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2066 windows.h file is present (where different names are used). [This was
2067 reversed later after testing - see 16 below.]
2068
2069 8. Changed all #include <config.h> to #include "config.h". There were also
2070 some further <pcre.h> cases that I changed to "pcre.h".
2071
2072 9. When pcregrep was used with the --colour option, it missed the line ending
2073 sequence off the lines that it output.
2074
2075 10. It was pointed out to me that arrays of string pointers cause lots of
2076 relocations when a shared library is dynamically loaded. A technique of
2077 using a single long string with a table of offsets can drastically reduce
2078 these. I have refactored PCRE in four places to do this. The result is
2079 dramatic:
2080
2081 Originally: 290
2082 After changing UCP table: 187
2083 After changing error message table: 43
2084 After changing table of "verbs" 36
2085 After changing table of Posix names 22
2086
2087 Thanks to the folks working on Gregex for glib for this insight.
2088
2089 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2090 unicode-properties was also set.
2091
2092 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2093
2094 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2095 checked only for CRLF.
2096
2097 14. Added casts to pcretest.c to avoid compiler warnings.
2098
2099 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2100
2101 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2102 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2103 entirely. This removes changes made in 7 above.
2104
2105 17. The CMake files have been updated, and there is now more information about
2106 building with CMake in the NON-UNIX-USE document.
2107
2108
2109 Version 7.3 28-Aug-07
2110 ---------------------
2111
2112 1. In the rejigging of the build system that eventually resulted in 7.1, the
2113 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2114 brackets there is not right, since it causes compilers to look for an
2115 installed pcre.h, not the version that is in the source that is being
2116 compiled (which of course may be different). I have changed it back to:
2117
2118 #include "pcre.h"
2119
2120 I have a vague recollection that the change was concerned with compiling in
2121 different directories, but in the new build system, that is taken care of
2122 by the VPATH setting the Makefile.
2123
2124 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2125 when the subject happened to end in the byte 0x85 (e.g. if the last
2126 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2127 characters but of course it shouldn't be taken as a newline when it is part
2128 of another character. The bug was that, for an unlimited repeat of . in
2129 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2130 characters when looking for a newline.
2131
2132 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2133
2134 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2135 in debug output.
2136
2137 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2138 long printing in the pcrecpp unittest when running under MinGW.
2139
2140 6. ESC_K was left out of the EBCDIC table.
2141
2142 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2143 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2144 limit also applies to "virtual nesting" when a pattern is recursive, and in
2145 this case 1000 isn't so big. I have been able to remove this limit at the
2146 expense of backing off one optimization in certain circumstances. Normally,
2147 when pcre_exec() would call its internal match() function recursively and
2148 immediately return the result unconditionally, it uses a "tail recursion"
2149 feature to save stack. However, when a subpattern that can match an empty
2150 string has an unlimited repetition quantifier, it no longer makes this
2151 optimization. That gives it a stack frame in which to save the data for
2152 checking that an empty string has been matched. Previously this was taken
2153 from the 1000-entry workspace that had been reserved. So now there is no
2154 explicit limit, but more stack is used.
2155
2156 8. Applied Daniel's patches to solve problems with the import/export magic
2157 syntax that is required for Windows, and which was going wrong for the
2158 pcreposix and pcrecpp parts of the library. These were overlooked when this
2159 problem was solved for the main library.
2160
2161 9. There were some crude static tests to avoid integer overflow when computing
2162 the size of patterns that contain repeated groups with explicit upper
2163 limits. As the maximum quantifier is 65535, the maximum group length was
2164 set at 30,000 so that the product of these two numbers did not overflow a
2165 32-bit integer. However, it turns out that people want to use groups that
2166 are longer than 30,000 bytes (though not repeat them that many times).
2167 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2168 made it possible to implement the integer overflow checks in a much more
2169 dynamic way, which I have now done. The artificial limitation on group
2170 length has been removed - we now have only the limit on the total length of
2171 the compiled pattern, which depends on the LINK_SIZE setting.
2172
2173 10. Fixed a bug in the documentation for get/copy named substring when
2174 duplicate names are permitted. If none of the named substrings are set, the
2175 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2176 empty string.
2177
2178 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2179 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2180 because the ] is interpreted as the first data character and the
2181 terminating ] is not found. PCRE has been made compatible with Perl in this
2182 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2183 cause memory overwriting.
2184
2185 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2186 string has been matched (to stop an infinite loop). It was not recognizing
2187 a conditional subpattern that could match an empty string if that
2188 subpattern was within another subpattern. For example, it looped when
2189 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2190 condition was not nested. This bug has been fixed.
2191
2192 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2193 past the start of the subject in the presence of bytes with the top bit
2194 set, for example "\x8aBCD".
2195
2196 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2197 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2198
2199 14. Optimized (?!) to (*FAIL).
2200
2201 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2202 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2203 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2204 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2205 does: it's just the validity check that is more restrictive.
2206
2207 16. Inserted checks for integer overflows during escape sequence (backslash)
2208 processing, and also fixed erroneous offset values for syntax errors during
2209 backslash processing.
2210
2211 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2212 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2213
2214 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2215 caused an overrun.
2216
2217 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2218 something other than just ASCII characters) inside a group that had an
2219 unlimited repeat caused a loop at compile time (while checking to see
2220 whether the group could match an empty string).
2221
2222 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2223 [\P{Any}] did so. (Error in the code for printing property names.)
2224
2225 21. An orphan \E inside a character class could cause a crash.
2226
2227 22. A repeated capturing bracket such as (A)? could cause a wild memory
2228 reference during compilation.
2229
2230 23. There are several functions in pcre_compile() that scan along a compiled
2231 expression for various reasons (e.g. to see if it's fixed length for look
2232 behind). There were bugs in these functions when a repeated \p or \P was
2233 present in the pattern. These operators have additional parameters compared
2234 with \d, etc, and these were not being taken into account when moving along
2235 the compiled data. Specifically:
2236
2237 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2238 length.
2239
2240 (b) An item such as \pL+ within a repeated group could cause crashes or
2241 loops.
2242
2243 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2244 "reference to non-existent subpattern" error.
2245
2246 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2247
2248 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2249 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2250
2251 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2252
2253 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2254 character were causing crashes (broken optimization).
2255
2256 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2257 \p or \P) caused a compile-time loop.
2258
2259 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2260 break. For example, the unstudied pattern [\r\n]A does not match the string
2261 "\r\nA" because change 7.0/46 below moves the current point on by two
2262 characters after failing to match at the start. However, the pattern \nA
2263 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2264 the same is true. There doesn't seem any very clean way out of this, but
2265 what I have chosen to do makes the common cases work: PCRE now takes note
2266 of whether there can be an explicit match for \r or \n anywhere in the
2267 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2268 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2269 pattern has explicit CR or LF references.
2270
2271 29. Added (*CR) etc for changing newline setting at start of pattern.
2272
2273
2274 Version 7.2 19-Jun-07
2275 ---------------------
2276
2277 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2278 which is apparently normally available under Windows.
2279
2280 2. Re-jig the pcregrep tests with different newline settings in an attempt
2281 to make them independent of the local environment's newline setting.
2282
2283 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2284
2285 4. Some of the "internals" tests were previously cut out when the link size
2286 was not 2, because the output contained actual offsets. The recent new
2287 "Z" feature of pcretest means that these can be cut out, making the tests
2288 usable with all link sizes.
2289
2290 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2291 stack recursion. This gives a massive performance boost under BSD, but just
2292 a small improvement under Linux. However, it saves one field in the frame
2293 in all cases.
2294
2295 6. Added more features from the forthcoming Perl 5.10:
2296
2297 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2298 recursion call. It refers to the nth most recently opened parentheses.
2299
2300 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2301 to be opened parentheses.
2302
2303 (c) Conditions that refer to capturing parentheses can be specified
2304 relatively, for example, (?(-2)... or (?(+3)...
2305
2306 (d) \K resets the start of the current match so that everything before
2307 is not part of it.
2308
2309 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2310
2311 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2312 reference syntax.
2313
2314 (g) (?| introduces a group in which the numbering of parentheses in each
2315 alternative starts with the same number.
2316
2317 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2318
2319 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2320 PCRE_INFO_JCHANGED.
2321
2322 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2323 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2324 for detecting groups that can match an empty string.
2325
2326 9. A pattern with a very large number of alternatives (more than several
2327 hundred) was running out of internal workspace during the pre-compile
2328 phase, where pcre_compile() figures out how much memory will be needed. A
2329 bit of new cunning has reduced the workspace needed for groups with
2330 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2331 workspace instead of running out of the 4096 that are available.
2332
2333 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2334
2335 11. Applied patch from Google to remove an optimization that didn't quite work.
2336 The report of the bug said:
2337
2338 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2339 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2340 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2341
2342 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2343 it matched the wrong number of bytes.
2344
2345
2346 Version 7.1 24-Apr-07
2347 ---------------------
2348
2349 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2350 that is more "standard", making use of automake and other Autotools. There
2351 is some re-arrangement of the files and adjustment of comments consequent
2352 on this.
2353
2354 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2355 for recursive directory scanning broke on some systems because the files
2356 are not scanned in any specific order and on different systems the order
2357 was different. A call to "sort" has been inserted into RunGrepTest for the
2358 approprate test as a short-term fix. In the longer term there may be an
2359 alternative.
2360
2361 3. I had an email from Eric Raymond about problems translating some of PCRE's
2362 man pages to HTML (despite the fact that I distribute HTML pages, some
2363 people do their own conversions for various reasons). The problems
2364 concerned the use of low-level troff macros .br and .in. I have therefore
2365 removed all such uses from the man pages (some were redundant, some could
2366 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2367 HTML has been updated to handle .nf/.fi and to complain if it encounters
2368 .br or .in.
2369
2370 4. Updated comments in configure.ac that get placed in config.h.in and also
2371 arranged for config.h to be included in the distribution, with the name
2372 config.h.generic, for the benefit of those who have to compile without
2373 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2374
2375 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2376 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2377 makevp.bat for the current PCRE, using the additional files
2378 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2379
2380 6. A Windows user reported a minor discrepancy with test 2, which turned out
2381 to be caused by a trailing space on an input line that had got lost in his
2382 copy. The trailing space was an accident, so I've just removed it.
2383
2384 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2385 that is needed.
2386
2387 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2388 as "const" (a) because they are and (b) because it helps the PHP
2389 maintainers who have recently made a script to detect big data structures
2390 in the php code that should be moved to the .rodata section. I remembered
2391 to update Builducptable as well, so it won't revert if ucptable.h is ever
2392 re-created.
2393
2394 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2395 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2396 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2397 support is not required. This saves 1.5-2K of code, which is important in
2398 some applications.
2399
2400 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2401 so as not to refer to the tables, even though these functions will never be
2402 called when UTF-8 support is disabled. Otherwise there are problems with a
2403 shared library.
2404
2405 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2406
2407 (a) It was defining its arguments as char * instead of void *.
2408
2409 (b) It was assuming that all moves were upwards in memory; this was true
2410 a long time ago when I wrote it, but is no longer the case.
2411
2412 The emulated memove() is provided for those environments that have neither
2413 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2414 is clearly not the case, as these two bugs were recently reported.
2415
2416 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2417 and Detrail to create the HTML documentation, the .txt form of the man
2418 pages, and it removes trailing spaces from listed files. It also creates
2419 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2420 case, it wraps all the #defines with #ifndefs. This script should be run
2421 before "make dist".
2422
2423 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2424 with Unicode property support.
2425
2426 (a) For a maximizing quantifier, if the two different cases of the
2427 character were of different lengths in their UTF-8 codings (there are
2428 some cases like this - I found 11), and the matching function had to
2429 back up over a mixture of the two cases, it incorrectly assumed they
2430 were both the same length.
2431
2432 (b) When PCRE was configured to use the heap rather than the stack for
2433 recursion during matching, it was not correctly preserving the data for
2434 the other case of a UTF-8 character when checking ahead for a match
2435 while processing a minimizing repeat. If the check also involved
2436 matching a wide character, but failed, corruption could cause an
2437 erroneous result when trying to check for a repeat of the original
2438 character.
2439
2440 13. Some tidying changes to the testing mechanism:
2441
2442 (a) The RunTest script now detects the internal link size and whether there
2443 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2444 values substituted by "configure". (The RunGrepTest script already did
2445 this for UTF-8.) The configure.ac script no longer substitutes the
2446 relevant variables.
2447
2448 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2449 with length and offset values. This means that the output is different
2450 for different internal link sizes. Test 2 is skipped for link sizes
2451 other than 2 because of this, bypassing the problem. Unfortunately,
2452 there was also a test in test 3 (the locale tests) that used /B and
2453 failed for link sizes other than 2. Rather than cut the whole test out,
2454 I have added a new /Z option to pcretest that replaces the length and
2455 offset values with spaces. This is now used to make test 3 independent
2456 of link size. (Test 2 will be tidied up later.)
2457
2458 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2459 segmentation fault instead of returning the appropriate error message.
2460
2461 15. In multiline mode when the newline sequence was set to "any", the pattern
2462 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2463 This doesn't seem right; it now treats the CRLF combination as the line
2464 ending, and so does not match in that case. It's only a pattern such as ^$
2465 that would hit this one: something like ^ABC$ would have failed after \r
2466 and then tried again after \r\n.
2467
2468 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2469 in an attempt to make files that differ only in their line terminators
2470 compare equal. This works on Linux.
2471
2472 17. Under certain error circumstances pcregrep might try to free random memory
2473 as it exited. This is now fixed, thanks to valgrind.
2474
2475 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2476 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2477 was because its rules for how to advance for /g after matching an empty
2478 string at the end of a line did not allow for this case. They now check for
2479 it specially.
2480
2481 20. pcretest is supposed to handle patterns and data of any length, by
2482 extending its buffers when necessary. It was getting this wrong when the
2483 buffer for a data line had to be extended.
2484
2485 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2486 CRLF as a newline sequence.
2487
2488 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2489 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2490 I have nevertheless tidied it up.
2491
2492 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2493
2494 24. Added a man page for pcre-config.
2495
2496
2497 Version 7.0 19-Dec-06
2498 ---------------------
2499
2500 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2501 moving to gcc 4.1.1.
2502
2503 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2504 sys/time.h, which is documented as needed for this function. It doesn't
2505 seem to matter on Linux, but it showed up on some releases of OS X.
2506
2507 3. It seems that there are systems where bytes whose values are greater than
2508 127 match isprint() in the "C" locale. The "C" locale should be the
2509 default when a C program starts up. In most systems, only ASCII printing
2510 characters match isprint(). This difference caused the output from pcretest
2511 to vary, making some of the tests fail. I have changed pcretest so that:
2512
2513 (a) When it is outputting text in the compiled version of a pattern, bytes
2514 other than 32-126 are always shown as hex escapes.
2515
2516 (b) When it is outputting text that is a matched part of a subject string,
2517 it does the same, unless a different locale has been set for the match
2518 (using the /L modifier). In this case, it uses isprint() to decide.
2519
2520 4. Fixed a major bug that caused incorrect computation of the amount of memory
2521 required for a compiled pattern when options that changed within the
2522 pattern affected the logic of the preliminary scan that determines the
2523 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2524 that the computed length was too small. The symptoms of this bug were
2525 either the PCRE error "internal error: code overflow" from pcre_compile(),
2526 or a glibc crash with a message such as "pcretest: free(): invalid next
2527 size (fast)". Examples of patterns that provoked this bug (shown in
2528 pcretest format) are:
2529
2530 /(?-x: )/x
2531 /(?x)(?-x: \s*#\s*)/
2532 /((?i)[\x{c0}])/8
2533 /(?i:[\x{c0}])/8
2534
2535 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2536 is now done differently.
2537
2538 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2539 wrapper classes; (b) implement a new function in the C++ scanner that is
2540 more efficient than the old way of doing things because it avoids levels of
2541 recursion in the regex matching; (c) add a paragraph to the documentation
2542 for the FullMatch() function.
2543
2544 6. The escape sequence \n was being treated as whatever was defined as
2545 "newline". Not only was this contrary to the documentation, which states
2546 that \n is character 10 (hex 0A), but it also went horribly wrong when
2547 "newline" was defined as CRLF. This has been fixed.
2548
2549 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2550 was being set to -1 for the "end of line" case (supposedly a value that no
2551 character can have). Though this value is never used (the check for end of
2552 line is "zero bytes in current character"), it caused compiler complaints.
2553 I've changed it to 0xffffffff.
2554
2555 8. In pcre_version.c, the version string was being built by a sequence of
2556 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2557 string (as it is for production releases) called a macro with an empty
2558 argument. The C standard says the result of this is undefined. The gcc
2559 compiler treats it as an empty string (which was what was wanted) but it is
2560 reported that Visual C gives an error. The source has been hacked around to
2561 avoid this problem.
2562
2563 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2564 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2565 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2566 of them did).
2567
2568 10. Originally, pcretest opened its input and output without "b"; then I was
2569 told that "b" was needed in some environments, so it was added for release
2570 5.0 to both the input and output. (It makes no difference on Unix-like
2571 systems.) Later I was told that it is wrong for the input on Windows. I've
2572 now abstracted the modes into two macros, to make it easier to fiddle with
2573 them, and removed "b" from the input mode under Windows.
2574
2575 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2576
2577 12. Added -help and --help to pcretest as an official way of being reminded
2578 of the options.
2579
2580 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2581 and pcrecpp.cc because they annoy compilers at high warning levels.
2582
2583 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2584
2585 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2586 scripts are not C programs :-) and which was not noticed because it works
2587 on Linux.
2588
2589 16. pcretest is supposed to handle any length of pattern and data line (as one
2590 line or as a continued sequence of lines) by extending its input buffer if
2591 necessary. This feature was broken for very long pattern lines, leading to
2592 a string of junk being passed to pcre_compile() if the pattern was longer
2593 than about 50K.
2594
2595 17. I have done a major re-factoring of the way pcre_compile() computes the
2596 amount of memory needed for a compiled pattern. Previously, there was code
2597 that made a preliminary scan of the pattern in order to do this. That was
2598 OK when PCRE was new, but as the facilities have expanded, it has become
2599 harder and harder to keep it in step with the real compile phase, and there
2600 have been a number of bugs (see for example, 4 above). I have now found a
2601 cunning way of running the real compile function in a "fake" mode that
2602 enables it to compute how much memory it would need, while actually only
2603 ever using a few hundred bytes of working memory and without too many
2604 tests of the mode. This should make future maintenance and development
2605 easier. A side effect of this work is that the limit of 200 on the nesting
2606 depth of parentheses has been removed (though this was never a serious
2607 limitation, I suspect). However, there is a downside: pcre_compile() now
2608 runs more slowly than before (30% or more, depending on the pattern). I
2609 hope this isn't a big issue. There is no effect on runtime performance.
2610
2611 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2612 newline (only possible for the last line of a file) and it was a
2613 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2614
2615 19. Added additional timing features to pcretest. (1) The -tm option now times
2616 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2617 separate command line item, by a number that specifies the number of
2618 repeats to use when timing. The default is 50000; this gives better
2619 precision, but takes uncomfortably long for very large patterns.
2620
2621 20. Extended pcre_study() to be more clever in cases where a branch of a
2622 subpattern has no definite first character. For example, (a*|b*)[cd] would
2623 previously give no result from pcre_study(). Now it recognizes that the
2624 first character must be a, b, c, or d.
2625
2626 21. There was an incorrect error "recursive call could loop indefinitely" if
2627 a subpattern (or the entire pattern) that was being tested for matching an
2628 empty string contained only one non-empty item after a nested subpattern.
2629 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2630 incorrectly, because the \d was being skipped in the check.
2631
2632 22. The pcretest program now has a new pattern option /B and a command line
2633 option -b, which is equivalent to adding /B to every pattern. This causes
2634 it to show the compiled bytecode, without the additional information that
2635 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2636 is the same as /B/I).
2637
2638 23. A new optimization is now able automatically to treat some sequences such
2639 as a*b as a*+b. More specifically, if something simple (such as a character
2640 or a simple class like \d) has an unlimited quantifier, and is followed by
2641 something that cannot possibly match the quantified thing, the quantifier
2642 is automatically "possessified".
2643
2644 24. A recursive reference to a subpattern whose number was greater than 39
2645 went wrong under certain circumstances in UTF-8 mode. This bug could also
2646 have affected the operation of pcre_study().
2647
2648 25. Realized that a little bit of performance could be had by replacing
2649 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2650
2651 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2652
2653 27. Possessive quantifiers such as a++ were previously implemented by turning
2654 them into atomic groups such as ($>a+). Now they have their own opcodes,
2655 which improves performance. This includes the automatically created ones
2656 from 23 above.
2657
2658 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2659 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2660 the first matched character to be a colon. This applied both to named and
2661 numbered groups.
2662
2663 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2664
2665 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2666 building PCRE on Windows easier, so I have included it in the distribution.
2667
2668 31. There is now a check in pcretest against a ridiculously large number being
2669 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2670 loop, the loop is abandoned.
2671
2672 32. Forward references to subpatterns in conditions such as (?(2)...) where
2673 subpattern 2 is defined later cause pcre_compile() to search forwards in
2674 the pattern for the relevant set of parentheses. This search went wrong
2675 when there were unescaped parentheses in a character class, parentheses
2676 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2677
2678 33. "Subroutine" calls and backreferences were previously restricted to
2679 referencing subpatterns earlier in the regex. This restriction has now
2680 been removed.
2681
2682 34. Added a number of extra features that are going to be in Perl 5.10. On the
2683 whole, these are just syntactic alternatives for features that PCRE had
2684 previously implemented using the Python syntax or my own invention. The
2685 other formats are all retained for compatibility.
2686
2687 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2688 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2689 also .NET compatible.
2690
2691 (b) A recursion or subroutine call to a named group can now be defined as
2692 (?&name) as well as (?P>name).
2693
2694 (c) A backreference to a named group can now be defined as \k<name> or
2695 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2696 5.10, are also .NET compatible.
2697
2698 (d) A conditional reference to a named group can now use the syntax
2699 (?(<name>) or (?('name') as well as (?(name).
2700
2701 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2702 groups (named and numbered) that are never evaluated inline, but can be
2703 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2704 is always false. There may be only one alternative in such a group.
2705
2706 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2707 as the simple (?(R). The condition is true only if the most recent
2708 recursion is that of the given number or name. It does not search out
2709 through the entire recursion stack.
2710
2711 (g) The escape \gN or \g{N} has been added, where N is a positive or
2712 negative number, specifying an absolute or relative reference.
2713
2714 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2715 some "unreachable code" warnings.
2716
2717 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2718 things, this adds five new scripts.
2719
2720 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2721 There were also incompatibilities regarding the handling of \Q..\E inside
2722 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2723 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2724
2725 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2726 matches an empty string, and forcibly breaks the loop. There were bugs in
2727 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2728 against aaaa the result was just "a" rather than "aaaa", for example. Two
2729 separate and independent bugs (that affected different cases) have been
2730 fixed.
2731
2732 39. Refactored the code to abolish the use of different opcodes for small
2733 capturing bracket numbers. This is a tidy that I avoided doing when I
2734 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2735 The new approach is not only tidier, it makes it possible to reduce the
2736 memory needed to fix the previous bug (38).
2737
2738 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2739 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2740 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2741 mode.
2742
2743 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2744 report.
2745
2746 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2747 copy construction and assignment in the C++ wrapper.
2748
2749 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2750 couple of bugs that could have given wrong results in the "--newline=crlf"
2751 case.
2752
2753 44. Added a number of casts and did some reorganization of signed/unsigned int
2754 variables following suggestions from Dair Grant. Also renamed the variable
2755 "this" as "item" because it is a C++ keyword.
2756
2757 45. Arranged for dftables to add
2758
2759 #include "pcre_internal.h"
2760
2761 to pcre_chartables.c because without it, gcc 4.x may remove the array
2762 definition from the final binary if PCRE is built into a static library and
2763 dead code stripping is activated.
2764
2765 46. For an unanchored pattern, if a match attempt fails at the start of a
2766 newline sequence, and the newline setting is CRLF or ANY, and the next two
2767 characters are CRLF, advance by two characters instead of one.
2768
2769
2770 Version 6.7 04-Jul-06
2771 ---------------------
2772
2773 1. In order to handle tests when input lines are enormously long, pcretest has
2774 been re-factored so that it automatically extends its buffers when
2775 necessary. The code is crude, but this _is_ just a test program. The
2776 default size has been increased from 32K to 50K.
2777
2778 2. The code in pcre_study() was using the value of the re argument before
2779 testing it for NULL. (Of course, in any sensible call of the function, it
2780 won't be NULL.)
2781
2782 3. The memmove() emulation function in pcre_internal.h, which is used on
2783 systems that lack both memmove() and bcopy() - that is, hardly ever -
2784 was missing a "static" storage class specifier.
2785
2786 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2787 containing an extended class (one that cannot be represented by a bitmap
2788 because it contains high-valued characters or Unicode property items, e.g.
2789 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2790 pattern, but PCRE should not loop if you do not (it no longer does).
2791 [Detail: two cases were found: (a) a repeated subpattern containing an
2792 extended class; (b) a recursive reference to a subpattern that followed a
2793 previous extended class. It wasn't skipping over the extended class
2794 correctly when UTF-8 mode was not set.]
2795
2796 5. A negated single-character class was not being recognized as fixed-length
2797 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2798 compile error "lookbehind assertion is not fixed length".
2799
2800 6. The RunPerlTest auxiliary script was showing an unexpected difference
2801 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2802 write a Perl script that can interpret lines of an input file either as
2803 byte characters or as UTF-8, which is what "perltest" was being required to
2804 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2805 can't do is switch easily at run time between having the "use utf8;" pragma
2806 or not. In the end, I fudged it by using the RunPerlTest script to insert
2807 "use utf8;" explicitly for the UTF-8 tests.
2808
2809 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2810 the end of the subject string, contrary to the documentation and to what
2811 Perl does. This was true of both matching functions. Now it matches only at
2812 the start of the subject and immediately after *internal* newlines.
2813
2814 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2815 a pointer to an int instead of a pointer to an unsigned long int. This
2816 caused problems on 64-bit systems.
2817
2818 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2819 instance of the 'standard' template library not being so standard".
2820
2821 10. There was no check on the number of named subpatterns nor the maximum
2822 length of a subpattern name. The product of these values is used to compute
2823 the size of the memory block for a compiled pattern. By supplying a very
2824 long subpattern name and a large number of named subpatterns, the size
2825 computation could be caused to overflow. This is now prevented by limiting
2826 the length of names to 32 characters, and the number of named subpatterns
2827 to 10,000.
2828
2829 11. Subpatterns that are repeated with specific counts have to be replicated in
2830 the compiled pattern. The size of memory for this was computed from the
2831 length of the subpattern and the repeat count. The latter is limited to
2832 65535, but there was no limit on the former, meaning that integer overflow
2833 could in principle occur. The compiled length of a repeated subpattern is
2834 now limited to 30,000 bytes in order to prevent this.
2835
2836 12. Added the optional facility to have named substrings with the same name.
2837
2838 13. Added the ability to use a named substring as a condition, using the
2839 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2840 are numbers (not recommended). Forward references are permitted.
2841
2842 14. Added forward references in named backreferences (if you see what I mean).
2843
2844 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2845 pattern could run off the end of the subject. For example, the pattern
2846 "(?s)(.{1,5})"8 did this with the subject "ab".
2847
2848 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2849 PCRE_CASELESS was set when matching characters that were quantified with ?
2850 or *.
2851
2852 17. A character class other than a single negated character that had a minimum
2853 but no maximum quantifier - for example [ab]{6,} - was not handled
2854 correctly by pce_dfa_exec(). It would match only one character.
2855
2856 18. A valid (though odd) pattern that looked like a POSIX character
2857 class but used an invalid character after [ (for example [[,abc,]]) caused
2858 pcre_compile() to give the error "Failed: internal error: code overflow" or
2859 in some cases to crash with a glibc free() error. This could even happen if
2860 the pattern terminated after [[ but there just happened to be a sequence of
2861 letters, a binary zero, and a closing ] in the memory that followed.
2862
2863 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2864 over the years. Originally (before any Unicode support), just the bottom 8
2865 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2866 output from "man perlunicode" includes this:
2867
2868 The regular expression compiler produces polymorphic opcodes. That
2869 is, the pattern adapts to the data and automatically switches to
2870 the Unicode character scheme when presented with Unicode data--or
2871 instead uses a traditional byte scheme when presented with byte
2872 data.
2873
2874 Sadly, a wide octal escape does not cause a switch, and in a string with
2875 no other multibyte characters, these octal escapes are treated as before.
2876 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2877 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2878 Unicode string.
2879
2880 I have not perpetrated such confusion in PCRE. Up till now, it took just
2881 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2882 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2883 translate to the appropriate multibyte character.
2884
2885 29. Applied some refactoring to reduce the number of warnings from Microsoft
2886 and Borland compilers. This has included removing the fudge introduced
2887 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2888 a warning about an unused variable.
2889
2890 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2891 characters since release 4.0, because Perl (from release 5.004) does not.
2892 [Or at least, is documented not to: some releases seem to be in conflict
2893 with the documentation.] However, when a pattern was studied with
2894 pcre_study() and all its branches started with \s, PCRE still included VT
2895 as a possible starting character. Of course, this did no harm; it just
2896 caused an unnecessary match attempt.
2897
2898 22. Removed a now-redundant internal flag bit that recorded the fact that case
2899 dependency changed within the pattern. This was once needed for "required
2900 byte" processing, but is no longer used. This recovers a now-scarce options
2901 bit. Also moved the least significant internal flag bit to the most-
2902 significant bit of the word, which was not previously used (hangover from
2903 the days when it was an int rather than a uint) to free up another bit for
2904 the future.
2905
2906 23. Added support for CRLF line endings as well as CR and LF. As well as the
2907 default being selectable at build time, it can now be changed at runtime
2908 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2909 specify that it is scanning data with non-default line endings.
2910
2911 24. Changed the definition of CXXLINK to make it agree with the definition of
2912 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2913
2914 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2915 recursions. This makes a big different to stack usage for some patterns.
2916
2917 26. If a subpattern containing a named recursion or subroutine reference such
2918 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2919 the space required for the compiled pattern went wrong and gave too small a
2920 value. Depending on the environment, this could lead to "Failed: internal
2921 error: code overflow at offset 49" or "glibc detected double free or
2922 corruption" errors.
2923
2924 27. Applied patches from Google (a) to support the new newline modes and (b) to
2925 advance over multibyte UTF-8 characters in GlobalReplace.
2926
2927 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2928 difference for some implementation of PCRE in some Windows version.
2929
2930 29. Added some extra testing facilities to pcretest:
2931
2932 \q<number> in a data line sets the "match limit" value
2933 \Q<number> in a data line sets the "match recursion limt" value
2934 -S <number> sets the stack size, where <number> is in megabytes
2935
2936 The -S option isn't available for Windows.
2937
2938
2939 Version 6.6 06-Feb-06
2940 ---------------------
2941
2942 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
2943 in pcreposix.h. I have copied the definition from pcre.h.
2944
2945 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
2946 because pcre.h is no longer a built file.
2947
2948 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
2949 not normally included in the compiled code.
2950
2951
2952 Version 6.5 01-Feb-06
2953 ---------------------
2954
2955 1. When using the partial match feature with pcre_dfa_exec(), it was not
2956 anchoring the second and subsequent partial matches at the new starting
2957 point. This could lead to incorrect results. For example, with the pattern
2958 /1234/, partially matching against "123" and then "a4" gave a match.
2959
2960 2. Changes to pcregrep:
2961
2962 (a) All non-match returns from pcre_exec() were being treated as failures
2963 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
2964 error message is output. Some extra information is given for the
2965 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
2966 probably the only errors that are likely to be caused by users (by
2967 specifying a regex that has nested indefinite repeats, for instance).
2968 If there are more than 20 of these errors, pcregrep is abandoned.
2969
2970 (b) A binary zero was treated as data while matching, but terminated the
2971 output line if it was written out. This has been fixed: binary zeroes
2972 are now no different to any other data bytes.
2973
2974 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
2975 used to set a locale for matching. The --locale=xxxx long option has
2976 been added (no short equivalent) to specify a locale explicitly on the
2977 pcregrep command, overriding the environment variables.
2978
2979 (d) When -B was used with -n, some line numbers in the output were one less
2980 than they should have been.
2981
2982 (e) Added the -o (--only-matching) option.
2983
2984 (f) If -A or -C was used with -c (count only), some lines of context were
2985 accidentally printed for the final match.
2986
2987 (g) Added the -H (--with-filename) option.
2988
2989 (h) The combination of options -rh failed to suppress file names for files
2990 that were found from directory arguments.
2991
2992 (i) Added the -D (--devices) and -d (--directories) options.
2993
2994 (j) Added the -F (--fixed-strings) option.
2995
2996 (k) Allow "-" to be used as a file name for -f as well as for a data file.
2997
2998 (l) Added the --colo(u)r option.
2999
3000 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3001 is not present by default.
3002
3003 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3004 items such as (?R) or (?1), when the recursion could match a number of
3005 alternatives. If it matched one of the alternatives, but subsequently,
3006 outside the recursion, there was a failure, the code tried to back up into
3007 the recursion. However, because of the way PCRE is implemented, this is not
3008 possible, and the result was an incorrect result from the match.
3009
3010 In order to prevent this happening, the specification of recursion has
3011 been changed so that all such subpatterns are automatically treated as
3012 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3013
3014 4. I had overlooked the fact that, in some locales, there are characters for
3015 which isalpha() is true but neither isupper() nor islower() are true. In
3016 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3017 and ordfeminine) are like this. This affected the treatment of \w and \W
3018 when they appeared in character classes, but not when they appeared outside
3019 a character class. The bit map for "word" characters is now created
3020 separately from the results of isalnum() instead of just taking it from the
3021 upper, lower, and digit maps. (Plus the underscore character, of course.)
3022
3023 5. The above bug also affected the handling of POSIX character classes such as
3024 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3025 permanent tables. Instead, the bit maps for such a class were previously
3026 created as the appropriate unions of the upper, lower, and digit bitmaps.
3027 Now they are created by subtraction from the [[:word:]] class, which has
3028 its own bitmap.
3029
3030 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3031 It is created by subtracting the vertical space characters (\x09, \x0a,
3032 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3033 subtraction was done in the overall bitmap for a character class, meaning
3034 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3035 be recognized. This bug has been fixed.
3036
3037 7. Patches from the folks at Google:
3038
3039 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3040 real life, but is still worth protecting against".
3041
3042 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3043 regular expressions".
3044
3045 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3046 have it.
3047
3048 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3049 "configure" and the latter not, in order to fix a problem somebody had
3050 with compiling the Arg class on HP-UX.
3051
3052 (e) Improve the error-handling of the C++ wrapper a little bit.
3053
3054 (f) New tests for checking recursion limiting.
3055
3056 8. The pcre_memmove() function, which is used only if the environment does not
3057 have a standard memmove() function (and is therefore rarely compiled),
3058 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3059 returning a result (though PCRE never actually uses the result).
3060
3061 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3062 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3063 returned instead of calling malloc() with an overflowing number that would
3064 most likely cause subsequent chaos.
3065
3066 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3067
3068 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3069 with this option is matched, the nmatch and pmatch options of regexec() are
3070 ignored.
3071
3072 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3073 provided in case anyone wants to the the POSIX interface with UTF-8
3074 strings.
3075
3076 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3077 C++ linking (needed for some HP-UX environments).
3078
3079 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3080 (unused parameter) and in the pcre_printint() function (omitted "default"
3081 switch label when the default is to do nothing).
3082
3083 15. Added some code to make it possible, when PCRE is compiled as a C++
3084 library, to replace subject pointers for pcre_exec() with a smart pointer
3085 class, thus making it possible to process discontinuous strings.
3086
3087 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3088 much the same function. They were added by different people who were trying
3089 to make PCRE easy to compile on non-Unix systems. It has been suggested
3090 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3091 for compiling on Windows systems. I have therefore replaced it with
3092 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3093 defaults to "extern" for C or "extern C" for C++, which works fine on
3094 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3095 SCOPE with something explicit in config.h. In addition:
3096
3097 (a) pcreposix.h still had just "extern" instead of either of these macros;
3098 I have replaced it with PCRE_DATA_SCOPE.
3099
3100 (b) Functions such as _pcre_xclass(), which are internal to the library,
3101 but external in the C sense, all had PCRE_EXPORT in their definitions.
3102 This is apparently wrong for the Windows case, so I have removed it.
3103 (It makes no difference on Unix-like systems.)
3104
3105 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3106 of recursive calls to match(). This is different to MATCH_LIMIT because
3107 that limits the total number of calls to match(), not all of which increase
3108 the depth of recursion. Limiting the recursion depth limits the amount of
3109 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3110 when PCRE is compiled, and changed at run time. A patch from Google adds
3111 this functionality to the C++ interface.
3112
3113 18. Changes to the handling of Unicode character properties:
3114
3115 (a) Updated the table to Unicode 4.1.0.
3116
3117 (b) Recognize characters that are not in the table as "Cn" (undefined).
3118
3119 (c) I revised the way the table is implemented to a much improved format
3120 which includes recognition of ranges. It now supports the ranges that
3121 are defined in UnicodeData.txt, and it also amalgamates other
3122 characters into ranges. This has reduced the number of entries in the
3123 table from around 16,000 to around 3,000, thus reducing its size
3124 considerably. I realized I did not need to use a tree structure after
3125 all - a binary chop search is just as efficient. Having reduced the
3126 number of entries, I extended their size from 6 bytes to 8 bytes to
3127 allow for more data.
3128
3129 (d) Added support for Unicode script names via properties such as \p{Han}.
3130
3131 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3132 matching that character.
3133
3134 20. When matching a repeated Unicode property with a minimum greater than zero,
3135 (for example \pL{2,}), PCRE could look past the end of the subject if it
3136 reached it while seeking the minimum number of characters. This could
3137 happen only if some of the characters were more than one byte long, because
3138 there is a check for at least the minimum number of bytes.
3139
3140 21. Refactored the implementation of \p and \P so as to be more general, to
3141 allow for more different types of property in future. This has changed the
3142 compiled form incompatibly. Anybody with saved compiled patterns that use
3143 \p or \P will have to recompile them.
3144
3145 22. Added "Any" and "L&" to the supported property types.
3146
3147 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3148 but give a compile time error if the value is greater than 0xff.
3149
3150 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3151 accidentally not being installed or uninstalled.
3152
3153 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3154 made were to insert the current release number. This seemed silly, because
3155 it made things harder for people building PCRE on systems that don't run
3156 "configure". I have turned pcre.h into a distributed file, no longer built
3157 by "configure", with the version identification directly included. There is
3158 no longer a pcre.h.in file.
3159
3160 However, this change necessitated a change to the pcre-config script as
3161 well. It is built from pcre-config.in, and one of the substitutions was the
3162 release number. I have updated configure.ac so that ./configure now finds
3163 the release number by grepping pcre.h.
3164
3165 26. Added the ability to run the tests under valgrind.
3166
3167
3168 Version 6.4 05-Sep-05
3169 ---------------------
3170
3171 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3172 "--" to be printed when multiple files were scanned, even when none of the
3173 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3174 consider it to be a bug, and have restored the previous behaviour.
3175
3176 2. A couple of code tidies to get rid of compiler warnings.
3177
3178 3. The pcretest program used to cheat by referring to symbols in the library
3179 whose names begin with _pcre_. These are internal symbols that are not
3180 really supposed to be visible externally, and in some environments it is
3181 possible to suppress them. The cheating is now confined to including
3182 certain files from the library's source, which is a bit cleaner.
3183
3184 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3185 file's purpose clearer.
3186
3187 5. Reorganized pcre_ucp_findchar().
3188
3189
3190 Version 6.3 15-Aug-05
3191 ---------------------
3192
3193 1. The file libpcre.pc.in did not have general read permission in the tarball.
3194
3195 2. There were some problems when building without C++ support:
3196
3197 (a) If C++ support was not built, "make install" and "make test" still
3198 tried to test it.
3199
3200 (b) There were problems when the value of CXX was explicitly set. Some
3201 changes have been made to try to fix these, and ...
3202
3203 (c) --disable-cpp can now be used to explicitly disable C++ support.
3204
3205 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3206 backslash in a target when C++ was disabled. This confuses some
3207 versions of "make", apparently. Using an intermediate variable solves
3208 this. (Same for CPP_LOBJ.)
3209
3210 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3211 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3212 necessary on certain architectures.
3213
3214 4. Added a setting of -export-symbols-regex to the link command to remove
3215 those symbols that are exported in the C sense, but actually are local
3216 within the library, and not documented. Their names all begin with
3217 "_pcre_". This is not a perfect job, because (a) we have to except some
3218 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3219 available (and never for static libraries). I have made a note to try to
3220 find a way round (a) in the future.
3221
3222
3223 Version 6.2 01-Aug-05
3224 ---------------------
3225
3226 1. There was no test for integer overflow of quantifier values. A construction
3227 such as {1111111111111111} would give undefined results. What is worse, if
3228 a minimum quantifier for a parenthesized subpattern overflowed and became
3229 negative, the calculation of the memory size went wrong. This could have
3230 led to memory overwriting.
3231
3232 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3233
3234 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3235 operating environments where this matters.
3236
3237 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3238 PCRE options from within the C++ wrapper.
3239
3240 5. Named capturing subpatterns were not being correctly counted when a pattern
3241 was compiled. This caused two problems: (a) If there were more than 100
3242 such subpatterns, the calculation of the memory needed for the whole
3243 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3244 back references of the form \12, where the number was greater than 9, were
3245 not recognized as back references, even though there were sufficient
3246 previous subpatterns.
3247
3248 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3249 versions of gcc, e.g. 2.95.4.
3250
3251
3252 Version 6.1 21-Jun-05
3253 ---------------------
3254
3255 1. There was one reference to the variable "posix" in pcretest.c that was not
3256 surrounded by "#if !defined NOPOSIX".
3257
3258 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3259 the cross-check on the old pcre_info() function, for the benefit of the
3260 cut-down version of PCRE that is currently imported into Exim.
3261
3262 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3263 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3264 patterns that start (?i) but I don't think that matters. The use of (?i) is
3265 just an example; this all applies to the other options as well.
3266
3267 4. Since libtool seems to echo the compile commands it is issuing, the output
3268 from "make" can be reduced a bit by putting "@" in front of each libtool
3269 compile command.
3270
3271 5. Patch from the folks at Google for configure.in to be a bit more thorough
3272 in checking for a suitable C++ installation before trying to compile the
3273 C++ stuff. This should fix a reported problem when a compiler was present,
3274 but no suitable headers.
3275
3276 6. The man pages all had just "PCRE" as their title. I have changed them to
3277 be the relevant file name. I have also arranged that these names are
3278 retained in the file doc/pcre.txt, which is a concatenation in text format
3279 of all the man pages except the little individual ones for each function.
3280
3281 7. The NON-UNIX-USE file had not been updated for the different set of source
3282 files that come with release 6. I also added a few comments about the C++
3283 wrapper.
3284
3285
3286 Version 6.0 07-Jun-05
3287 ---------------------
3288
3289 1. Some minor internal re-organization to help with my DFA experiments.
3290
3291 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3292 didn't matter for the library itself when fully configured, but did matter
3293 when compiling without UCP support, or within Exim, where the ucp files are
3294 not imported.
3295
3296 3. Refactoring of the library code to split up the various functions into
3297 different source modules. The addition of the new DFA matching code (see
3298 below) to a single monolithic source would have made it really too
3299 unwieldy, quite apart from causing all the code to be include in a
3300 statically linked application, when only some functions are used. This is
3301 relevant even without the DFA addition now that patterns can be compiled in
3302 one application and matched in another.
3303
3304 The downside of splitting up is that there have to be some external
3305 functions and data tables that are used internally in different modules of
3306 the library but which are not part of the API. These have all had their
3307 names changed to start with "_pcre_" so that they are unlikely to clash
3308 with other external names.
3309
3310 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3311 a different (DFA) algorithm. Although it is slower than the original
3312 function, it does have some advantages for certain types of matching
3313 problem.
3314
3315 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3316 including restarting after a partial match.
3317
3318 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3319 defined when compiling for Windows was sent to me. I have put it into the
3320 code, though I have no means of testing or verifying it.
3321
3322 7. Added the pcre_refcount() auxiliary function.
3323
3324 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3325 match before or at the first newline in the subject string. In pcretest,
3326 the /f option on a pattern can be used to set this.
3327
3328 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3329 would behave wrongly. This has been present in PCRE since release 4.0.
3330
3331 10. A number of changes to the pcregrep command:
3332
3333 (a) Refactored how -x works; insert ^(...)$ instead of setting
3334 PCRE_ANCHORED and checking the length, in preparation for adding
3335 something similar for -w.
3336
3337 (b) Added the -w (match as a word) option.
3338
3339 (c) Refactored the way lines are read and buffered so as to have more
3340 than one at a time available.
3341
3342 (d) Implemented a pcregrep test script.
3343
3344 (e) Added the -M (multiline match) option. This allows patterns to match
3345 over several lines of the subject. The buffering ensures that at least
3346 8K, or the rest of the document (whichever is the shorter) is available
3347 for matching (and similarly the previous 8K for lookbehind assertions).
3348
3349 (f) Changed the --help output so that it now says
3350
3351 -w, --word-regex(p)
3352
3353 instead of two lines, one with "regex" and the other with "regexp"
3354 because that confused at least one person since the short forms are the
3355 same. (This required a bit of code, as the output is generated
3356 automatically from a table. It wasn't just a text change.)
3357
3358 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3359 option but starts with a hyphen. Could be a pattern or a path name
3360 starting with a hyphen, for instance.
3361
3362 (h) "-" can be given as a file name to represent stdin.
3363
3364 (i) When file names are being printed, "(standard input)" is used for
3365 the standard input, for compatibility with GNU grep. Previously
3366 "<stdin>" was used.
3367
3368 (j) The option --label=xxx can be used to supply a name to be used for
3369 stdin when file names are being printed. There is no short form.
3370
3371 (k) Re-factored the options decoding logic because we are going to add
3372 two more options that take data. Such options can now be given in four
3373 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3374
3375 (l) Added the -A, -B, and -C options for requesting that lines of context
3376 around matches be printed.
3377
3378 (m) Added the -L option to print the names of files that do not contain
3379 any matching lines, that is, the complement of -l.
3380
3381 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3382 continue to scan other files.
3383
3384 (o) The -s option was incorrectly implemented. For compatibility with other
3385 greps, it now suppresses the error message for a non-existent or non-
3386 accessible file (but not the return code). There is a new option called
3387 -q that suppresses the output of matching lines, which was what -s was
3388 previously doing.
3389
3390 (p) Added --include and --exclude options to specify files for inclusion
3391 and exclusion when recursing.
3392
3393 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3394 Hopefully, it now does.
3395
3396 12. Missing cast in pcre_study().
3397
3398 13. Added an "uninstall" target to the makefile.
3399
3400 14. Replaced "extern" in the function prototypes in Makefile.in with
3401 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3402 world, but is set differently for Windows.
3403
3404 15. Added a second compiling function called pcre_compile2(). The only
3405 difference is that it has an extra argument, which is a pointer to an
3406 integer error code. When there is a compile-time failure, this is set
3407 non-zero, in addition to the error test pointer being set to point to an
3408 error message. The new argument may be NULL if no error number is required
3409 (but then you may as well call pcre_compile(), which is now just a
3410 wrapper). This facility is provided because some applications need a
3411 numeric error indication, but it has also enabled me to tidy up the way
3412 compile-time errors are handled in the POSIX wrapper.
3413
3414 16. Added VPATH=.libs to the makefile; this should help when building with one
3415 prefix path and installing with another. (Or so I'm told by someone who
3416 knows more about this stuff than I do.)
3417
3418 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3419 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3420 match everything, including newlines. This is not POSIX-compatible, but
3421 somebody wanted the feature. From pcretest it can be activated by using
3422 both the P and the s flags.
3423
3424 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3425
3426 19. libpcre.pc was being incorrectly installed as executable.
3427
3428 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3429 it now also looks for '\r' so that it will work unmodified on Windows.
3430
3431 21. Added Google's contributed C++ wrapper to the distribution.
3432
3433 22. Added some untidy missing memory free() calls in pcretest, to keep
3434 Electric Fence happy when testing.
3435
3436
3437
3438 Version 5.0 13-Sep-04
3439 ---------------------
3440
3441 1. Internal change: literal characters are no longer packed up into items
3442 containing multiple characters in a single byte-string. Each character
3443 is now matched using a separate opcode. However, there may be more than one
3444 byte in the character in UTF-8 mode.
3445
3446 2. The pcre_callout_block structure has two new fields: pattern_position and
3447 next_item_length. These contain the offset in the pattern to the next match
3448 item, and its length, respectively.
3449
3450 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3451 insertion of callouts before each pattern item. Added the /C option to
3452 pcretest to make use of this.
3453
3454 4. On the advice of a Windows user, the lines
3455
3456 #if defined(_WIN32) || defined(WIN32)
3457 _setmode( _fileno( stdout ), 0x8000 );
3458 #endif /* defined(_WIN32) || defined(WIN32) */
3459
3460 have been added to the source of pcretest. This apparently does useful
3461 magic in relation to line terminators.
3462
3463 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3464 for the benefit of those environments where the "b" makes a difference.
3465
3466 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3467 to know about it. I have put a hack into configure.in that adds in code
3468 to set GCC=yes if CC=icc. This seems to end up at a point in the
3469 generated configure script that is early enough to affect the setting of
3470 compiler options, which is what is needed, but I have no means of testing
3471 whether it really works. (The user who reported this had patched the
3472 generated configure script, which of course I cannot do.)
3473
3474 LATER: After change 22 below (new libtool files), the configure script
3475 seems to know about icc (and also ecc). Therefore, I have commented out
3476 this hack in configure.in.
3477
3478 7. Added support for pkg-config (2 patches were sent in).
3479
3480 8. Negated POSIX character classes that used a combination of internal tables
3481 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3482 [[:^ascii]]. Typically, they would match almost any characters. The other
3483 POSIX classes were not broken in this way.
3484
3485 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3486 to find the match, as PCRE was deluded into thinking that the match had to
3487 start at the start point or following a newline. The same bug applied to
3488 patterns with negative forward assertions or any backward assertions
3489 preceding ".*" at the start, unless the pattern required a fixed first
3490 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3491
3492 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3493 starting at the last subject character, bytes beyond the end of the subject
3494 string were read.
3495
3496 11. Renamed the variable "class" as "classbits" to make life easier for C++
3497 users. (Previously there was a macro definition, but it apparently wasn't
3498 enough.)
3499
3500 12. Added the new field "tables" to the extra data so that tables can be passed
3501 in at exec time, or the internal tables can be re-selected. This allows
3502 a compiled regex to be saved and re-used at a later time by a different
3503 program that might have everything at different addresses.
3504
3505 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3506 -R library as well as a -L library.
3507
3508 14. The debugging options of pcretest (-d on the command line or D on a
3509 pattern) showed incorrect output for anything following an extended class
3510 that contained multibyte characters and which was followed by a quantifier.
3511
3512 15. Added optional support for general category Unicode character properties
3513 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3514 support. It adds about 90K to the size of the library. The meanings of the
3515 inbuilt class escapes such as \d and \s have NOT been changed.
3516
3517 16. Updated pcredemo.c to include calls to free() to release the memory for the
3518 compiled pattern.
3519
3520 17. The generated file chartables.c was being created in the source directory
3521 instead of in the building directory. This caused the build to fail if the
3522 source directory was different from the building directory, and was
3523 read-only.
3524
3525 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3526 file. No doubt somebody will tell me if they don't make sense... Also added
3527 Dan Mooney's comments about building on OpenVMS.
3528
3529 19. Added support for partial matching via the PCRE_PARTIAL option for
3530 pcre_exec() and the \P data escape in pcretest.
3531
3532 20. Extended pcretest with 3 new pattern features:
3533
3534 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3535 write the compiled pattern to the file whose name is "rest-of-line".
3536 This is a straight binary dump of the data, with the saved pointer to
3537 the character tables forced to be NULL. The study data, if any, is
3538 written too. After writing, pcretest reads a new pattern.
3539
3540 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3541 compiled pattern from the given file. There must not be any
3542 occurrences of "<" in the file name (pretty unlikely); if there are,
3543 pcretest will instead treat the initial "<" as a pattern delimiter.
3544 After reading in the pattern, pcretest goes on to read data lines as
3545 usual.
3546
3547 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3548 and 16-bit fields in a compiled pattern, to simulate a pattern that
3549 was compiled on a host of opposite endianness.
3550
3551 21. The pcre-exec() function can now cope with patterns that were compiled on
3552 hosts of opposite endianness, with this restriction:
3553
3554 As for any compiled expression that is saved and used later, the tables
3555 pointer field cannot be preserved; the extra_data field in the arguments
3556 to pcre_exec() should be used to pass in a tables address if a value
3557 other than the default internal tables were used at compile time.
3558
3559 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3560 now diagnosed as an error. Previously, most of the time, a negative number
3561 would have been treated as zero, but if in addition "ovector" was passed as
3562 NULL, a crash could occur.
3563
3564 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3565 new versions from the libtool 1.5 distribution (the last one is a copy of
3566 a file called libtool.m4). This seems to have fixed the need to patch
3567 "configure" to support Darwin 1.3 (which I used to do). However, I still
3568 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3569 workstation).
3570
3571 24. Changed the PCRE licence to be the more standard "BSD" licence.
3572
3573
3574 Version 4.5 01-Dec-03
3575 ---------------------
3576
3577 1. There has been some re-arrangement of the code for the match() function so
3578 that it can be compiled in a version that does not call itself recursively.
3579 Instead, it keeps those local variables that need separate instances for
3580 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3581 needs to "recurse". Keeping track of where control must go is done by means
3582 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3583 hide most of the details from the main code, and operates only if
3584 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3585 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3586 operating.
3587
3588 To make it easier for callers to provide specially tailored get/free
3589 functions for this usage, two new functions, pcre_stack_malloc, and
3590 pcre_stack_free, are used. They are always called in strict stacking order,
3591 and the size of block requested is always the same.
3592
3593 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3594 PCRE has been compiled to use the stack or the heap for recursion. The
3595 -C option of pcretest uses this to show which version is compiled.
3596
3597 A new data escape \S, is added to pcretest; it causes the amounts of store
3598 obtained and freed by both kinds of malloc/free at match time to be added
3599 to the output.
3600
3601 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3602 what's available on my current Linux desktop machine.
3603
3604 3. When matching a UTF-8 string, the test for a valid string at the start has
3605 been extended. If start_offset is not zero, PCRE now checks that it points
3606 to a byte that is the start of a UTF-8 character. If not, it returns
3607 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3608 this is necessary because there may be backward assertions in the pattern.
3609 When matching the same subject several times, it may save resources to use
3610 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3611
3612 4. The code for checking the validity of UTF-8 strings has been tightened so
3613 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3614 containing "overlong sequences".
3615
3616 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3617 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3618 should have been "&", but it just so happened that all the cases this let
3619 through by mistake were picked up later in the function.
3620
3621 6. I had used a variable called "isblank" - this is a C99 function, causing
3622 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3623
3624 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3625 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3626 so the version is shown; (c) stop "make test" echoing "./RunTest".
3627
3628 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3629
3630 9. The prototype for memmove() for systems that don't have it was using
3631 size_t, but the inclusion of the header that defines size_t was later. I've
3632 moved the #includes for the C headers earlier to avoid this.
3633
3634 10. Added some adjustments to the code to make it easier to compiler on certain
3635 special systems:
3636
3637 (a) Some "const" qualifiers were missing.
3638 (b) Added the macro EXPORT before all exported functions; by default this
3639 is defined to be empty.
3640 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3641 that it reads its output file name as an argument instead of writing
3642 to the standard output and assuming this can be redirected.
3643
3644 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3645 class containing characters with values greater than 255, PCRE compilation
3646 went into a loop.
3647
3648 12. A recursive reference to a subpattern that was within another subpattern
3649 that had a minimum quantifier of zero caused PCRE to crash. For example,
3650 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3651 recursion. If the recursively-called subpattern itself had a zero repeat,
3652 that was OK.
3653
3654 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3655 buffer into which it was copied (for escape processing) was still set at
3656 1024, so long lines caused crashes.
3657
3658 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3659 "internal error: code overflow...". This applied to any character class
3660 that was followed by a possessive quantifier.
3661
3662 15. Modified the Makefile to add libpcre.la as a prerequisite for
3663 libpcreposix.la because I was told this is needed for a parallel build to
3664 work.
3665
3666 16. If a pattern that contained .* following optional items at the start was
3667 studied, the wrong optimizing data was generated, leading to matching
3668 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3669 matching string must start with a or b or c. The correct conclusion for
3670 this pattern is that a match can start with any character.
3671
3672
3673 Version 4.4 13-Aug-03
3674 ---------------------
3675
3676 1. In UTF-8 mode, a character class containing characters with values between
3677 127 and 255 was not handled correctly if the compiled pattern was studied.
3678 In fixing this, I have also improved the studying algorithm for such
3679 classes (slightly).
3680
3681 2. Three internal functions had redundant arguments passed to them. Removal
3682 might give a very teeny performance improvement.
3683
3684 3. Documentation bug: the value of the capture_top field in a callout is *one
3685 more than* the number of the hightest numbered captured substring.
3686
3687 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3688 in incorrectly linking with a previously installed version. They now link
3689 explicitly with libpcre.la.
3690
3691 5. configure.in no longer needs to recognize Cygwin specially.
3692
3693 6. A problem in pcre.in for Windows platforms is fixed.
3694
3695 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3696 pcretest, it used to include the size of the study block as part of its
3697 output. Unfortunately, the structure contains a field that has a different
3698 size on different hardware architectures. This meant that the tests that
3699 showed this size failed. As the block is currently always of a fixed size,
3700 this information isn't actually particularly useful in pcretest output, so
3701 I have just removed it.
3702
3703 8. Three pre-processor statements accidentally did not start in column 1.
3704 Sadly, there are *still* compilers around that complain, even though
3705 standard C has not required this for well over a decade. Sigh.
3706
3707 9. In pcretest, the code for checking callouts passed small integers in the
3708 callout_data field, which is a void * field. However, some picky compilers
3709 complained about the casts involved for this on 64-bit systems. Now
3710 pcretest passes the address of the small integer instead, which should get
3711 rid of the warnings.
3712
3713 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3714 both compile and run time, and gives an error if an invalid UTF-8 sequence
3715 is found. There is a option for disabling this check in cases where the
3716 string is known to be correct and/or the maximum performance is wanted.
3717
3718 11. In response to a bug report, I changed one line in Makefile.in from
3719
3720 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3721 to
3722 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3723
3724 to look similar to other lines, but I have no way of telling whether this
3725 is the right thing to do, as I do not use Windows. No doubt I'll get told
3726 if it's wrong...
3727
3728
3729 Version 4.3 21-May-03
3730 ---------------------
3731
3732 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3733 Makefile.
3734
3735 2. Some refactoring to improve the quality of the code:
3736
3737 (i) The utf8_table... variables are now declared "const".
3738
3739 (ii) The code for \cx, which used the "case flipping" table to upper case
3740 lower case letters, now just substracts 32. This is ASCII-specific,
3741 but the whole concept of \cx is ASCII-specific, so it seems
3742 reasonable.
3743
3744 (iii) PCRE was using its character types table to recognize decimal and
3745 hexadecimal digits in the pattern. This is silly, because it handles
3746 only 0-9, a-f, and A-F, but the character types table is locale-
3747 specific, which means strange things might happen. A private
3748 table is now used for this - though it costs 256 bytes, a table is
3749 much faster than multiple explicit tests. Of course, the standard
3750 character types table is still used for matching digits in subject
3751 strings against \d.
3752
3753 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3754 ending in _t are). So I've renamed it as ESC_tee.
3755
3756 3. The first argument for regexec() in the POSIX wrapper should have been
3757 defined as "const".
3758
3759 4. Changed pcretest to use malloc() for its buffers so that they can be
3760 Electric Fenced for debugging.
3761
3762 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3763 to read one or more bytes before the start of the subject string. Often this
3764 had no effect on PCRE's behaviour, but in some circumstances it could
3765 provoke a segmentation fault.
3766
3767 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3768 to try to read one or more bytes before the start of the subject string.
3769
3770 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3771 UTF-8 support could misbehave in various ways if the subject string
3772 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3773 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3774 back over UTF-8 characters.)
3775
3776
3777 Version 4.2 14-Apr-03
3778 ---------------------
3779
3780 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3781
3782 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3783 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3784 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3785 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3786 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3787 and BUILD_EXEEXT
3788 Note: automatic setting of the BUILD variables is not yet working
3789 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3790 compile-time but not at link-time
3791 [LINK]: use for linking executables only
3792 make different versions for Windows and non-Windows
3793 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3794 libraries
3795 [LINK_FOR_BUILD]: new variable
3796 [OBJEXT]: use throughout
3797 [EXEEXT]: use throughout
3798 <winshared>: new target
3799 <wininstall>: new target
3800 <dftables.o>: use native compiler
3801 <dftables>: use native linker
3802 <install>: handle Windows platform correctly
3803 <clean>: ditto
3804 <check>: ditto
3805 copy DLL to top builddir before testing
3806
3807 As part of these changes, -no-undefined was removed again. This was reported
3808 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3809 in any case.
3810
3811 3. Some tidies to get rid of compiler warnings:
3812
3813 . In the match_data structure, match_limit was an unsigned long int, whereas
3814 match_call_count was an int. I've made them both unsigned long ints.
3815
3816 . In pcretest the fact that a const uschar * doesn't automatically cast to
3817 a void * provoked a warning.
3818
3819 . Turning on some more compiler warnings threw up some "shadow" variables
3820 and a few more missing casts.
3821
3822 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3823 option, a class that contained a single character with a value between 128
3824 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3825
3826 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3827 option, a class that contained several characters, but with at least one
3828 whose value was between 128 and 255 caused PCRE to crash.
3829
3830
3831 Version 4.1 12-Mar-03
3832 ---------------------
3833
3834 1. Compiling with gcc -pedantic found a couple of places where casts were
3835 needed, and a string in dftables.c that was longer than standard compilers are
3836 required to support.
3837
3838 2. Compiling with Sun's compiler found a few more places where the code could
3839 be tidied up in order to avoid warnings.
3840
3841 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3842 first of these names is deprecated in the latest Autoconf in favour of the name
3843 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3844 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3845 analogy I have changed it to CFLAGS_FOR_BUILD.
3846
3847 4. Added -no-undefined to the linking command in the Makefile, because this is
3848 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3849 linking step for the pcreposix library.
3850
3851 5. PCRE was failing to diagnose the case of two named groups with the same
3852 name.
3853
3854 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3855 literal character that is needed in the subject for a match, and scans along to
3856 ensure that it is present before embarking on the full matching process. This
3857 saves time in cases of nested unlimited repeats that are never going to match.
3858 Problem: the scan can take a lot of time if the subject is very long (e.g.
3859 megabytes), thus penalizing straightforward matches. It is now done only if the
3860 amount of subject to be scanned is less than 1000 bytes.
3861
3862 7. A lesser problem with the same optimization is that it was recording the
3863 first character of an anchored pattern as "needed", thus provoking a search
3864 right along the subject, even when the first match of the pattern was going to
3865 fail. The "needed" character is now not set for anchored patterns, unless it
3866 follows something in the pattern that is of non-fixed length. Thus, it still
3867 fulfils its original purpose of finding quick non-matches in cases of nested
3868 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3869
3870
3871 Version 4.0 17-Feb-03
3872 ---------------------
3873
3874 1. If a comment in an extended regex that started immediately after a meta-item
3875 extended to the end of string, PCRE compiled incorrect data. This could lead to
3876 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3877
3878 2. Moved to autoconf 2.53 and libtool 1.4.2.
3879
3880 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3881 the special perltest8 script is no longer needed - all the tests can be run
3882 from a single perltest script.
3883
3884 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3885 by \s. It has now been removed in PCRE. This means it isn't recognized as
3886 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3887 class [:space:] *does* include VT, thereby creating a mess.
3888
3889 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3890 space and tab.
3891
3892 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3893 its new features into the main test script, reducing the number of scripts.
3894
3895 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3896 were backward compatible, and made the (?i) apply to the whole pattern, as if
3897 /i were given. Now it behaves more logically, and applies the option setting
3898 only to what follows. PCRE has been changed to follow suit. However, if it
3899 finds options settings right at the start of the pattern, it extracts them into
3900 the global options, as before. Thus, they show up in the info data.
3901
3902 8. Added support for the \Q...\E escape sequence. Characters in between are
3903 treated as literals. This is slightly different from Perl in that $ and @ are
3904 also handled as literals inside the quotes. In Perl, they will cause variable
3905 interpolation. Note the following examples:
3906
3907 Pattern PCRE matches Perl matches
3908
3909 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3910 \Qabc\$xyz\E abc\$xyz abc\$xyz
3911 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3912
3913 For compatibility with Perl, \Q...\E sequences are recognized inside character
3914 classes as well as outside them.
3915
3916 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3917 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3918 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3919 signed/unsigned warnings.
3920
3921 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3922 option for pcretest, so I've replaced it by a simple function that does just
3923 that job.
3924
3925 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3926 "pcregrep -".
3927
3928 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3929 Java package. This provides some syntactic sugar for simple cases of what my
3930 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3931 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3932 item, you can use this simplified notation. Note that only makes sense with
3933 greedy quantifiers. Consequently, the use of the possessive quantifier forces
3934 greediness, whatever the setting of the PCRE_UNGREEDY option.
3935
3936 13. A change of greediness default within a pattern was not taking effect at
3937 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3938 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
3939 was abstracted outside.
3940
3941 14. PCRE now supports the \G assertion. It is true when the current matching
3942 position is at the start point of the match. This differs from \A when the
3943 starting offset is non-zero. Used with the /g option of pcretest (or similar
3944 code), it works in the same way as it does for Perl's /g option. If all
3945 alternatives of a regex begin with \G, the expression is anchored to the start
3946 match position, and the "anchored" flag is set in the compiled expression.
3947
3948 15. Some bugs concerning the handling of certain option changes within patterns
3949 have been fixed. These applied to options other than (?ims). For example,
3950 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
3951 been the other way round. Some of this was related to change 7 above.
3952
3953 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
3954 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
3955 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
3956 POSIX classes only within a class (e.g. /[[:alpha:]]/).
3957
3958 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
3959 mode. Unlike ".", it always matches newline, whatever the setting of
3960 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
3961 assertions. Perl allows it, but it doesn't (in general) work because it can't
3962 calculate the length of the lookbehind. At least, that's the case for Perl
3963 5.8.0 - I've been told they are going to document that it doesn't work in
3964 future.
3965
3966 18. Added an error diagnosis for escapes that PCRE does not support: these are
3967 \L, \l, \N, \P, \p, \U, \u, and \X.
3968
3969 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
3970 reading past the end of the pattern in cases such as /[abcd/.
3971
3972 20. PCRE was getting more memory than necessary for patterns with classes that
3973 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
3974
3975 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
3976 compiling PCRE for use with Virtual Pascal.
3977
3978 22. Small fix to the Makefile to make it work properly if the build is done
3979 outside the source tree.
3980
3981 23. Added a new extension: a condition to go with recursion. If a conditional
3982 subpattern starts with (?(R) the "true" branch is used if recursion has
3983 happened, whereas the "false" branch is used only at the top level.
3984
3985 24. When there was a very long string of literal characters (over 255 bytes
3986 without UTF support, over 250 bytes with UTF support), the computation of how
3987 much memory was required could be incorrect, leading to segfaults or other
3988 strange effects.
3989
3990 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
3991 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
3992 there was a subsequent back reference to those brackets. This meant that, for
3993 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
3994 possible to check for precisely this case. All we can do is abandon the
3995 optimization if .* occurs inside capturing brackets when there are any back
3996 references whatsoever. (See below for a better fix that came later.)
3997
3998 26. The handling of the optimization for finding the first character of a
3999 non-anchored pattern, and for finding a character that is required later in the
4000 match were failing in some cases. This didn't break the matching; it just
4001 failed to optimize when it could. The way this is done has been re-implemented.
4002
4003 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4004
4005 28. Added a new feature that provides some of the functionality that Perl
4006 provides with (?{...}). The facility is termed a "callout". The way it is done
4007 in PCRE is for the caller to provide an optional function, by setting
4008 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4009 global variable. By default it is unset, which disables all calling out. To get
4010 the function called, the regex must include (?C) at appropriate points. This
4011 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4012 This provides a means of identifying different callout points. When PCRE
4013 reaches such a point in the regex, if pcre_callout has been set, the external
4014 function is called. It is provided with data in a structure called
4015 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4016 matching continues; if it returns a non-zero value, the match at the current
4017 point fails. However, backtracking will occur if possible. [This was changed
4018 later and other features added - see item 49 below.]
4019
4020 29. pcretest is upgraded to test the callout functionality. It provides a
4021 callout function that displays information. By default, it shows the start of
4022 the match and the current position in the text. There are some new data escapes
4023 to vary what happens:
4024
4025 \C+ in addition, show current contents of captured substrings
4026 \C- do not supply a callout function
4027 \C!n return 1 when callout number n is reached
4028 \C!n!m return 1 when callout number n is reached for the mth time
4029
4030 30. If pcregrep was called with the -l option and just a single file name, it
4031 output "<stdin>" if a match was found, instead of the file name.
4032
4033 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4034 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4035 pcre_exec(). This saves a malloc/free per call. The default value of
4036 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4037 when configuring.
4038
4039 32. The default maximum size of a compiled pattern is 64K. There have been a
4040 few cases of people hitting this limit. The code now uses macros to handle the
4041 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4042 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4043 configuring. Tests 2 and 5 work only with 2-byte links because they output
4044 debugging information about compiled patterns.
4045
4046 33. Internal code re-arrangements:
4047
4048 (a) Moved the debugging function for printing out a compiled regex into
4049 its own source file (printint.c) and used #include to pull it into
4050 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4051 separate copies.
4052
4053 (b) Defined the list of op-code names for debugging as a macro in
4054 internal.h so that it is next to the definition of the opcodes.
4055
4056 (c) Defined a table of op-code lengths for simpler skipping along compiled
4057 code. This is again a macro in internal.h so that it is next to the
4058 definition of the opcodes.
4059
4060 34. Added support for recursive calls to individual subpatterns, along the
4061 lines of Robin Houston's patch (but implemented somewhat differently).
4062
4063 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4064 allow it to read and process whole directories in Win32. This code was
4065 contributed by Lionel Fourquaux; it has not been tested by me.
4066
4067 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4068 used to name a group. Names consist of alphanumerics and underscores, and must
4069 be unique. Back references use the syntax (?P=name) and recursive calls use
4070 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4071 numbers. The function pcre_fullinfo() can be used after compilation to extract
4072 a name/number map. There are three relevant calls:
4073
4074 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4075 PCRE_INFO_NAMECOUNT yields the number of entries
4076 PCRE_INFO_NAMETABLE yields a pointer to the map.
4077
4078 The map is a vector of fixed-size entries. The size of each entry depends on
4079 the length of the longest name used. The first two bytes of each entry are the
4080 group number, most significant byte first. There follows the corresponding
4081 name, zero terminated. The names are in alphabetical order.
4082
4083 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4084 case instead of 255. Making it the same both with and without UTF-8 support
4085 means that the same test output works with both.
4086
4087 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4088 calling malloc() with a zero argument.
4089
4090 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4091 optimization. I've improved things by keeping a bitmap of backreferences with
4092 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4093 fact referenced, the optimization can be applied. It is unlikely that a
4094 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4095 the match to follow \n) will appear inside brackets with a number greater than
4096 31, but if it does, any back reference > 31 suppresses the optimization.
4097
4098 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4099 of disabling numbered capturing parentheses. Any opening parenthesis that is
4100 not followed by ? behaves as if it were followed by ?: but named parentheses
4101 can still be used for capturing (and they will acquire numbers in the usual
4102 way).
4103
4104 41. Redesigned the return codes from the match() function into yes/no/error so
4105 that errors can be passed back from deep inside the nested calls. A malloc
4106 failure while inside a recursive subpattern call now causes the
4107 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4108
4109 42. It is now possible to set a limit on the number of times the match()
4110 function is called in a call to pcre_exec(). This facility makes it possible to
4111 limit the amount of recursion and backtracking, though not in a directly
4112 obvious way, because the match() function is used in a number of different
4113 circumstances. The count starts from zero for each position in the subject
4114 string (for non-anchored patterns). The default limit is, for compatibility, a
4115 large number, namely 10 000 000. You can change this in two ways:
4116
4117 (a) When configuring PCRE before making, you can use --with-match-limit=n
4118 to set a default value for the compiled library.
4119
4120 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4121 a different value is set. See 45 below.
4122
4123 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4124
4125 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4126 of things that can be changed at compile time. The first argument specifies
4127 what is wanted and the second points to where the information is to be placed.
4128 The current list of available information is:
4129
4130 PCRE_CONFIG_UTF8
4131
4132 The output is an integer that is set to one if UTF-8 support is available;
4133 otherwise it is set to zero.
4134
4135 PCRE_CONFIG_NEWLINE
4136
4137 The output is an integer that it set to the value of the code that is used for
4138 newline. It is either LF (10) or CR (13).
4139
4140 PCRE_CONFIG_LINK_SIZE
4141
4142 The output is an integer that contains the number of bytes used for internal
4143 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4144
4145 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4146
4147 The output is an integer that contains the threshold above which the POSIX
4148 interface uses malloc() for output vectors. See item 31 above.
4149
4150 PCRE_CONFIG_MATCH_LIMIT
4151
4152 The output is an unsigned integer that contains the default limit of the number
4153 of match() calls in a pcre_exec() execution. See 42 above.
4154
4155 44. pcretest has been upgraded by the addition of the -C option. This causes it
4156 to extract all the available output from the new pcre_config() function, and to
4157 output it. The program then exits immediately.
4158
4159 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4160 order to support additional features. One way would have been to define
4161 pcre_exec2() (for example) with extra arguments, but this would not have been
4162 extensible, and would also have required all calls to the original function to
4163 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4164 is used for passing in "extra" data from pcre_study().
4165
4166 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4167 contains the following fields:
4168
4169 flags a bitmap indicating which of the following fields are set
4170 study_data opaque data from pcre_study()
4171 match_limit a way of specifying a limit on match() calls for a specific
4172 call to pcre_exec()
4173 callout_data data for callouts (see 49 below)
4174
4175 The flag bits are also defined in pcre.h, and are
4176
4177 PCRE_EXTRA_STUDY_DATA
4178 PCRE_EXTRA_MATCH_LIMIT
4179 PCRE_EXTRA_CALLOUT_DATA
4180
4181 The pcre_study() function now returns one of these new pcre_extra blocks, with
4182 the actual study data pointed to by the study_data field, and the
4183 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4184 before. That is, this change is entirely upwards-compatible and requires no
4185 change to existing code.
4186
4187 If you want to pass in additional data to pcre_exec(), you can either place it
4188 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4189 block.
4190
4191 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4192 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4193 times with different match limits, until it finds the minimum value needed for
4194 pcre_exec() to complete. The value is then output. This can be instructive; for
4195 most simple matches the number is quite small, but for pathological cases it
4196 gets very large very quickly.
4197
4198 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4199 returns the size of the data block pointed to by the study_data field in a
4200 pcre_extra block, that is, the value that was passed as the argument to
4201 pcre_malloc() when PCRE was getting memory in which to place the information
4202 created by pcre_study(). The fourth argument should point to a size_t variable.
4203 pcretest has been extended so that this information is shown after a successful
4204 pcre_study() call when information about the compiled regex is being displayed.
4205
4206 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4207 because what follows is always an absolute path. (Later: it turns out that this
4208 is more than cosmetic for MinGW, because it doesn't like empty path
4209 components.)
4210
4211 49. Some changes have been made to the callout feature (see 28 above):
4212
4213 (i) A callout function now has three choices for what it returns:
4214
4215 0 => success, carry on matching
4216 > 0 => failure at this point, but backtrack if possible
4217 < 0 => serious error, return this value from pcre_exec()
4218
4219 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4220 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4221 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4222 use by callout functions. It will never be used by PCRE itself.
4223
4224 (ii) The pcre_extra structure (see 45 above) has a void * field called
4225 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4226 pcre_callout_block structure has a field of the same name. The contents of
4227 the field passed in the pcre_extra structure are passed to the callout
4228 function in the corresponding field in the callout block. This makes it
4229 easier to use the same callout-containing regex from multiple threads. For
4230 testing, the pcretest program has a new data escape
4231
4232 \C*n pass the number n (may be negative) as callout_data
4233
4234 If the callout function in pcretest receives a non-zero value as
4235 callout_data, it returns that value.
4236
4237 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4238 there were some redundant $(CFLAGS) in commands that are now specified as
4239 $(LINK), which already includes $(CFLAGS).
4240
4241 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4242 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4243 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4244 one-byte characters throughout. Note that case-insensitive matching applies
4245 only to characters whose values are less than 256. PCRE doesn't support the
4246 notion of cases for higher-valued characters.
4247
4248 (i) A character class whose characters are all within 0-255 is handled as
4249 a bit map, and the map is inverted for negative classes. Previously, a
4250 character > 255 always failed to match such a class; however it should
4251 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4252
4253 (ii) A negated character class with a single character < 255 is coded as
4254 "not this character" (OP_NOT). This wasn't working properly when the test
4255 character was multibyte, either singly or repeated.
4256
4257 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4258 mode, for example: \x{100}{2,3}.
4259
4260 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4261 singly or repeated) now correctly test multibyte characters. However,
4262 PCRE doesn't recognize any characters with values greater than 255 as
4263 digits, spaces, or word characters. Such characters always match \D, \S,
4264 and \W, and never match \d, \s, or \w.
4265
4266 (v) Classes may now contain characters and character ranges with values
4267 greater than 255. For example: [ab\x{100}-\x{400}].
4268
4269 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4270 PCRE in UTF-8 mode.
4271
4272 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4273 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4274 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4275 value.)
4276
4277 53. The single man page has become too large. I have therefore split it up into
4278 a number of separate man pages. These also give rise to individual HTML pages;
4279 these are now put in a separate directory, and there is an index.html page that
4280 lists them all. Some hyperlinking between the pages has been installed.
4281
4282 54. Added convenience functions for handling named capturing parentheses.
4283
4284 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4285 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4286 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4287 are faulted.
4288
4289 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4290 calling configure. These values are used when compiling the dftables.c program
4291 which is run to generate the source of the default character tables. They
4292 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4293 you will need to set these values.
4294
4295 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4296
4297
4298 Version 3.9 02-Jan-02
4299 ---------------------
4300
4301 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4302
4303 2. If --disable-static was given, the building process failed when trying to
4304 build pcretest and pcregrep. (For some reason it was using libtool to compile
4305 them, which is not right, as they aren't part of the library.)
4306
4307
4308 Version 3.8 18-Dec-01
4309 ---------------------
4310
4311 1. The experimental UTF-8 code was completely screwed up. It was packing the
4312 bytes in the wrong order. How dumb can you get?
4313
4314
4315 Version 3.7 29-Oct-01
4316 ---------------------
4317
4318 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4319 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4320 this didn't happen under Solaris 8, where I normally test things.
4321
4322 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4323 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4324 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4325 this fix an hour or so after the initial 3.7 release.)
4326
4327
4328 Version 3.6 23-Oct-01
4329 ---------------------
4330
4331 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4332 offsets passed as NULL with zero offset count.
4333
4334 2. The config.guess and config.sub files had not been updated when I moved to
4335 the latest autoconf.
4336
4337
4338 Version 3.5 15-Aug-01
4339 ---------------------
4340
4341 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4342 had been forgotten.
4343
4344 2. By using declared but undefined structures, we can avoid using "void"
4345 definitions in pcre.h while keeping the internal definitions of the structures
4346 private.
4347
4348 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4349 user point of view, this means that both static and shared libraries are built
4350 by default, but this can be individually controlled. More of the work of
4351 handling this static/shared cases is now inside libtool instead of PCRE's make
4352 file.
4353
4354 4. The pcretest utility is now installed along with pcregrep because it is
4355 useful for users (to test regexs) and by doing this, it automatically gets
4356 relinked by libtool. The documentation has been turned into a man page, so
4357 there are now .1, .txt, and .html versions in /doc.
4358
4359 5. Upgrades to pcregrep:
4360 (i) Added long-form option names like gnu grep.
4361 (ii) Added --help to list all options with an explanatory phrase.
4362 (iii) Added -r, --recursive to recurse into sub-directories.
4363 (iv) Added -f, --file to read patterns from a file.
4364
4365 6. pcre_exec() was referring to its "code" argument before testing that
4366 argument for NULL (and giving an error if it was NULL).
4367
4368 7. Upgraded Makefile.in to allow for compiling in a different directory from
4369 the source directory.
4370
4371 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4372 options bits, the pointer it was passed was to an int instead of to an unsigned
4373 long int. This mattered only on 64-bit systems.
4374
4375 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4376 generated) instead of pcre.in, which it its source. Also made the same change
4377 in several of the .c files.
4378
4379 10. A new release of gcc defines printf() as a macro, which broke pcretest
4380 because it had an ifdef in the middle of a string argument for printf(). Fixed
4381 by using separate calls to printf().
4382
4383 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4384 script, to force use of CR or LF instead of \n in the source. On non-Unix
4385 systems, the value can be set in config.h.
4386
4387 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4388 absolute limit. Changed the text of the error message to make this clear, and
4389 likewise updated the man page.
4390
4391 13. The limit of 99 on the number of capturing subpatterns has been removed.
4392 The new limit is 65535, which I hope will not be a "real" limit.
4393
4394
4395 Version 3.4 22-Aug-00
4396 ---------------------
4397
4398 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4399
4400 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4401
4402
4403 Version 3.3 01-Aug-00
4404 ---------------------
4405
4406 1. If an octal character was given, but the value was greater than \377, it
4407 was not getting masked to the least significant bits, as documented. This could
4408 lead to crashes in some systems.
4409
4410 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4411 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4412
4413 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4414 These just pass their arguments on to (pcre_free)(), but they are provided
4415 because some uses of PCRE bind it to non-C systems that can call its functions,
4416 but cannot call free() or pcre_free() directly.
4417
4418 4. Add "make test" as a synonym for "make check". Corrected some comments in
4419 the Makefile.
4420
4421 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4422 Makefile.
4423
4424 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4425 command called pgrep for grepping around the active processes.
4426
4427 7. Added the beginnings of support for UTF-8 character strings.
4428
4429 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4430 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4431 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4432 out for the ar command.)
4433
4434
4435 Version 3.2 12-May-00
4436 ---------------------
4437
4438 This is purely a bug fixing release.
4439
4440 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4441 of ZA. This was just one example of several cases that could provoke this bug,
4442 which was introduced by change 9 of version 2.00. The code for breaking
4443 infinite loops after an iteration that matches an empty string was't working
4444 correctly.
4445
4446 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4447 when matched against abbab (for example). After matching an empty string, it
4448 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4449 caused it to match further down the string than it should.
4450
4451 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4452 was there because it doesn't seem to be needed, and it causes trouble on some
4453 systems, as it is not a Standard C header. It has been removed.
4454
4455 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4456 were reported on the Macintosh. The changes were from
4457
4458 while ((c = *(++ptr)) != 0 && c != '\n');
4459 to
4460 while ((c = *(++ptr)) != 0 && c != '\n') ;
4461
4462 Totally extraordinary, but if that's what it takes...
4463
4464 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4465 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4466 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4467 assumes the way PCRE uses memmove() (always moving upwards).
4468
4469 6. PCRE is being used in one environment where strchr() is not available. There
4470 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4471 faster code anyway.
4472
4473
4474 Version 3.1 09-Feb-00
4475 ---------------------
4476
4477 The only change in this release is the fixing of some bugs in Makefile.in for
4478 the "install" target:
4479
4480 (1) It was failing to install pcreposix.h.
4481
4482 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4483
4484
4485 Version 3.0 01-Feb-00
4486 ---------------------
4487
4488 1. Add support for the /+ modifier to perltest (to output $` like it does in
4489 pcretest).
4490
4491 2. Add support for the /g modifier to perltest.
4492
4493 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4494 matches null strings.
4495
4496 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4497 pattern. Perl treats empty patterns specially - it reuses the most recent
4498 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4499 effect.
4500
4501 5. The POSIX interface was broken in that it was just handing over the POSIX
4502 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4503 required a bigger vector, with some working space on the end. This means that
4504 the POSIX wrapper now has to get and free some memory, and copy the results.
4505
4506 6. Added some simple autoconf support, placing the test data and the
4507 documentation in separate directories, re-organizing some of the
4508 information files, and making it build pcre-config (a GNU standard). Also added
4509 libtool support for building PCRE as a shared library, which is now the
4510 default.
4511
4512 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4513 09 are not valid octal constants. Single digits will be used for minor values
4514 less than 10.
4515
4516 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4517 existing programs that set these in the POSIX interface can use PCRE without
4518 modification.
4519
4520 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4521 return all that pcre_info() returns, plus additional data. The pcre_info()
4522 function is retained for compatibility, but is considered to be obsolete.
4523
4524 10. Added experimental recursion feature (?R) to handle one common case that
4525 Perl 5.6 will be able to do with (?p{...}).
4526
4527 11. Added support for POSIX character classes like [:alpha:], which Perl is
4528 adopting.
4529
4530
4531 Version 2.08 31-Aug-99
4532 ----------------------
4533
4534 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4535 trying to match at the startoffset position, but instead was moving forward to
4536 the next newline as if a previous match had failed.
4537
4538 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4539 and could get into a loop if a null string was matched other than at the start
4540 of the subject.
4541
4542 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4543 be distinguished at compile time, and for completeness also added PCRE_DATE.
4544
4545 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4546 in GnuWin32 environments.
4547
4548
4549 Version 2.07 29-Jul-99
4550 ----------------------
4551
4552 1. The documentation is now supplied in plain text form and HTML as well as in
4553 the form of man page sources.
4554
4555 2. C++ compilers don't like assigning (void *) values to other pointer types.
4556 In particular this affects malloc(). Although there is no problem in Standard
4557 C, I've put in casts to keep C++ compilers happy.
4558
4559 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4560 should be (const char *).
4561
4562 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4563 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4564 However, I haven't made this a standard facility. The documentation doesn't
4565 mention it, and the Makefile doesn't support it.
4566
4567 5. The Makefile now contains an "install" target, with editable destinations at
4568 the top of the file. The pcretest program is not installed.
4569
4570 6. pgrep -V now gives the PCRE version number and date.
4571
4572 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4573 causing the entire string to be ignored, instead of just the last character.
4574
4575 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4576 non-matching string, it can take a very, very long time, even for strings of
4577 quite modest length, because of the nested recursion. PCRE now does better in
4578 some of these cases. It does this by remembering the last required literal
4579 character in the pattern, and pre-searching the subject to ensure it is present
4580 before running the real match. In other words, it applies a heuristic to detect
4581 some types of certain failure quickly, and in the above example, if presented
4582 with a string that has no trailing " it gives "no match" very quickly.
4583
4584 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4585 other alternatives are tried instead.
4586
4587
4588 Version 2.06 09-Jun-99
4589 ----------------------
4590
4591 1. Change pcretest's output for amount of store used to show just the code
4592 space, because the remainder (the data block) varies in size between 32-bit and
4593 64-bit systems.
4594
4595 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4596 start matching at. This allows lookbehinds to work when searching for multiple
4597 occurrences in a string.
4598
4599 3. Added additional options to pcretest for testing multiple occurrences:
4600
4601 /+ outputs the rest of the string that follows a match
4602 /g loops for multiple occurrences, using the new startoffset argument
4603 /G loops for multiple occurrences by passing an incremented pointer
4604
4605 4. PCRE wasn't doing the "first character" optimization for patterns starting
4606 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4607 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4608 the letter 'x'. On long subject strings, this gives a significant speed-up.
4609
4610
4611 Version 2.05 21-Apr-99
4612 ----------------------
4613
4614 1. Changed the type of magic_number from int to long int so that it works
4615 properly on 16-bit systems.
4616
4617 2. Fixed a bug which caused patterns starting with .* not to work correctly
4618 when the subject string contained newline characters. PCRE was assuming
4619 anchoring for such patterns in all cases, which is not correct because .* will
4620 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4621 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4622 must be retried after every newline in the subject.
4623
4624
4625 Version 2.04 18-Feb-99
4626 ----------------------
4627
4628 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4629 computation of the store needed to hold the pattern was incorrect (too large).
4630 If such patterns were nested a few deep, this could multiply and become a real
4631 problem.
4632
4633 2. Added /M option to pcretest to show the memory requirement of a specific
4634 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4635
4636 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4637 compiled in such a way that the backtracking after subsequent failure was
4638 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4639 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4640
4641
4642 Version 2.03 02-Feb-99
4643 ----------------------
4644
4645 1. Fixed typo and small mistake in man page.
4646
4647 2. Added 4th condition (GPL supersedes if conflict) and created separate
4648 LICENCE file containing the conditions.
4649
4650 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4651 Perl, that is the internal \ allows the delimiter to be included in the
4652 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4653 the final delimiter, add \ to the end of the pattern (to test the error).
4654
4655 4. Added the convenience functions for extracting substrings after a successful
4656 match. Updated pcretest to make it able to test these functions.
4657
4658
4659 Version 2.02 14-Jan-99
4660 ----------------------
4661
4662 1. Initialized the working variables associated with each extraction so that
4663 their saving and restoring doesn't refer to uninitialized store.
4664
4665 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4666 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4667 fix the problem.
4668
4669 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4670 calls, and wasn't printing the correct value for compiling calls. Increased the
4671 default value of LOOPREPEAT, and the number of significant figures in the
4672 times.
4673
4674 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4675
4676 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4677 a building problem on Windows NT with a FAT file system.
4678
4679
4680 Version 2.01 21-Oct-98
4681 ----------------------
4682
4683 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4684 to character tables built by pcre_maketables() in the current locale. If NULL
4685 is passed, the default tables are used.
4686
4687
4688 Version 2.00 24-Sep-98
4689 ----------------------
4690
4691 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4692 it any more.
4693
4694 2. Allow quantification of (?>) groups, and make it work correctly.
4695
4696 3. The first character computation wasn't working for (?>) groups.
4697
4698 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4699 end of the subject) and add 5.005's \z, which really does match only at the
4700 very end of the subject.
4701
4702 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4703
4704 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4705 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4706 localized options. All options to pcre_study() were also removed.
4707
4708 7. Add other new features from 5.005:
4709
4710 $(?<= positive lookbehind
4711 $(?<! negative lookbehind
4712 (?imsx-imsx) added the unsetting capability
4713 such a setting is global if at outer level; local otherwise
4714 (?imsx-imsx:) non-capturing groups with option setting
4715 (?(cond)re|re) conditional pattern matching
4716
4717 A backreference to itself in a repeated group matches the previous
4718 captured string.
4719
4720 8. General tidying up of studying (both automatic and via "study")
4721 consequential on the addition of new assertions.
4722
4723 9. As in 5.005, unlimited repeated groups that could match an empty substring
4724 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4725 runtime if any iteration does actually match an empty substring.
4726
4727 10. Include the RunTest script in the distribution.
4728
4729 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4730 discrepancies, some of which were old and were also with respect to 5.004. They
4731 have now been fixed.
4732
4733
4734 Version 1.09 28-Apr-98
4735 ----------------------
4736
4737 1. A negated single character class followed by a quantifier with a minimum
4738 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4739 program crashes, or just wrong answers. This did not apply to negated classes
4740 containing more than one character, or to minima other than one.
4741
4742
4743 Version 1.08 27-Mar-98
4744 ----------------------
4745
4746 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4747
4748 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4749 latter must appear before anything that relies on it in the pattern.
4750
4751
4752 Version 1.07 16-Feb-98
4753 ----------------------
4754
4755 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4756 repeat of a potentially empty string).
4757
4758
4759 Version 1.06 23-Jan-98
4760 ----------------------
4761
4762 1. Added Markus Oberhumer's little patches for C++.
4763
4764 2. Literal strings longer than 255 characters were broken.
4765
4766
4767 Version 1.05 23-Dec-97
4768 ----------------------
4769
4770 1. Negated character classes containing more than one character were failing if
4771 PCRE_CASELESS was set at run time.
4772
4773
4774 Version 1.04 19-Dec-97
4775 ----------------------
4776
4777 1. Corrected the man page, where some "const" qualifiers had been omitted.
4778
4779 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4780 input syntax.
4781
4782 3. Fixed memory leak which occurred when a regex with back references was
4783 matched with an offsets vector that wasn't big enough. The temporary memory
4784 that is used in this case wasn't being freed if the match failed.
4785
4786 4. Tidied pcretest to ensure it frees memory that it gets.
4787
4788 5. Temporary memory was being obtained in the case where the passed offsets
4789 vector was exactly big enough.
4790
4791 6. Corrected definition of offsetof() from change 5 below.
4792
4793 7. I had screwed up change 6 below and broken the rules for the use of
4794 setjmp(). Now fixed.
4795
4796
4797 Version 1.03 18-Dec-97
4798 ----------------------
4799
4800 1. A erroneous regex with a missing opening parenthesis was correctly
4801 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4802 on some systems.
4803
4804 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4805 it was reported that one broken compiler failed on the former because "code" is
4806 also an independent variable.
4807
4808 3. The erroneous regex a[]b caused an array overrun reference.
4809
4810 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4811 fail on data ending with that character. (It was going on too far, and checking
4812 the next character, typically a binary zero.) This was specific to the
4813 optimized code for single-character negative classes.
4814
4815 5. Added a contributed patch from the TIN world which does the following:
4816
4817 + Add an undef for memmove, in case the the system defines a macro for it.
4818
4819 + Add a definition of offsetof(), in case there isn't one. (I don't know
4820 the reason behind this - offsetof() is part of the ANSI standard - but
4821 it does no harm).
4822
4823 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4824 most of the places where whitespace preceded '#'. I have given up and
4825 allowed the remaining 2 cases to be at the margin.
4826
4827 + Rename some variables in pcre to eliminate shadowing. This seems very
4828 pedantic, but does no harm, of course.
4829
4830 6. Moved the call to setjmp() into its own function, to get rid of warnings
4831 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4832
4833 7. Constructs such as \d{8,} were compiling into the equivalent of
4834 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4835 outcome, but in this particular case used more store than had been allocated,
4836 which caused the bug to be discovered because it threw up an internal error.
4837
4838 8. The debugging code in both pcre and pcretest for outputting the compiled
4839 form of a regex was going wrong in the case of back references followed by
4840 curly-bracketed repeats.
4841
4842
4843 Version 1.02 12-Dec-97
4844 ----------------------
4845
4846 1. Typos in pcre.3 and comments in the source fixed.
4847
4848 2. Applied a contributed patch to get rid of places where it used to remove
4849 'const' from variables, and fixed some signed/unsigned and uninitialized
4850 variable warnings.
4851
4852 3. Added the "runtest" target to Makefile.
4853
4854 4. Set default compiler flag to -O2 rather than just -O.
4855
4856
4857 Version 1.01 19-Nov-97
4858 ----------------------
4859
4860 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4861 like /([ab]*)*/, that is, for classes with more than one character in them.
4862
4863 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4864 as /((?>a*))*/ (a PCRE_EXTRA facility).
4865
4866
4867 Version 1.00 18-Nov-97
4868 ----------------------
4869
4870 1. Added compile-time macros to support systems such as SunOS4 which don't have
4871 memmove() or strerror() but have other things that can be used instead.
4872
4873 2. Arranged that "make clean" removes the executables.
4874
4875
4876 Version 0.99 27-Oct-97
4877 ----------------------
4878
4879 1. Fixed bug in code for optimizing classes with only one character. It was
4880 initializing a 32-byte map regardless, which could cause it to run off the end
4881 of the memory it had got.
4882
4883 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4884
4885
4886 Version 0.98 22-Oct-97
4887 ----------------------
4888
4889 1. Fixed bug in code for handling temporary memory usage when there are more
4890 back references than supplied space in the ovector. This could cause segfaults.
4891
4892
4893 Version 0.97 21-Oct-97
4894 ----------------------
4895
4896 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4897
4898 2. Optimized negated single characters not to use a bit map.
4899
4900 3. Brought error texts together as macro definitions; clarified some of them;
4901 fixed one that was wrong - it said "range out of order" when it meant "invalid
4902 escape sequence".
4903
4904 4. Changed some char * arguments to const char *.
4905
4906 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4907
4908 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4909 pcretest.
4910
4911
4912 Version 0.96 16-Oct-97
4913 ----------------------
4914
4915 1. Added a simple "pgrep" utility to the distribution.
4916
4917 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4918 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4919 where "ddd" means "one or more decimal digits".
4920
4921 3. Fixed serious bug. If a pattern had a back reference, but the call to
4922 pcre_exec() didn't supply a large enough ovector to record the related
4923 identifying subpattern, the match always failed. PCRE now remembers the number
4924 of the largest back reference, and gets some temporary memory in which to save
4925 the offsets during matching if necessary, in order to ensure that
4926 backreferences always work.
4927
4928 4. Increased the compatibility with Perl in a number of ways:
4929
4930 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4931 to request this handling. The option can be set at compile or exec time.
4932
4933 (b) $ matches before a terminating newline by default; an option
4934 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
4935 mode). The option can be set at compile or exec time.
4936
4937 (c) The handling of \ followed by a digit other than 0 is now supposed to be
4938 the same as Perl's. If the decimal number it represents is less than 10
4939 or there aren't that many previous left capturing parentheses, an octal
4940 escape is read. Inside a character class, it's always an octal escape,
4941 even if it is a single digit.
4942
4943 (d) An escaped but undefined alphabetic character is taken as a literal,
4944 unless PCRE_EXTRA is set. Currently this just reserves the remaining
4945 escapes.
4946
4947 (e) {0} is now permitted. (The previous item is removed from the compiled
4948 pattern).
4949
4950 5. Changed all the names of code files so that the basic parts are no longer
4951 than 10 characters, and abolished the teeny "globals.c" file.
4952
4953 6. Changed the handling of character classes; they are now done with a 32-byte
4954 bit map always.
4955
4956 7. Added the -d and /D options to pcretest to make it possible to look at the
4957 internals of compilation without having to recompile pcre.
4958
4959
4960 Version 0.95 23-Sep-97
4961 ----------------------
4962
4963 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
4964 \x20 at the start of a run of normal characters. These were being treated as
4965 real characters, instead of the source characters being re-checked.
4966
4967
4968 Version 0.94 18-Sep-97
4969 ----------------------
4970
4971 1. The functions are now thread-safe, with the caveat that the global variables
4972 containing pointers to malloc() and free() or alternative functions are the
4973 same for all threads.
4974
4975 2. Get pcre_study() to generate a bitmap of initial characters for non-
4976 anchored patterns when this is possible, and use it if passed to pcre_exec().
4977
4978
4979 Version 0.93 15-Sep-97
4980 ----------------------
4981
4982 1. /(b)|(:+)/ was computing an incorrect first character.
4983
4984 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
4985 but not actually doing anything yet.
4986
4987 3. Treat "-" characters in classes that cannot be part of ranges as literals,
4988 as Perl does (e.g. [-az] or [az-]).
4989
4990 4. Set the anchored flag if a branch starts with .* or .*? because that tests
4991 all possible positions.
4992
4993 5. Split up into different modules to avoid including unneeded functions in a
4994 compiled binary. However, compile and exec are still in one module. The "study"
4995 function is split off.
4996
4997 6. The character tables are now in a separate module whose source is generated
4998 by an auxiliary program - but can then be edited by hand if required. There are
4999 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5000 toupper() in the code.
5001
5002 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5003 make them global. Abolish the function for setting them, as the caller can now
5004 set them directly.
5005
5006
5007 Version 0.92 11-Sep-97
5008 ----------------------
5009
5010 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5011 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5012
5013 2. Caseless matching was not working in character classes if the characters in
5014 the pattern were in upper case.
5015
5016 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5017
5018 4. Make PCRE_ANCHORED public and accept as a compile option.
5019
5020 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5021 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5022 pass them.
5023
5024 6. Give an error if bad option bits passed at compile or run time.
5025
5026 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5027 pcretest to cause it to pass that flag.
5028
5029 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5030 options, and the first character, if set.
5031
5032 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5033
5034
5035 Version 0.91 10-Sep-97
5036 ----------------------
5037
5038 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5039 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5040
5041 2. PCRE was looping on encountering an indefinitely repeated back reference to
5042 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5043 Perl does - treats the match as successful.
5044
5045 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5