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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1306 - (show annotations)
Mon Apr 1 17:04:17 2013 UTC (6 years, 8 months ago) by zherczeg
File size: 245825 byte(s)
Error occurred while calculating annotation data.
Auto-detect and optimize limited repetitions in JIT.
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 17. The \A escape now records a lookbehind value of 1, though its execution
68 does not actually inspect the previous character. This is to ensure that,
69 in partial multi-segment matching, at least one character from the old
70 segment is retained when a new segment is processed. Otherwise, if there
71 are no lookbehinds in the pattern, \A might match incorrectly at the start
72 of a new segment.
73
74 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
75
76 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
77 gives some modest performance improvement in 8-bit mode.
78
79 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
80 be expressed in certain programming languages using Universal Character
81 Names.
82
83 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
84 which points out that "non characters" are not "characters that may not
85 appear in Unicode strings" but rather "characters that are reserved for
86 internal use and have only local meaning".
87
88 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
89 there was a conditional group that depended on an assertion, if the
90 assertion was false, the callout that immediately followed the alternation
91 in the condition was skipped when pcre_exec() was used for matching.
92
93 23. Allow an explicit callout to be inserted before an assertion that is the
94 condition for a conditional group, for compatibility with automatic
95 callouts, which always insert a callout at this point.
96
97 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
98 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
99
100 25. (*PRUNE) is now supported by the JIT compiler.
101
102 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
103
104 27. Fix the case where there are two or more SKIPs with arguments that may be
105 ignored.
106
107 28. (*SKIP) is now supported by the JIT compiler.
108
109 29. (*THEN) is now supported by the JIT compiler.
110
111 30. Update RunTest with additional test selector options.
112
113 31. The way PCRE handles backtracking verbs has been changed in two ways.
114
115 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
116 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
117 backtracking. In some cases this makes it more Perl-compatible, but Perl's
118 rather obscure rules do not always do the same thing.
119
120 (2) Previously, backtracking verbs were confined within assertions. This is
121 no longer the case for positive assertions, except for (*ACCEPT). Again,
122 this sometimes improves Perl compatibility, and sometimes does not.
123
124 32. A number of tests that were in test 2 because Perl did things differently
125 have been moved to test 1, because either Perl or PCRE has changed, and
126 these tests are now compatible.
127
128 32. Control verbs are handled in the same way in JIT and interpreter.
129
130 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
131 contained a forward subroutine reference caused a compile error.
132
133 34. Auto-detect and optimize limited repetitions in JIT.
134
135
136 Version 8.32 30-November-2012
137 -----------------------------
138
139 1. Improved JIT compiler optimizations for first character search and single
140 character iterators.
141
142 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
143 Patch by Daniel Richard G.
144
145 3. Single character iterator optimizations in the JIT compiler.
146
147 4. Improved JIT compiler optimizations for character ranges.
148
149 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
150 Reported by Giuseppe D'Angelo.
151
152 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
153 of a line, was being set incorrectly in cases where .* appeared inside
154 atomic brackets at the start of a pattern, or where there was a subsequent
155 *PRUNE or *SKIP.
156
157 7. Improved instruction cache flush for POWER/PowerPC.
158 Patch by Daniel Richard G.
159
160 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
161 grep:
162
163 (a) There is now no limit to the number of patterns to be matched.
164
165 (b) An error is given if a pattern is too long.
166
167 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
168 are now supported.
169
170 (d) --exclude-from and --include-from (multiple use) have been added.
171
172 (e) Exclusions and inclusions now apply to all files and directories, not
173 just to those obtained from scanning a directory recursively.
174
175 (f) Multiple uses of -f and --file-list are now supported.
176
177 (g) In a Windows environment, the default for -d has been changed from
178 "read" (the GNU grep default) to "skip", because otherwise the presence
179 of a directory in the file list provokes an error.
180
181 (h) The documentation has been revised and clarified in places.
182
183 9. Improve the matching speed of capturing brackets.
184
185 10. Changed the meaning of \X so that it now matches a Unicode extended
186 grapheme cluster.
187
188 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
189 out POSIX threads when JIT support is configured.
190
191 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
192
193 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
194 structure could go wrong in environments where size_t is not the same size
195 as int.
196
197 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
198 set.
199
200 15. The EBCDIC support had decayed; later updates to the code had included
201 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
202 general tidy up of EBCDIC-related issues, and the documentation was also
203 not quite right. There is now a test that can be run on ASCII systems to
204 check some of the EBCDIC-related things (but is it not a full test).
205
206 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
207 in a small tidy to the code.
208
209 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
210
211 18. If the --only-matching (-o) option in pcregrep is specified multiple
212 times, each one causes appropriate output. For example, -o1 -o2 outputs the
213 substrings matched by the 1st and 2nd capturing parentheses. A separating
214 string can be specified by --om-separator (default empty).
215
216 19. Improving the first n character searches.
217
218 20. Turn case lists for horizontal and vertical white space into macros so that
219 they are defined only once.
220
221 21. This set of changes together give more compatible Unicode case-folding
222 behaviour for characters that have more than one other case when UCP
223 support is available.
224
225 (a) The Unicode property table now has offsets into a new table of sets of
226 three or more characters that are case-equivalent. The MultiStage2.py
227 script that generates these tables (the pcre_ucd.c file) now scans
228 CaseFolding.txt instead of UnicodeData.txt for character case
229 information.
230
231 (b) The code for adding characters or ranges of characters to a character
232 class has been abstracted into a generalized function that also handles
233 case-independence. In UTF-mode with UCP support, this uses the new data
234 to handle characters with more than one other case.
235
236 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
237 whose other case is greater than 256 are now correctly matched
238 caselessly. Previously, the high codepoint matched the low one, but not
239 vice versa.
240
241 (d) The processing of \h, \H, \v, and \ in character classes now makes use
242 of the new class addition function, using character lists defined as
243 macros alongside the case definitions of 20 above.
244
245 (e) Caseless back references now work with characters that have more than
246 one other case.
247
248 (f) General caseless matching of characters with more than one other case
249 is supported.
250
251 22. Unicode character properties were updated from Unicode 6.2.0
252
253 23. Improved CMake support under Windows. Patch by Daniel Richard G.
254
255 24. Add support for 32-bit character strings, and UTF-32
256
257 25. Major JIT compiler update (code refactoring and bugfixing).
258 Experimental Sparc 32 support is added.
259
260 26. Applied a modified version of Daniel Richard G's patch to create
261 pcre.h.generic and config.h.generic by "make" instead of in the
262 PrepareRelease script.
263
264 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
265 pcre_compile.c when checking for a zero character.
266
267 28. Introducing a native interface for JIT. Through this interface, the compiled
268 machine code can be directly executed. The purpose of this interface is to
269 provide fast pattern matching, so several sanity checks are not performed.
270 However, feature tests are still performed. The new interface provides
271 1.4x speedup compared to the old one.
272
273 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
274 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
275 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
276
277 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
278 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
279 the "old" RFC 2279). Instead, it ended up passing a negative length to
280 pcre_exec().
281
282 31. Add support for GCC's visibility feature to hide internal functions.
283
284 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
285 "unknown -C option" after outputting 0 or 1.
286
287 33. There is now support for generating a code coverage report for the test
288 suite in environments where gcc is the compiler and lcov is installed. This
289 is mainly for the benefit of the developers.
290
291 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
292 unaddressable using valgrind annotations, allowing valgrind to detect
293 invalid memory accesses. This is mainly for the benefit of the developers.
294
295 25. (*UTF) can now be used to start a pattern in any of the three libraries.
296
297 26. Give configure error if --enable-cpp but no C++ compiler found.
298
299
300 Version 8.31 06-July-2012
301 -------------------------
302
303 1. Fixing a wrong JIT test case and some compiler warnings.
304
305 2. Removed a bashism from the RunTest script.
306
307 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
308 to unsigned type, result still unsigned" that was given by an MS compiler
309 on encountering the code "-sizeof(xxx)".
310
311 4. Partial matching support is added to the JIT compiler.
312
313 5. Fixed several bugs concerned with partial matching of items that consist
314 of more than one character:
315
316 (a) /^(..)\1/ did not partially match "aba" because checking references was
317 done on an "all or nothing" basis. This also applied to repeated
318 references.
319
320 (b) \R did not give a hard partial match if \r was found at the end of the
321 subject.
322
323 (c) \X did not give a hard partial match after matching one or more
324 characters at the end of the subject.
325
326 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
327 a partial match for the string "\r".
328
329 (e) When newline was set to CRLF, the metacharacter "." did not recognize
330 a partial match for a CR character at the end of the subject string.
331
332 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
333 running pcretest, the text "(JIT)" added to the output whenever JIT is
334 actually used to run the match.
335
336 7. Individual JIT compile options can be set in pcretest by following -s+[+]
337 or /S+[+] with a digit between 1 and 7.
338
339 8. OP_NOT now supports any UTF character not just single-byte ones.
340
341 9. (*MARK) control verb is now supported by the JIT compiler.
342
343 10. The command "./RunTest list" lists the available tests without actually
344 running any of them. (Because I keep forgetting what they all are.)
345
346 11. Add PCRE_INFO_MAXLOOKBEHIND.
347
348 12. Applied a (slightly modified) user-supplied patch that improves performance
349 when the heap is used for recursion (compiled with --disable-stack-for-
350 recursion). Instead of malloc and free for each heap frame each time a
351 logical recursion happens, frames are retained on a chain and re-used where
352 possible. This sometimes gives as much as 30% improvement.
353
354 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
355 call.
356
357 14. As documented, (*COMMIT) is now confined to within a positive assertion.
358
359 15. It is now possible to link pcretest with libedit as an alternative to
360 libreadline.
361
362 16. (*COMMIT) control verb is now supported by the JIT compiler.
363
364 17. The Unicode data tables have been updated to Unicode 6.1.0.
365
366 18. Added --file-list option to pcregrep.
367
368 19. Added binary file support to pcregrep, including the -a, --binary-files,
369 -I, and --text options.
370
371 20. The madvise function is renamed for posix_madvise for QNX compatibility
372 reasons. Fixed by Giuseppe D'Angelo.
373
374 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
375
376 22. Changed the option for creating symbolic links for 16-bit man pages from
377 -s to -sf so that re-installing does not cause issues.
378
379 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
380
381 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
382 matches in certain environments (the workspace was not being correctly
383 retained). Also added to pcre_dfa_exec() a simple plausibility check on
384 some of the workspace data at the beginning of a restart.
385
386 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
387 was not doing so when it should - probably a typo introduced by SVN 528
388 (change 8.10/14).
389
390 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
391 \w+ when the character tables indicated that \x{c4} was a word character.
392 There were several related cases, all because the tests for doing a table
393 lookup were testing for characters less than 127 instead of 255.
394
395 27. If a pattern contains capturing parentheses that are not used in a match,
396 their slots in the ovector are set to -1. For those that are higher than
397 any matched groups, this happens at the end of processing. In the case when
398 there were back references that the ovector was too small to contain
399 (causing temporary malloc'd memory to be used during matching), and the
400 highest capturing number was not used, memory off the end of the ovector
401 was incorrectly being set to -1. (It was using the size of the temporary
402 memory instead of the true size.)
403
404 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
405 ovector size, it uses memory at the end of the block that it has got.
406
407 29. Check for an overlong MARK name and give an error at compile time. The
408 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
409
410 30. JIT compiler update.
411
412 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
413 Rill for the patch.
414
415 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
416
417 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
418
419 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
420 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
421 was enabled.
422
423 35. Improve JIT code generation for greedy plus quantifier.
424
425 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
426 1 to "aa" instead of to an empty string. The bug affected repeated groups
427 that could potentially match an empty string.
428
429 37. Optimizing single character iterators in JIT.
430
431 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
432 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
433 codepoints that are too big for the mode are faulted, and in a UTF mode,
434 disallowed codepoints are also faulted.
435
436 39. If PCRE was compiled with UTF support, in three places in the DFA
437 matcher there was code that should only have been obeyed in UTF mode, but
438 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
439 processing when bytes with values greater than 127 were present. In 16-bit
440 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
441 both cases the values are those that cannot be the first data item in a UTF
442 character. The three items that might have provoked this were recursions,
443 possessively repeated groups, and atomic groups.
444
445 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
446 and pcregrep, because some OS require shared objects to be explicitly
447 passed to ld, causing the link step to fail if they are not.
448
449 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
450 mode, patterns that started with \h* or \R* might be incorrectly matched.
451
452
453 Version 8.30 04-February-2012
454 -----------------------------
455
456 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
457 name is defined in ctype.h.
458
459 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
460 only in quite long subpatterns.
461
462 3. Removed the function pcre_info(), which has been obsolete and deprecated
463 since it was replaced by pcre_fullinfo() in February 2000.
464
465 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
466 match a (*MARK), and the match failed at the start of the subject, a
467 reference to memory before the start of the subject could occur. This bug
468 was introduced by fix 17 of release 8.21.
469
470 5. A reference to an unset group with zero minimum repetition was giving
471 totally wrong answers (in non-JavaScript-compatibility mode). For example,
472 /(another)?(\1?)test/ matched against "hello world test". This bug was
473 introduced in release 8.13.
474
475 6. Add support for 16-bit character strings (a large amount of work involving
476 many changes and refactorings).
477
478 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
479 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
480 from a file.
481
482 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
483 rounding is not applied in this particular case).
484
485 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
486 if they appear, or are escaped, in patterns.
487
488 10. Get rid of a number of -Wunused-but-set-variable warnings.
489
490 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
491 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
492 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
493 also returns the mark "x". This bug applied to capturing parentheses,
494 non-capturing parentheses, and atomic parentheses. It also applied to some
495 assertions.
496
497 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
498 information out of configure.ac instead of relying on pcre.h.generic, which
499 is not stored in the repository.
500
501 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
502 -lreadline.
503
504 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
505
506 15. Applied Graycode's patch to put the top-level frame on the stack rather
507 than the heap when not using the stack for recursion. This gives a
508 performance improvement in many cases when recursion is not deep.
509
510 16. Experimental code added to "pcretest -C" to output the stack frame size.
511
512
513 Version 8.21 12-Dec-2011
514 ------------------------
515
516 1. Updating the JIT compiler.
517
518 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
519 are added as well.
520
521 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
522 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
523 calling _pcre_jit_exec. Some extra comments are added.
524
525 4. (*MARK) settings inside atomic groups that do not contain any capturing
526 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
527 was introduced by change 18 for 8.20.
528
529 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
530 ECMA-262 standard.
531
532 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
533 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
534 This bug was probably introduced by change 9 of 8.13.
535
536 7. While fixing 6 above, I noticed that a number of other items were being
537 incorrectly rejected as "not fixed length". This arose partly because newer
538 opcodes had not been added to the fixed-length checking code. I have (a)
539 corrected the bug and added tests for these items, and (b) arranged for an
540 error to occur if an unknown opcode is encountered while checking for fixed
541 length instead of just assuming "not fixed length". The items that were
542 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
543 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
544 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
545
546 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
547 being incorrectly compiled and would have given unpredicatble results.
548
549 9. A possessively repeated subpattern with minimum repeat count greater than
550 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
551 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
552 the first (A) could occur when it should not.
553
554 10. Add a cast and remove a redundant test from the code.
555
556 11. JIT should use pcre_malloc/pcre_free for allocation.
557
558 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
559 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
560 is anything other than /usr, -L is still shown).
561
562 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
563
564 14. Perl does not support \N without a following name in a [] class; PCRE now
565 also gives an error.
566
567 15. If a forward reference was repeated with an upper limit of around 2000,
568 it caused the error "internal error: overran compiling workspace". The
569 maximum number of forward references (including repeats) was limited by the
570 internal workspace, and dependent on the LINK_SIZE. The code has been
571 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
572 the default depends on LINK_SIZE. There is a new upper limit (for safety)
573 of around 200,000 forward references. While doing this, I also speeded up
574 the filling in of repeated forward references.
575
576 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
577 incorrectly expecting the subject to contain another "a" after the start.
578
579 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
580 in the match, the SKIP should be ignored. This was not happening; instead
581 the SKIP was being treated as NOMATCH. For patterns such as
582 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
583 tested.
584
585 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
586 now much more compatible with Perl, in particular in cases where the result
587 is a non-match for a non-anchored pattern. For example, if
588 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
589 "m", where previously it did not return a name. A side effect of this
590 change is that for partial matches, the last encountered mark name is
591 returned, as for non matches. A number of tests that were previously not
592 Perl-compatible have been moved into the Perl-compatible test files. The
593 refactoring has had the pleasing side effect of removing one argument from
594 the match() function, thus reducing its stack requirements.
595
596 19. If the /S+ option was used in pcretest to study a pattern using JIT,
597 subsequent uses of /S (without +) incorrectly behaved like /S+.
598
599 21. Retrieve executable code size support for the JIT compiler and fixing
600 some warnings.
601
602 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
603 not work when the shorter character appeared right at the end of the
604 subject string.
605
606 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
607 systems.
608
609 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
610 output it when the /M option is used in pcretest.
611
612 25. The CheckMan script was not being included in the distribution. Also, added
613 an explicit "perl" to run Perl scripts from the PrepareRelease script
614 because this is reportedly needed in Windows.
615
616 26. If study data was being save in a file and studying had not found a set of
617 "starts with" bytes for the pattern, the data written to the file (though
618 never used) was taken from uninitialized memory and so caused valgrind to
619 complain.
620
621 27. Updated RunTest.bat as provided by Sheri Pierce.
622
623 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
624
625 29. Computation of memory usage for the table of capturing group names was
626 giving an unnecessarily large value.
627
628
629 Version 8.20 21-Oct-2011
630 ------------------------
631
632 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
633 a POSIX class. After further experiments with Perl, which convinced me that
634 Perl has bugs and confusions, a closing square bracket is no longer allowed
635 in a POSIX name. This bug also affected patterns with classes that started
636 with full stops.
637
638 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
639 captured substring, but while checking the failing first alternative,
640 substring 1 is temporarily captured. If the output vector supplied to
641 pcre_exec() was not big enough for this capture, the yield of the function
642 was still zero ("insufficient space for captured substrings"). This cannot
643 be totally fixed without adding another stack variable, which seems a lot
644 of expense for a edge case. However, I have improved the situation in cases
645 such as /(a)(b)x|abc/ matched against "abc", where the return code
646 indicates that fewer than the maximum number of slots in the ovector have
647 been set.
648
649 3. Related to (2) above: when there are more back references in a pattern than
650 slots in the output vector, pcre_exec() uses temporary memory during
651 matching, and copies in the captures as far as possible afterwards. It was
652 using the entire output vector, but this conflicts with the specification
653 that only 2/3 is used for passing back captured substrings. Now it uses
654 only the first 2/3, for compatibility. This is, of course, another edge
655 case.
656
657 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
658 main code base, and can be used by building with --enable-jit. When this is
659 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
660 runtime --no-jit option is given.
661
662 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
663 ovector, the return from the function was zero, implying that there were
664 other matches that did not fit. The correct "exactly full" value is now
665 returned.
666
667 6. If a subpattern that was called recursively or as a subroutine contained
668 (*PRUNE) or any other control that caused it to give a non-standard return,
669 invalid errors such as "Error -26 (nested recursion at the same subject
670 position)" or even infinite loops could occur.
671
672 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
673 computing the minimum length on reaching *ACCEPT, and so ended up with the
674 wrong value of 1 rather than 0. Further investigation indicates that
675 computing a minimum subject length in the presence of *ACCEPT is difficult
676 (think back references, subroutine calls), and so I have changed the code
677 so that no minimum is registered for a pattern that contains *ACCEPT.
678
679 8. If (*THEN) was present in the first (true) branch of a conditional group,
680 it was not handled as intended. [But see 16 below.]
681
682 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
683 Sheri Pierce.
684
685 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
686 the first byte in a match must be "a".
687
688 11. Change 17 for 8.13 increased the recursion depth for patterns like
689 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
690 pattern contains any instances of (*THEN). If it does not, the old
691 optimizations are restored. It would be nice to do this on a per-group
692 basis, but at the moment that is not feasible.
693
694 12. In some environments, the output of pcretest -C is CRLF terminated. This
695 broke RunTest's code that checks for the link size. A single white space
696 character after the value is now allowed for.
697
698 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
699 For "fr", it uses the Windows-specific input and output files.
700
701 14. If (*THEN) appeared in a group that was called recursively or as a
702 subroutine, it did not work as intended. [But see next item.]
703
704 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
705 pattern fragments (but not containing any | characters). If A and B are
706 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
707 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
708 D. In other words, Perl considers parentheses that do not contain any |
709 characters to be part of a surrounding alternative, whereas PCRE was
710 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
711 differently. PCRE now behaves in the same way as Perl, except in the case
712 of subroutine/recursion calls such as (?1) which have in any case always
713 been different (but PCRE had them first :-).
714
715 16. Related to 15 above: Perl does not treat the | in a conditional group as
716 creating alternatives. Such a group is treated in the same way as an
717 ordinary group without any | characters when processing (*THEN). PCRE has
718 been changed to match Perl's behaviour.
719
720 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
721 RunGrepTest script failed.
722
723 18. Change 22 for version 13 caused atomic groups to use more stack. This is
724 inevitable for groups that contain captures, but it can lead to a lot of
725 stack use in large patterns. The old behaviour has been restored for atomic
726 groups that do not contain any capturing parentheses.
727
728 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
729 suppress the check for a minimum subject length at run time. (If it was
730 given to pcre_exec() or pcre_dfa_exec() it did work.)
731
732 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
733 fail to work when decoding hex characters in data strings in EBCDIC
734 environments.
735
736 21. It appears that in at least one Mac OS environment, the isxdigit() function
737 is implemented as a macro that evaluates to its argument more than once,
738 contravening the C 90 Standard (I haven't checked a later standard). There
739 was an instance in pcretest which caused it to go wrong when processing
740 \x{...} escapes in subject strings. The has been rewritten to avoid using
741 things like p++ in the argument of isxdigit().
742
743
744 Version 8.13 16-Aug-2011
745 ------------------------
746
747 1. The Unicode data tables have been updated to Unicode 6.0.0.
748
749 2. Two minor typos in pcre_internal.h have been fixed.
750
751 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
752 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
753 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
754
755 4. There were a number of related bugs in the code for matching backrefences
756 caselessly in UTF-8 mode when codes for the characters concerned were
757 different numbers of bytes. For example, U+023A and U+2C65 are an upper
758 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
759 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
760 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
761 2-byte code at the end of the subject (it thought there wasn't enough data
762 left).
763
764 5. Comprehensive information about what went wrong is now returned by
765 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
766 as the output vector has at least 2 elements. The offset of the start of
767 the failing character and a reason code are placed in the vector.
768
769 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
770 now returned is for the first byte of the failing character, instead of the
771 last byte inspected. This is an incompatible change, but I hope it is small
772 enough not to be a problem. It makes the returned offset consistent with
773 pcre_exec() and pcre_dfa_exec().
774
775 7. pcretest now gives a text phrase as well as the error number when
776 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
777 failure, the offset and reason code are output.
778
779 8. When \R was used with a maximizing quantifier it failed to skip backwards
780 over a \r\n pair if the subsequent match failed. Instead, it just skipped
781 back over a single character (\n). This seems wrong (because it treated the
782 two characters as a single entity when going forwards), conflicts with the
783 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
784 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
785 has been changed.
786
787 9. Some internal refactoring has changed the processing so that the handling
788 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
789 time (the PCRE_DOTALL option was changed this way some time ago: version
790 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
791 which was always a bit of a fudge. It also means that there is one less
792 argument for the match() function, which reduces its stack requirements
793 slightly. This change also fixes an incompatibility with Perl: the pattern
794 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
795
796 10. More internal refactoring has drastically reduced the number of recursive
797 calls to match() for possessively repeated groups such as (abc)++ when
798 using pcre_exec().
799
800 11. While implementing 10, a number of bugs in the handling of groups were
801 discovered and fixed:
802
803 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
804 (a|)*(?1) gave a compile-time internal error.
805 ((a|)+)+ did not notice that the outer group could match an empty string.
806 (^a|^)+ was not marked as anchored.
807 (.*a|.*)+ was not marked as matching at start or after a newline.
808
809 12. Yet more internal refactoring has removed another argument from the match()
810 function. Special calls to this function are now indicated by setting a
811 value in a variable in the "match data" data block.
812
813 13. Be more explicit in pcre_study() instead of relying on "default" for
814 opcodes that mean there is no starting character; this means that when new
815 ones are added and accidentally left out of pcre_study(), testing should
816 pick them up.
817
818 14. The -s option of pcretest has been documented for ages as being an old
819 synonym of -m (show memory usage). I have changed it to mean "force study
820 for every regex", that is, assume /S for every regex. This is similar to -i
821 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
822 using it. It makes it easier to run collections of tests with and without
823 study enabled, and thereby test pcre_study() more easily. All the standard
824 tests are now run with and without -s (but some patterns can be marked as
825 "never study" - see 20 below).
826
827 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
828 restoration of the capturing data to the outer values was not happening
829 correctly.
830
831 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
832 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
833 pattern had matched an empty string, and so incorrectly returned a no
834 match.
835
836 17. There was optimizing code for the last branch of non-capturing parentheses,
837 and also for the obeyed branch of a conditional subexpression, which used
838 tail recursion to cut down on stack usage. Unfortunately, now that there is
839 the possibility of (*THEN) occurring in these branches, tail recursion is
840 no longer possible because the return has to be checked for (*THEN). These
841 two optimizations have therefore been removed. [But see 8.20/11 above.]
842
843 18. If a pattern containing \R was studied, it was assumed that \R always
844 matched two bytes, thus causing the minimum subject length to be
845 incorrectly computed because \R can also match just one byte.
846
847 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
848 was incorrectly computed.
849
850 20. If /S is present twice on a test pattern in pcretest input, it now
851 *disables* studying, thereby overriding the use of -s on the command line
852 (see 14 above). This is necessary for one or two tests to keep the output
853 identical in both cases.
854
855 21. When (*ACCEPT) was used in an assertion that matched an empty string and
856 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
857
858 22. When an atomic group that contained a capturing parenthesis was
859 successfully matched, but the branch in which it appeared failed, the
860 capturing was not being forgotten if a higher numbered group was later
861 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
862 group 1 to "a", when in fact it should be unset. This applied to multi-
863 branched capturing and non-capturing groups, repeated or not, and also to
864 positive assertions (capturing in negative assertions does not happen
865 in PCRE) and also to nested atomic groups.
866
867 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
868 subject after a captured substring, to make it easier to tell which of a
869 number of identical substrings has been captured.
870
871 24. The way atomic groups are processed by pcre_exec() has been changed so that
872 if they are repeated, backtracking one repetition now resets captured
873 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
874 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
875 "aaa". Previously, it would have been "a". As part of this code
876 refactoring, the way recursive calls are handled has also been changed.
877
878 25. If an assertion condition captured any substrings, they were not passed
879 back unless some other capturing happened later. For example, if
880 (?(?=(a))a) was matched against "a", no capturing was returned.
881
882 26. When studying a pattern that contained subroutine calls or assertions,
883 the code for finding the minimum length of a possible match was handling
884 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
885 group 1 called group 2 while simultaneously a separate group 2 called group
886 1). A stack overflow occurred in this case. I have fixed this by limiting
887 the recursion depth to 10.
888
889 27. Updated RunTest.bat in the distribution to the version supplied by Tom
890 Fortmann. This supports explicit test numbers on the command line, and has
891 argument validation and error reporting.
892
893 28. An instance of \X with an unlimited repeat could fail if at any point the
894 first character it looked at was a mark character.
895
896 29. Some minor code refactoring concerning Unicode properties and scripts
897 should reduce the stack requirement of match() slightly.
898
899 30. Added the '=' option to pcretest to check the setting of unused capturing
900 slots at the end of the pattern, which are documented as being -1, but are
901 not included in the return count.
902
903 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
904 compiled something random. Now it gives a compile-time error (as does
905 Perl).
906
907 32. A *MARK encountered during the processing of a positive assertion is now
908 recorded and passed back (compatible with Perl).
909
910 33. If --only-matching or --colour was set on a pcregrep call whose pattern
911 had alternative anchored branches, the search for a second match in a line
912 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
913 matched the line "0102" twice. The same bug affected patterns that started
914 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
915 twice.
916
917 34. Previously, PCRE did not allow quantification of assertions. However, Perl
918 does, and because of capturing effects, quantifying parenthesized
919 assertions may at times be useful. Quantifiers are now allowed for
920 parenthesized assertions.
921
922 35. A minor code tidy in pcre_compile() when checking options for \R usage.
923
924 36. \g was being checked for fancy things in a character class, when it should
925 just be a literal "g".
926
927 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
928 appearance of a nested POSIX class supersedes an apparent external class.
929 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
930 unescaped square brackets may also appear as part of class names. For
931 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
932 more like Perl. (But see 8.20/1 above.)
933
934 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
935 was because it thought it was \N{name}, which is not supported).
936
937 39. Add minix to OS list not supporting the -S option in pcretest.
938
939 40. PCRE tries to detect cases of infinite recursion at compile time, but it
940 cannot analyze patterns in sufficient detail to catch mutual recursions
941 such as ((?1))((?2)). There is now a runtime test that gives an error if a
942 subgroup is called recursively as a subpattern for a second time at the
943 same position in the subject string. In previous releases this might have
944 been caught by the recursion limit, or it might have run out of stack.
945
946 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
947 happen only once. PCRE was, however incorrectly giving a compile time error
948 "recursive call could loop indefinitely" because it cannot analyze the
949 pattern in sufficient detail. The compile time test no longer happens when
950 PCRE is compiling a conditional subpattern, but actual runaway loops are
951 now caught at runtime (see 40 above).
952
953 42. It seems that Perl allows any characters other than a closing parenthesis
954 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
955 has been changed to be the same.
956
957 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
958 as not to get warnings when autogen.sh is called. Also changed
959 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
960
961 44. To help people who use pcregrep to scan files containing exceedingly long
962 lines, the following changes have been made:
963
964 (a) The default value of the buffer size parameter has been increased from
965 8K to 20K. (The actual buffer used is three times this size.)
966
967 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
968 PCRE is built.
969
970 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
971 to be set at run time.
972
973 (d) Numerical values in pcregrep options can be followed by K or M, for
974 example --buffer-size=50K.
975
976 (e) If a line being scanned overflows pcregrep's buffer, an error is now
977 given and the return code is set to 2.
978
979 45. Add a pointer to the latest mark to the callout data block.
980
981 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
982 partial match of an empty string instead of no match. This was specific to
983 the use of ".".
984
985 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
986 complete match instead of a partial match. This bug was dependent on both
987 the PCRE_UTF8 and PCRE_DOTALL options being set.
988
989 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
990 starting byte set, because \b was not being ignored.
991
992
993 Version 8.12 15-Jan-2011
994 ------------------------
995
996 1. Fixed some typos in the markup of the man pages, and wrote a script that
997 checks for such things as part of the documentation building process.
998
999 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1000 --match-limit and --recursion-limit options (added for 8.11). In
1001 particular, this made one of the standard tests fail. (The integer value
1002 went into the wrong half of a long int.)
1003
1004 3. If the --colour option was given to pcregrep with -v (invert match), it
1005 did strange things, either producing crazy output, or crashing. It should,
1006 of course, ignore a request for colour when reporting lines that do not
1007 match.
1008
1009 4. Another pcregrep bug caused similar problems if --colour was specified with
1010 -M (multiline) and the pattern match finished with a line ending.
1011
1012 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1013 matched in multiline mode, the following line was shown as part of the
1014 match. This seems wrong, so I have changed it.
1015
1016 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1017 the check for further matches in the same line (so they could be coloured)
1018 to overrun the end of the current line. If another match was found, it was
1019 incorrectly shown (and then shown again when found in the next line).
1020
1021 7. If pcregrep was compiled under Windows, there was a reference to the
1022 function pcregrep_exit() before it was defined. I am assuming this was
1023 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1024 reported by a user. I've moved the definition above the reference.
1025
1026
1027 Version 8.11 10-Dec-2010
1028 ------------------------
1029
1030 1. (*THEN) was not working properly if there were untried alternatives prior
1031 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1032 backtracked to try for "b" instead of moving to the next alternative branch
1033 at the same level (in this case, to look for "c"). The Perl documentation
1034 is clear that when (*THEN) is backtracked onto, it goes to the "next
1035 alternative in the innermost enclosing group".
1036
1037 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1038 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1039 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1040 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1041 (*THEN).
1042
1043 3. If \s appeared in a character class, it removed the VT character from
1044 the class, even if it had been included by some previous item, for example
1045 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1046 of \s, but is part of the POSIX "space" class.)
1047
1048 4. A partial match never returns an empty string (because you can always
1049 match an empty string at the end of the subject); however the checking for
1050 an empty string was starting at the "start of match" point. This has been
1051 changed to the "earliest inspected character" point, because the returned
1052 data for a partial match starts at this character. This means that, for
1053 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1054 (previously it gave "no match").
1055
1056 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1057 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1058 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1059 has an implication that the given string is incomplete (because a partial
1060 match is preferred over a full match). For this reason, these items now
1061 give a partial match in this situation. [Aside: previously, the one case
1062 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1063 match rather than a full match, which was wrong by the old rules, but is
1064 now correct.]
1065
1066 6. There was a bug in the handling of #-introduced comments, recognized when
1067 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1068 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1069 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1070 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1071 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1072 places in pcre_compile().
1073
1074 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1075 comments when looking ahead for named forward references to subpatterns,
1076 the only newline sequence it recognized was NL. It now handles newlines
1077 according to the set newline convention.
1078
1079 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1080 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1081 cater for a lack of strerror(). These oversights have been fixed.
1082
1083 9. Added --match-limit and --recursion-limit to pcregrep.
1084
1085 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1086
1087 11. When the -o option was used, pcregrep was setting a return code of 1, even
1088 when matches were found, and --line-buffered was not being honoured.
1089
1090 12. Added an optional parentheses number to the -o and --only-matching options
1091 of pcregrep.
1092
1093 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1094 can match an empty string. The code to do it in pcretest and pcredemo
1095 needed fixing:
1096
1097 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1098 only one byte after an empty string match just before CRLF (this case
1099 just got forgotten; "any" and "anycrlf" were OK).
1100
1101 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1102 mode when an empty string match preceded an ASCII character followed by
1103 a non-ASCII character. (The code for advancing by one character rather
1104 than one byte was nonsense.)
1105
1106 (c) The pcredemo.c sample program did not have any code at all to handle
1107 the cases when CRLF is a valid newline sequence.
1108
1109 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1110 as a starting offset was within the subject string. There is now a new
1111 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1112 negative or greater than the length of the string. In order to test this,
1113 pcretest is extended to allow the setting of negative starting offsets.
1114
1115 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1116 starting offset points to the beginning of a UTF-8 character was
1117 unnecessarily clumsy. I tidied it up.
1118
1119 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1120 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1121
1122 17. Nobody had reported that the --include_dir option, which was added in
1123 release 7.7 should have been called --include-dir (hyphen, not underscore)
1124 for compatibility with GNU grep. I have changed it to --include-dir, but
1125 left --include_dir as an undocumented synonym, and the same for
1126 --exclude-dir, though that is not available in GNU grep, at least as of
1127 release 2.5.4.
1128
1129 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1130 characters from a string of bytes) have been redefined so as not to use
1131 loops, in order to improve performance in some environments. At the same
1132 time, I abstracted some of the common code into auxiliary macros to save
1133 repetition (this should not affect the compiled code).
1134
1135 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1136 compile-time error is now given if \c is not followed by an ASCII
1137 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1138 different, and any byte value is allowed.)
1139
1140 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1141 START_OPTIMIZE option, which is now allowed at compile time - but just
1142 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1143 to pcregrep and other applications that have no direct access to PCRE
1144 options. The new /Y option in pcretest sets this option when calling
1145 pcre_compile().
1146
1147 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1148 back references. Groups containing recursive back references were forced to
1149 be atomic by that change, but in the case of named groups, the amount of
1150 memory required was incorrectly computed, leading to "Failed: internal
1151 error: code overflow". This has been fixed.
1152
1153 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1154 pcretest.c, to avoid build problems in some Borland environments.
1155
1156
1157 Version 8.10 25-Jun-2010
1158 ------------------------
1159
1160 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1161 THEN.
1162
1163 2. (*ACCEPT) was not working when inside an atomic group.
1164
1165 3. Inside a character class, \B is treated as a literal by default, but
1166 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1167 causes the error). The code is unchanged, but I tidied the documentation.
1168
1169 4. Inside a character class, PCRE always treated \R and \X as literals,
1170 whereas Perl faults them if its -w option is set. I have changed PCRE so
1171 that it faults them when PCRE_EXTRA is set.
1172
1173 5. Added support for \N, which always matches any character other than
1174 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1175
1176 6. When compiling pcregrep with newer versions of gcc which may have
1177 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1178 declared with attribute warn_unused_result" were given. Just casting the
1179 result to (void) does not stop the warnings; a more elaborate fudge is
1180 needed. I've used a macro to implement this.
1181
1182 7. Minor change to pcretest.c to avoid a compiler warning.
1183
1184 8. Added four artifical Unicode properties to help with an option to make
1185 \s etc use properties (see next item). The new properties are: Xan
1186 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1187
1188 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1189 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1190 this option. Modified pcretest to add /W to test this facility. Added
1191 REG_UCP to make it available via the POSIX interface.
1192
1193 10. Added --line-buffered to pcregrep.
1194
1195 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1196 studied, and the match started with a letter with a code point greater than
1197 127 whose first byte was different to the first byte of the other case of
1198 the letter, the other case of this starting letter was not recognized
1199 (#976).
1200
1201 12. If a pattern that was studied started with a repeated Unicode property
1202 test, for example, \p{Nd}+, there was the theoretical possibility of
1203 setting up an incorrect bitmap of starting bytes, but fortunately it could
1204 not have actually happened in practice until change 8 above was made (it
1205 added property types that matched character-matching opcodes).
1206
1207 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1208 possible starting bytes for non-anchored patterns.
1209
1210 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1211 \R, and also a number of cases that involve Unicode properties, both
1212 explicit and implicit when PCRE_UCP is set.
1213
1214 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1215 input, it could crash or give wrong results if characters with values
1216 greater than 0xc0 were present in the subject string. (Detail: it assumed
1217 UTF-8 input when processing these items.)
1218
1219 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1220 size_t is 64-bit (#991).
1221
1222 17. Added a check for running out of memory when PCRE is compiled with
1223 --disable-stack-for-recursion (#990).
1224
1225 18. If the last data line in a file for pcretest does not have a newline on
1226 the end, a newline was missing in the output.
1227
1228 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1229 less than 128) in its various bitmaps. However, there is a facility for
1230 generating tables according to the current locale when PCRE is compiled. It
1231 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1232 characters, are recognized by isspace() and therefore were getting set in
1233 these tables, and indeed these tables seem to approximate to ISO 8859. This
1234 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1235 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1236 which of course cannot start UTF-8 characters. I have changed the code so
1237 that only real ASCII characters (less than 128) and the correct starting
1238 bytes for UTF-8 encodings are set for characters greater than 127 when in
1239 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1240 altogether.)
1241
1242 20. Added the /T option to pcretest so as to be able to run tests with non-
1243 standard character tables, thus making it possible to include the tests
1244 used for 19 above in the standard set of tests.
1245
1246 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1247 reference to a subpattern the other side of a comment that contains an
1248 opening parenthesis caused either an internal compiling error, or a
1249 reference to the wrong subpattern.
1250
1251
1252 Version 8.02 19-Mar-2010
1253 ------------------------
1254
1255 1. The Unicode data tables have been updated to Unicode 5.2.0.
1256
1257 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1258 configured.
1259
1260 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1261 original author of that file, following a query about its status.
1262
1263 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1264 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1265
1266 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1267 quantifier applied to a forward-referencing subroutine call, could compile
1268 incorrect code or give the error "internal error: previously-checked
1269 referenced subpattern not found".
1270
1271 6. Both MS Visual Studio and Symbian OS have problems with initializing
1272 variables to point to external functions. For these systems, therefore,
1273 pcre_malloc etc. are now initialized to local functions that call the
1274 relevant global functions.
1275
1276 7. There were two entries missing in the vectors called coptable and poptable
1277 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1278 I've fixed the data, and added a kludgy way of testing at compile time that
1279 the lengths are correct (equal to the number of opcodes).
1280
1281 8. Following on from 7, I added a similar kludge to check the length of the
1282 eint vector in pcreposix.c.
1283
1284 9. Error texts for pcre_compile() are held as one long string to avoid too
1285 much relocation at load time. To find a text, the string is searched,
1286 counting zeros. There was no check for running off the end of the string,
1287 which could happen if a new error number was added without updating the
1288 string.
1289
1290 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1291
1292 11. \K was not working if it appeared in an atomic group or in a group that
1293 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1294 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1295 the assertion is positive, but not if it is negative.
1296
1297 12. Change 11 fortuitously reduced the size of the stack frame used in the
1298 "match()" function of pcre_exec.c by one pointer. Forthcoming
1299 implementation of support for (*MARK) will need an extra pointer on the
1300 stack; I have reserved it now, so that the stack frame size does not
1301 decrease.
1302
1303 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1304 item in branch that calls a recursion is a subroutine call - as in the
1305 second branch in the above example - was incorrectly given the compile-
1306 time error "recursive call could loop indefinitely" because pcre_compile()
1307 was not correctly checking the subroutine for matching a non-empty string.
1308
1309 14. The checks for overrunning compiling workspace could trigger after an
1310 overrun had occurred. This is a "should never occur" error, but it can be
1311 triggered by pathological patterns such as hundreds of nested parentheses.
1312 The checks now trigger 100 bytes before the end of the workspace.
1313
1314 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1315
1316
1317 Version 8.01 19-Jan-2010
1318 ------------------------
1319
1320 1. If a pattern contained a conditional subpattern with only one branch (in
1321 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1322 computed the wrong minimum data length (which is of course zero for such
1323 subpatterns). This could cause incorrect "no match" results.
1324
1325 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1326 the pattern is reset in the first branch, pcre_compile() failed with
1327 "internal error: code overflow at offset...". This happened only when
1328 the reset was to the original external option setting. (An optimization
1329 abstracts leading options settings into an external setting, which was the
1330 cause of this.)
1331
1332 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1333 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1334 assertion pattern did not match (meaning that the assertion was true), it
1335 was incorrectly treated as false if the SKIP had been reached during the
1336 matching. This also applied to assertions used as conditions.
1337
1338 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1339 assertion subpattern, including such a pattern used as a condition,
1340 unpredictable results occurred, instead of the error return
1341 PCRE_ERROR_DFA_UITEM.
1342
1343 5. The C++ GlobalReplace function was not working like Perl for the special
1344 situation when an empty string is matched. It now does the fancy magic
1345 stuff that is necessary.
1346
1347 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1348 removed. (These were left over from very, very early versions of PCRE.)
1349
1350 7. Some cosmetic changes to the code to make life easier when compiling it
1351 as part of something else:
1352
1353 (a) Change DEBUG to PCRE_DEBUG.
1354
1355 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1356 called "current" as "current_branch", to prevent a collision with the
1357 Linux macro when compiled as a kernel module.
1358
1359 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1360 prevent a collision with the Linux macro when compiled as a kernel
1361 module.
1362
1363 8. In pcre_compile() there are some checks for integer overflows that used to
1364 cast potentially large values to (double). This has been changed to that
1365 when building, a check for int64_t is made, and if it is found, it is used
1366 instead, thus avoiding the use of floating point arithmetic. (There is no
1367 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1368 double.
1369
1370 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1371 2005 (difference between two addresses compared to an unsigned value).
1372
1373 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1374 custom one, because of the following reported problem in Windows:
1375
1376 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1377 under Win32.
1378 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1379 therefore missing the function definition.
1380 - The compiler thus generates a "C" signature for the test function.
1381 - The linker fails to find the "C" function.
1382 - PCRE fails to configure if asked to do so against libbz2.
1383
1384 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1385 messages were output:
1386
1387 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1388 rerunning libtoolize, to keep the correct libtool macros in-tree.
1389 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1390
1391 I have done both of these things.
1392
1393 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1394 most of the time, it *can* run out if it is given a pattern that contains a
1395 runaway infinite recursion. I updated the discussion in the pcrestack man
1396 page.
1397
1398 13. Now that we have gone to the x.xx style of version numbers, the minor
1399 version may start with zero. Using 08 or 09 is a bad idea because users
1400 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1401 interpreted as invalid octal numbers. I've updated the previous comment in
1402 configure.ac, and also added a check that gives an error if 08 or 09 are
1403 used.
1404
1405 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1406 causing partial matching to fail when the end of the subject matched \W
1407 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1408
1409 15. There were some discrepancies between the declarations in pcre_internal.h
1410 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1411 their definitions. The declarations used "const uschar *" and the
1412 definitions used USPTR. Even though USPTR is normally defined as "const
1413 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1414 reported that: "This difference in casting confuses some C++ compilers, for
1415 example, SunCC recognizes above declarations as different functions and
1416 generates broken code for hbpcre." I have changed the declarations to use
1417 USPTR.
1418
1419 16. GNU libtool is named differently on some systems. The autogen.sh script now
1420 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1421 (FreeBSD).
1422
1423 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1424 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1425 comment: "Figure out how to create a longlong from a string: strtoll and
1426 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1427 instance, but it only takes 2 args instead of 3!"
1428
1429 18. A subtle bug concerned with back references has been fixed by a change of
1430 specification, with a corresponding code fix. A pattern such as
1431 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1432 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1433 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1434 same bug. Such groups have to be quantified to be useful, or contained
1435 inside another quantified group. (If there's no repetition, the reference
1436 can never match.) The problem arises because, having left the group and
1437 moved on to the rest of the pattern, a later failure that backtracks into
1438 the group uses the captured value from the final iteration of the group
1439 rather than the correct earlier one. I have fixed this in PCRE by forcing
1440 any group that contains a reference to itself to be an atomic group; that
1441 is, there cannot be any backtracking into it once it has completed. This is
1442 similar to recursive and subroutine calls.
1443
1444
1445 Version 8.00 19-Oct-09
1446 ----------------------
1447
1448 1. The table for translating pcre_compile() error codes into POSIX error codes
1449 was out-of-date, and there was no check on the pcre_compile() error code
1450 being within the table. This could lead to an OK return being given in
1451 error.
1452
1453 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1454 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1455 in a Windows environment.
1456
1457 3. The pcregrep --count option prints the count for each file even when it is
1458 zero, as does GNU grep. However, pcregrep was also printing all files when
1459 --files-with-matches was added. Now, when both options are given, it prints
1460 counts only for those files that have at least one match. (GNU grep just
1461 prints the file name in this circumstance, but including the count seems
1462 more useful - otherwise, why use --count?) Also ensured that the
1463 combination -clh just lists non-zero counts, with no names.
1464
1465 4. The long form of the pcregrep -F option was incorrectly implemented as
1466 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1467 but it seems right to fix it, and I didn't think it was worth preserving
1468 the old behaviour.
1469
1470 5. The command line items --regex=pattern and --regexp=pattern were not
1471 recognized by pcregrep, which required --regex pattern or --regexp pattern
1472 (with a space rather than an '='). The man page documented the '=' forms,
1473 which are compatible with GNU grep; these now work.
1474
1475 6. No libpcreposix.pc file was created for pkg-config; there was just
1476 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1477
1478 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1479 when UCP support is not needed, by modifying the Python script that
1480 generates it from Unicode data files. This should not matter if the module
1481 is correctly used as a library, but I received one complaint about 50K of
1482 unwanted data. My guess is that the person linked everything into his
1483 program rather than using a library. Anyway, it does no harm.
1484
1485 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1486 was a minimum greater than 1 for a wide character in a possessive
1487 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1488 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1489 character. Chaos in the form of incorrect output or a compiling loop could
1490 result.
1491
1492 9. The restrictions on what a pattern can contain when partial matching is
1493 requested for pcre_exec() have been removed. All patterns can now be
1494 partially matched by this function. In addition, if there are at least two
1495 slots in the offset vector, the offset of the earliest inspected character
1496 for the match and the offset of the end of the subject are set in them when
1497 PCRE_ERROR_PARTIAL is returned.
1498
1499 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1500 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1501 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1502 and may be more useful for multi-segment matching.
1503
1504 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1505 used to be given if ever the end of the subject was reached; now it is
1506 given only if matching could not proceed because another character was
1507 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1508 string "Z", which now yields "no match" instead of "partial match". In the
1509 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1510 final character ended with (*FAIL).
1511
1512 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1513 if the pattern had a "must contain" character that was already found in the
1514 earlier partial match, unless partial matching was again requested. For
1515 example, with the pattern /dog.(body)?/, the "must contain" character is
1516 "g". If the first part-match was for the string "dog", restarting with
1517 "sbody" failed. This bug has been fixed.
1518
1519 13. The string returned by pcre_dfa_exec() after a partial match has been
1520 changed so that it starts at the first inspected character rather than the
1521 first character of the match. This makes a difference only if the pattern
1522 starts with a lookbehind assertion or \b or \B (\K is not supported by
1523 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1524 matching functions compatible, and I think it's the right thing to do.
1525
1526 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1527 so that the demonstration program is easily available in environments where
1528 PCRE has not been installed from source.
1529
1530 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1531 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1532 library.
1533
1534 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1535 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1536 is not the first non-POSIX option to be added. Clearly some people find
1537 these options useful.
1538
1539 17. If a caller to the POSIX matching function regexec() passes a non-zero
1540 value for nmatch with a NULL value for pmatch, the value of
1541 nmatch is forced to zero.
1542
1543 18. RunGrepTest did not have a test for the availability of the -u option of
1544 the diff command, as RunTest does. It now checks in the same way as
1545 RunTest, and also checks for the -b option.
1546
1547 19. If an odd number of negated classes containing just a single character
1548 interposed, within parentheses, between a forward reference to a named
1549 subpattern and the definition of the subpattern, compilation crashed with
1550 an internal error, complaining that it could not find the referenced
1551 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1552 [The bug was that it was starting one character too far in when skipping
1553 over the character class, thus treating the ] as data rather than
1554 terminating the class. This meant it could skip too much.]
1555
1556 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1557 /g option in pcretest when the pattern contains \K, which makes it possible
1558 to have an empty string match not at the start, even when the pattern is
1559 anchored. Updated pcretest and pcredemo to use this option.
1560
1561 21. If the maximum number of capturing subpatterns in a recursion was greater
1562 than the maximum at the outer level, the higher number was returned, but
1563 with unset values at the outer level. The correct (outer level) value is
1564 now given.
1565
1566 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1567 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1568 make it do so. The string so far is captured, making this feature
1569 compatible with Perl.
1570
1571 23. The tests have been re-organized, adding tests 11 and 12, to make it
1572 possible to check the Perl 5.10 features against Perl 5.10.
1573
1574 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1575 pattern matches a fixed length string. PCRE did not allow this; now it
1576 does. Neither allows recursion.
1577
1578 25. I finally figured out how to implement a request to provide the minimum
1579 length of subject string that was needed in order to match a given pattern.
1580 (It was back references and recursion that I had previously got hung up
1581 on.) This code has now been added to pcre_study(); it finds a lower bound
1582 to the length of subject needed. It is not necessarily the greatest lower
1583 bound, but using it to avoid searching strings that are too short does give
1584 some useful speed-ups. The value is available to calling programs via
1585 pcre_fullinfo().
1586
1587 26. While implementing 25, I discovered to my embarrassment that pcretest had
1588 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1589 study optimizations had never been tested with that matching function.
1590 Oops. What is worse, even when it was passed study data, there was a bug in
1591 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1592 were also very few tests of studied patterns with pcre_dfa_exec().
1593
1594 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1595 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1596 on the other side of the coin, they are no longer allowed to have different
1597 names, because these cannot be distinguished in PCRE, and this has caused
1598 confusion. (This is a difference from Perl.)
1599
1600 28. When duplicate subpattern names are present (necessarily with different
1601 numbers, as required by 27 above), and a test is made by name in a
1602 conditional pattern, either for a subpattern having been matched, or for
1603 recursion in such a pattern, all the associated numbered subpatterns are
1604 tested, and the overall condition is true if the condition is true for any
1605 one of them. This is the way Perl works, and is also more like the way
1606 testing by number works.
1607
1608
1609 Version 7.9 11-Apr-09
1610 ---------------------
1611
1612 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1613 (pcretest), all targets were linked against these libraries. This included
1614 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1615 libraries. This caused unwanted dependencies to be created. This problem
1616 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1617 pcretest is linked with readline.
1618
1619 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1620 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1621 moved outside it again, because FALSE and TRUE are already defined in AIX,
1622 but BOOL is not.
1623
1624 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1625 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1626
1627 4. The pcregrep documentation said spaces were inserted as well as colons (or
1628 hyphens) following file names and line numbers when outputting matching
1629 lines. This is not true; no spaces are inserted. I have also clarified the
1630 wording for the --colour (or --color) option.
1631
1632 5. In pcregrep, when --colour was used with -o, the list of matching strings
1633 was not coloured; this is different to GNU grep, so I have changed it to be
1634 the same.
1635
1636 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1637 each matching line was coloured. Now it goes on to look for further matches
1638 of any of the test patterns, which is the same behaviour as GNU grep.
1639
1640 7. A pattern that could match an empty string could cause pcregrep to loop; it
1641 doesn't make sense to accept an empty string match in pcregrep, so I have
1642 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1643 seems to be how GNU grep behaves.
1644
1645 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1646 start or after a newline", because the conditional assertion was not being
1647 correctly handled. The rule now is that both the assertion and what follows
1648 in the first alternative must satisfy the test.
1649
1650 9. If auto-callout was enabled in a pattern with a conditional group whose
1651 condition was an assertion, PCRE could crash during matching, both with
1652 pcre_exec() and pcre_dfa_exec().
1653
1654 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1655 used for matching.
1656
1657 11. Unicode property support in character classes was not working for
1658 characters (bytes) greater than 127 when not in UTF-8 mode.
1659
1660 12. Added the -M command line option to pcretest.
1661
1662 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1663
1664 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1665
1666 16. Added comments and documentation about mis-use of no_arg in the C++
1667 wrapper.
1668
1669 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1670 from Martin Jerabek that uses macro names for all relevant character and
1671 string constants.
1672
1673 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1674 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1675 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1676 these, but not everybody uses configure.
1677
1678 19. A conditional group that had only one branch was not being correctly
1679 recognized as an item that could match an empty string. This meant that an
1680 enclosing group might also not be so recognized, causing infinite looping
1681 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1682 with the subject "ab", where knowledge that the repeated group can match
1683 nothing is needed in order to break the loop.
1684
1685 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1686 exec(), but without supplying a callout function, matching went wrong.
1687
1688 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1689 leak if the size of the offset vector was greater than 30. When the vector
1690 is smaller, the saved offsets during recursion go onto a local stack
1691 vector, but for larger vectors malloc() is used. It was failing to free
1692 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1693 error, in fact).
1694
1695 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1696 heapframe that is used only when UTF-8 support is enabled. This caused no
1697 problem, but was untidy.
1698
1699 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1700 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1701 included within another project.
1702
1703 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1704 slightly modified by me:
1705
1706 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1707 not building pcregrep.
1708
1709 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1710 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1711
1712 25. Forward references, both numeric and by name, in patterns that made use of
1713 duplicate group numbers, could behave incorrectly or give incorrect errors,
1714 because when scanning forward to find the reference group, PCRE was not
1715 taking into account the duplicate group numbers. A pattern such as
1716 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1717
1718 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1719 the feature of a custom pointer more persuasive (as requested by a user).
1720
1721 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1722 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1723 pre-defined.
1724
1725 28. Added support for (*UTF8) at the start of a pattern.
1726
1727 29. Arrange for flags added by the "release type" setting in CMake to be shown
1728 in the configuration summary.
1729
1730
1731 Version 7.8 05-Sep-08
1732 ---------------------
1733
1734 1. Replaced UCP searching code with optimized version as implemented for Ad
1735 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1736 stage table and inline lookup instead of a function, giving speed ups of 2
1737 to 5 times on some simple patterns that I tested. Permission was given to
1738 distribute the MultiStage2.py script that generates the tables (it's not in
1739 the tarball, but is in the Subversion repository).
1740
1741 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1742 scripts.
1743
1744 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1745 a group with a zero qualifier. The result of the study could be incorrect,
1746 or the function might crash, depending on the pattern.
1747
1748 4. Caseless matching was not working for non-ASCII characters in back
1749 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1750 It now works when Unicode Property Support is available.
1751
1752 5. In pcretest, an escape such as \x{de} in the data was always generating
1753 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1754 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1755 truncation.
1756
1757 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1758
1759 7. Added two (int) casts to pcregrep when printing the difference of two
1760 pointers, in case they are 64-bit values.
1761
1762 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1763 test 2 if it fails.
1764
1765 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1766 and a #define of that name to empty if it is not externally set. This is to
1767 allow users of MSVC to set it if necessary.
1768
1769 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1770 the convenience functions in the pcre_get.c source file.
1771
1772 11. An option change at the start of a pattern that had top-level alternatives
1773 could cause overwriting and/or a crash. This command provoked a crash in
1774 some environments:
1775
1776 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1777
1778 This potential security problem was recorded as CVE-2008-2371.
1779
1780 12. For a pattern where the match had to start at the beginning or immediately
1781 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1782 pcre_dfa_exec() could read past the end of the passed subject if there was
1783 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1784 pcretest so that it places the subject at the end of its malloc-ed buffer.
1785
1786 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1787 exec() might read past the end of the data buffer in UTF-8 mode.
1788
1789 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1790 the data contained the byte 0x85 as part of a UTF-8 character within its
1791 first line. This applied both to normal and DFA matching.
1792
1793 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1794 /^[^d]*?$/8 failed to match "abc".
1795
1796 16. Added a missing copyright notice to pcrecpp_internal.h.
1797
1798 17. Make it more clear in the documentation that values returned from
1799 pcre_exec() in ovector are byte offsets, not character counts.
1800
1801 18. Tidied a few places to stop certain compilers from issuing warnings.
1802
1803 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1804 supplied by Stefan Weber. I made a further small update for 7.8 because
1805 there is a change of source arrangements: the pcre_searchfuncs.c module is
1806 replaced by pcre_ucd.c.
1807
1808
1809 Version 7.7 07-May-08
1810 ---------------------
1811
1812 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1813 a string to a long long, pretend we don't even have a long long." This is
1814 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1815
1816 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1817 pre-7.6 versions, which defined a global no_arg variable instead of putting
1818 it in the RE class. (See also #8 below.)
1819
1820 3. Remove a line of dead code, identified by coverity and reported by Nuno
1821 Lopes.
1822
1823 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1824
1825 (1) The include/exclude patterns were being applied to the whole pathnames
1826 of files, instead of just to the final components.
1827
1828 (2) If there was more than one level of directory, the subdirectories were
1829 skipped unless they satisfied the include/exclude conditions. This is
1830 inconsistent with GNU grep (and could even be seen as contrary to the
1831 pcregrep specification - which I improved to make it absolutely clear).
1832 The action now is always to scan all levels of directory, and just
1833 apply the include/exclude patterns to regular files.
1834
1835 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1836 --exclude_dir in the tests to avoid scanning .svn directories.
1837
1838 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1839 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1840 doesn't support NULs in patterns.
1841
1842 7. Added some missing "const"s to declarations of static tables in
1843 pcre_compile.c and pcre_dfa_exec.c.
1844
1845 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1846 caused by fix #2 above. (Subsequently also a second patch to fix the
1847 first patch. And a third patch - this was a messy problem.)
1848
1849 9. Applied Craig's patch to remove the use of push_back().
1850
1851 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1852 matching function regexec().
1853
1854 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1855 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1856 references. PCRE supports relative numbers with this syntax (I don't think
1857 Oniguruma does).
1858
1859 12. Previously, a group with a zero repeat such as (...){0} was completely
1860 omitted from the compiled regex. However, this means that if the group
1861 was called as a subroutine from elsewhere in the pattern, things went wrong
1862 (an internal error was given). Such groups are now left in the compiled
1863 pattern, with a new opcode that causes them to be skipped at execution
1864 time.
1865
1866 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1867 to the way PCRE behaves:
1868
1869 (a) A lone ] character is dis-allowed (Perl treats it as data).
1870
1871 (b) A back reference to an unmatched subpattern matches an empty string
1872 (Perl fails the current match path).
1873
1874 (c) A data ] in a character class must be notated as \] because if the
1875 first data character in a class is ], it defines an empty class. (In
1876 Perl it is not possible to have an empty class.) The empty class []
1877 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1878 The negative empty class [^] matches any one character, independently
1879 of the DOTALL setting.
1880
1881 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1882 non-existent subpattern following a character class starting with ']' and
1883 containing () gave an internal compiling error instead of "reference to
1884 non-existent subpattern". Fortunately, when the pattern did exist, the
1885 compiled code was correct. (When scanning forwards to check for the
1886 existencd of the subpattern, it was treating the data ']' as terminating
1887 the class, so got the count wrong. When actually compiling, the reference
1888 was subsequently set up correctly.)
1889
1890 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1891 it was being rejected as not supported by pcre_dfa_exec(), even though
1892 other assertions are supported. I have made pcre_dfa_exec() support
1893 (*FAIL).
1894
1895 16. The implementation of 13c above involved the invention of a new opcode,
1896 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1897 cannot be changed at match time, I realized I could make a small
1898 improvement to matching performance by compiling OP_ALLANY instead of
1899 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1900 on the OP_ANY path.
1901
1902 17. Compiling pcretest on Windows with readline support failed without the
1903 following two fixes: (1) Make the unistd.h include conditional on
1904 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1905
1906 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1907 ncurses library to be included for pcretest when ReadLine support is
1908 requested, but also to allow for it to be overridden. This patch came from
1909 Daniel Bergström.
1910
1911 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1912 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1913 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1914 spotting this.
1915
1916
1917 Version 7.6 28-Jan-08
1918 ---------------------
1919
1920 1. A character class containing a very large number of characters with
1921 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1922 overflow.
1923
1924 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1925 HAVE_LONG_LONG is not defined.
1926
1927 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1928 bring them up to date and include new features. This patch includes:
1929
1930 - Fixed PH's badly added libz and libbz2 support.
1931 - Fixed a problem with static linking.
1932 - Added pcredemo. [But later removed - see 7 below.]
1933 - Fixed dftables problem and added an option.
1934 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1935 HAVE_LONG_LONG.
1936 - Added readline support for pcretest.
1937 - Added an listing of the option settings after cmake has run.
1938
1939 4. A user submitted a patch to Makefile that makes it easy to create
1940 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1941 Makefile.am that cause it to include this special target, without
1942 affecting anything else. Note that the same mingw target plus all
1943 the other distribution libraries and programs are now supported
1944 when configuring with CMake (see 6 below) instead of with
1945 Configure/Make.
1946
1947 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1948 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1949 exported in the Windows port". It has not yet been confirmed that the patch
1950 solves the problem, but it does no harm.
1951
1952 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1953 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1954 with CMake, and also correct the comment about stack recursion.
1955
1956 7. Remove the automatic building of pcredemo from the ./configure system and
1957 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1958 of a program that users should build themselves after PCRE is installed, so
1959 building it automatically is not really right. What is more, it gave
1960 trouble in some build environments.
1961
1962 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1963
1964
1965 Version 7.5 10-Jan-08
1966 ---------------------
1967
1968 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1969 values in parens when parsing an RE using the C++ wrapper."
1970
1971 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1972 Characters greater than 255 were excluded from the class instead of being
1973 included.
1974
1975 3. The same bug as (2) above applied to negated POSIX classes such as
1976 [:^space:].
1977
1978 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1979 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1980 I have changed it.
1981
1982 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1983 first named subpattern) and a construct such as (?&a) would reference the
1984 first named subpattern whose name started with "a" (in other words, the
1985 length check was missing). Both these problems are fixed. "Subpattern name
1986 expected" is now given for (?&) (a zero-length name), and this patch also
1987 makes it give the same error for \k'' (previously it complained that that
1988 was a reference to a non-existent subpattern).
1989
1990 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1991 this is right because (?- can be followed by option settings as well as by
1992 digits. I have, however, made the messages clearer.
1993
1994 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1995 than the number used in the conditional) now cause a compile-time error.
1996 This is actually not compatible with Perl, which accepts such patterns, but
1997 treats the conditional as always being FALSE (as PCRE used to), but it
1998 seems to me that giving a diagnostic is better.
1999
2000 8. Change "alphameric" to the more common word "alphanumeric" in comments
2001 and messages.
2002
2003 9. Fix two occurrences of "backslash" in comments that should have been
2004 "backspace".
2005
2006 10. Remove two redundant lines of code that can never be obeyed (their function
2007 was moved elsewhere).
2008
2009 11. The program that makes PCRE's Unicode character property table had a bug
2010 which caused it to generate incorrect table entries for sequences of
2011 characters that have the same character type, but are in different scripts.
2012 It amalgamated them into a single range, with the script of the first of
2013 them. In other words, some characters were in the wrong script. There were
2014 thirteen such cases, affecting characters in the following ranges:
2015
2016 U+002b0 - U+002c1
2017 U+0060c - U+0060d
2018 U+0061e - U+00612
2019 U+0064b - U+0065e
2020 U+0074d - U+0076d
2021 U+01800 - U+01805
2022 U+01d00 - U+01d77
2023 U+01d9b - U+01dbf
2024 U+0200b - U+0200f
2025 U+030fc - U+030fe
2026 U+03260 - U+0327f
2027 U+0fb46 - U+0fbb1
2028 U+10450 - U+1049d
2029
2030 12. The -o option (show only the matching part of a line) for pcregrep was not
2031 compatible with GNU grep in that, if there was more than one match in a
2032 line, it showed only the first of them. It now behaves in the same way as
2033 GNU grep.
2034
2035 13. If the -o and -v options were combined for pcregrep, it printed a blank
2036 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2037 does the same. The return code can be used to tell if there were any
2038 non-matching lines.
2039
2040 14. Added --file-offsets and --line-offsets to pcregrep.
2041
2042 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2043 infinitely looping recursion. The bug was that positive lookaheads were not
2044 being skipped when checking for a possible empty match (negative lookaheads
2045 and both kinds of lookbehind were skipped).
2046
2047 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2048 inclusion of <windows.h> to before rather than after the definition of
2049 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2050
2051 17. Specifying a possessive quantifier with a specific limit for a Unicode
2052 character property caused pcre_compile() to compile bad code, which led at
2053 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2054 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2055 caused the error; without that there was no problem.
2056
2057 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2058
2059 19. Added --enable-pcretest-libreadline.
2060
2061 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2062 RE::GlobalReplace(). As a result, the number of replacements returned was
2063 double what it should be. I removed one of the increments, but Craig sent a
2064 later patch that removed the other one (the right fix) and added unit tests
2065 that check the return values (which was not done before).
2066
2067 21. Several CMake things:
2068
2069 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2070 the names libpcre and libpcreposix, not just pcre and pcreposix.
2071
2072 (2) The above change means that pcretest and pcregrep are now correctly
2073 linked with the newly-built libraries, not previously installed ones.
2074
2075 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2076
2077 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2078 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2079 UTF-8 newline character). The key issue is that the pattern starts .*;
2080 this means that the match must be either at the beginning, or after a
2081 newline. The bug was in the code for advancing after a failed match and
2082 checking that the new position followed a newline. It was not taking
2083 account of UTF-8 characters correctly.
2084
2085 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2086 character classes. PCRE was not treating the sequence [:...:] as a
2087 character class unless the ... were all letters. Perl, however, seems to
2088 allow any characters between [: and :], though of course it rejects as
2089 unknown any "names" that contain non-letters, because all the known class
2090 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2091 for example, whereas PCRE did not - it did not recognize a POSIX character
2092 class. This seemed a bit dangerous, so the code has been changed to be
2093 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2094 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2095 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2096 Perl does, and where it didn't before.
2097
2098 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2099 Windows environments %n is disabled by default.
2100
2101
2102 Version 7.4 21-Sep-07
2103 ---------------------
2104
2105 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2106 means that a class such as [\s] counted as "explicit reference to CR or
2107 LF". That isn't really right - the whole point of the change was to try to
2108 help when there was an actual mention of one of the two characters. So now
2109 the change happens only if \r or \n (or a literal CR or LF) character is
2110 encountered.
2111
2112 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2113 of both had grown to the point where there were only 3 bits left.
2114 Fortunately, there was spare space in the data structure, and so I have
2115 moved the internal flags into a new 16-bit field to free up more option
2116 bits.
2117
2118 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2119 but did not set the internal JCHANGED flag - either of these is enough to
2120 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2121 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2122 start sets both bits.
2123
2124 4. Added options (at build time, compile time, exec time) to change \R from
2125 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2126
2127 5. doc/pcresyntax.html was missing from the distribution.
2128
2129 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2130 compatibility, even though it is no longer used.
2131
2132 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2133 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2134 windows.h file is present (where different names are used). [This was
2135 reversed later after testing - see 16 below.]
2136
2137 8. Changed all #include <config.h> to #include "config.h". There were also
2138 some further <pcre.h> cases that I changed to "pcre.h".
2139
2140 9. When pcregrep was used with the --colour option, it missed the line ending
2141 sequence off the lines that it output.
2142
2143 10. It was pointed out to me that arrays of string pointers cause lots of
2144 relocations when a shared library is dynamically loaded. A technique of
2145 using a single long string with a table of offsets can drastically reduce
2146 these. I have refactored PCRE in four places to do this. The result is
2147 dramatic:
2148
2149 Originally: 290
2150 After changing UCP table: 187
2151 After changing error message table: 43
2152 After changing table of "verbs" 36
2153 After changing table of Posix names 22
2154
2155 Thanks to the folks working on Gregex for glib for this insight.
2156
2157 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2158 unicode-properties was also set.
2159
2160 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2161
2162 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2163 checked only for CRLF.
2164
2165 14. Added casts to pcretest.c to avoid compiler warnings.
2166
2167 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2168
2169 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2170 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2171 entirely. This removes changes made in 7 above.
2172
2173 17. The CMake files have been updated, and there is now more information about
2174 building with CMake in the NON-UNIX-USE document.
2175
2176
2177 Version 7.3 28-Aug-07
2178 ---------------------
2179
2180 1. In the rejigging of the build system that eventually resulted in 7.1, the
2181 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2182 brackets there is not right, since it causes compilers to look for an
2183 installed pcre.h, not the version that is in the source that is being
2184 compiled (which of course may be different). I have changed it back to:
2185
2186 #include "pcre.h"
2187
2188 I have a vague recollection that the change was concerned with compiling in
2189 different directories, but in the new build system, that is taken care of
2190 by the VPATH setting the Makefile.
2191
2192 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2193 when the subject happened to end in the byte 0x85 (e.g. if the last
2194 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2195 characters but of course it shouldn't be taken as a newline when it is part
2196 of another character. The bug was that, for an unlimited repeat of . in
2197 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2198 characters when looking for a newline.
2199
2200 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2201
2202 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2203 in debug output.
2204
2205 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2206 long printing in the pcrecpp unittest when running under MinGW.
2207
2208 6. ESC_K was left out of the EBCDIC table.
2209
2210 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2211 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2212 limit also applies to "virtual nesting" when a pattern is recursive, and in
2213 this case 1000 isn't so big. I have been able to remove this limit at the
2214 expense of backing off one optimization in certain circumstances. Normally,
2215 when pcre_exec() would call its internal match() function recursively and
2216 immediately return the result unconditionally, it uses a "tail recursion"
2217 feature to save stack. However, when a subpattern that can match an empty
2218 string has an unlimited repetition quantifier, it no longer makes this
2219 optimization. That gives it a stack frame in which to save the data for
2220 checking that an empty string has been matched. Previously this was taken
2221 from the 1000-entry workspace that had been reserved. So now there is no
2222 explicit limit, but more stack is used.
2223
2224 8. Applied Daniel's patches to solve problems with the import/export magic
2225 syntax that is required for Windows, and which was going wrong for the
2226 pcreposix and pcrecpp parts of the library. These were overlooked when this
2227 problem was solved for the main library.
2228
2229 9. There were some crude static tests to avoid integer overflow when computing
2230 the size of patterns that contain repeated groups with explicit upper
2231 limits. As the maximum quantifier is 65535, the maximum group length was
2232 set at 30,000 so that the product of these two numbers did not overflow a
2233 32-bit integer. However, it turns out that people want to use groups that
2234 are longer than 30,000 bytes (though not repeat them that many times).
2235 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2236 made it possible to implement the integer overflow checks in a much more
2237 dynamic way, which I have now done. The artificial limitation on group
2238 length has been removed - we now have only the limit on the total length of
2239 the compiled pattern, which depends on the LINK_SIZE setting.
2240
2241 10. Fixed a bug in the documentation for get/copy named substring when
2242 duplicate names are permitted. If none of the named substrings are set, the
2243 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2244 empty string.
2245
2246 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2247 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2248 because the ] is interpreted as the first data character and the
2249 terminating ] is not found. PCRE has been made compatible with Perl in this
2250 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2251 cause memory overwriting.
2252
2253 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2254 string has been matched (to stop an infinite loop). It was not recognizing
2255 a conditional subpattern that could match an empty string if that
2256 subpattern was within another subpattern. For example, it looped when
2257 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2258 condition was not nested. This bug has been fixed.
2259
2260 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2261 past the start of the subject in the presence of bytes with the top bit
2262 set, for example "\x8aBCD".
2263
2264 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2265 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2266
2267 14. Optimized (?!) to (*FAIL).
2268
2269 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2270 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2271 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2272 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2273 does: it's just the validity check that is more restrictive.
2274
2275 16. Inserted checks for integer overflows during escape sequence (backslash)
2276 processing, and also fixed erroneous offset values for syntax errors during
2277 backslash processing.
2278
2279 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2280 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2281
2282 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2283 caused an overrun.
2284
2285 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2286 something other than just ASCII characters) inside a group that had an
2287 unlimited repeat caused a loop at compile time (while checking to see
2288 whether the group could match an empty string).
2289
2290 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2291 [\P{Any}] did so. (Error in the code for printing property names.)
2292
2293 21. An orphan \E inside a character class could cause a crash.
2294
2295 22. A repeated capturing bracket such as (A)? could cause a wild memory
2296 reference during compilation.
2297
2298 23. There are several functions in pcre_compile() that scan along a compiled
2299 expression for various reasons (e.g. to see if it's fixed length for look
2300 behind). There were bugs in these functions when a repeated \p or \P was
2301 present in the pattern. These operators have additional parameters compared
2302 with \d, etc, and these were not being taken into account when moving along
2303 the compiled data. Specifically:
2304
2305 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2306 length.
2307
2308 (b) An item such as \pL+ within a repeated group could cause crashes or
2309 loops.
2310
2311 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2312 "reference to non-existent subpattern" error.
2313
2314 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2315
2316 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2317 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2318
2319 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2320
2321 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2322 character were causing crashes (broken optimization).
2323
2324 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2325 \p or \P) caused a compile-time loop.
2326
2327 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2328 break. For example, the unstudied pattern [\r\n]A does not match the string
2329 "\r\nA" because change 7.0/46 below moves the current point on by two
2330 characters after failing to match at the start. However, the pattern \nA
2331 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2332 the same is true. There doesn't seem any very clean way out of this, but
2333 what I have chosen to do makes the common cases work: PCRE now takes note
2334 of whether there can be an explicit match for \r or \n anywhere in the
2335 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2336 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2337 pattern has explicit CR or LF references.
2338
2339 29. Added (*CR) etc for changing newline setting at start of pattern.
2340
2341
2342 Version 7.2 19-Jun-07
2343 ---------------------
2344
2345 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2346 which is apparently normally available under Windows.
2347
2348 2. Re-jig the pcregrep tests with different newline settings in an attempt
2349 to make them independent of the local environment's newline setting.
2350
2351 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2352
2353 4. Some of the "internals" tests were previously cut out when the link size
2354 was not 2, because the output contained actual offsets. The recent new
2355 "Z" feature of pcretest means that these can be cut out, making the tests
2356 usable with all link sizes.
2357
2358 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2359 stack recursion. This gives a massive performance boost under BSD, but just
2360 a small improvement under Linux. However, it saves one field in the frame
2361 in all cases.
2362
2363 6. Added more features from the forthcoming Perl 5.10:
2364
2365 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2366 recursion call. It refers to the nth most recently opened parentheses.
2367
2368 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2369 to be opened parentheses.
2370
2371 (c) Conditions that refer to capturing parentheses can be specified
2372 relatively, for example, (?(-2)... or (?(+3)...
2373
2374 (d) \K resets the start of the current match so that everything before
2375 is not part of it.
2376
2377 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2378
2379 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2380 reference syntax.
2381
2382 (g) (?| introduces a group in which the numbering of parentheses in each
2383 alternative starts with the same number.
2384
2385 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2386
2387 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2388 PCRE_INFO_JCHANGED.
2389
2390 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2391 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2392 for detecting groups that can match an empty string.
2393
2394 9. A pattern with a very large number of alternatives (more than several
2395 hundred) was running out of internal workspace during the pre-compile
2396 phase, where pcre_compile() figures out how much memory will be needed. A
2397 bit of new cunning has reduced the workspace needed for groups with
2398 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2399 workspace instead of running out of the 4096 that are available.
2400
2401 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2402
2403 11. Applied patch from Google to remove an optimization that didn't quite work.
2404 The report of the bug said:
2405
2406 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2407 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2408 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2409
2410 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2411 it matched the wrong number of bytes.
2412
2413
2414 Version 7.1 24-Apr-07
2415 ---------------------
2416
2417 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2418 that is more "standard", making use of automake and other Autotools. There
2419 is some re-arrangement of the files and adjustment of comments consequent
2420 on this.
2421
2422 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2423 for recursive directory scanning broke on some systems because the files
2424 are not scanned in any specific order and on different systems the order
2425 was different. A call to "sort" has been inserted into RunGrepTest for the
2426 approprate test as a short-term fix. In the longer term there may be an
2427 alternative.
2428
2429 3. I had an email from Eric Raymond about problems translating some of PCRE's
2430 man pages to HTML (despite the fact that I distribute HTML pages, some
2431 people do their own conversions for various reasons). The problems
2432 concerned the use of low-level troff macros .br and .in. I have therefore
2433 removed all such uses from the man pages (some were redundant, some could
2434 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2435 HTML has been updated to handle .nf/.fi and to complain if it encounters
2436 .br or .in.
2437
2438 4. Updated comments in configure.ac that get placed in config.h.in and also
2439 arranged for config.h to be included in the distribution, with the name
2440 config.h.generic, for the benefit of those who have to compile without
2441 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2442
2443 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2444 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2445 makevp.bat for the current PCRE, using the additional files
2446 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2447
2448 6. A Windows user reported a minor discrepancy with test 2, which turned out
2449 to be caused by a trailing space on an input line that had got lost in his
2450 copy. The trailing space was an accident, so I've just removed it.
2451
2452 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2453 that is needed.
2454
2455 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2456 as "const" (a) because they are and (b) because it helps the PHP
2457 maintainers who have recently made a script to detect big data structures
2458 in the php code that should be moved to the .rodata section. I remembered
2459 to update Builducptable as well, so it won't revert if ucptable.h is ever
2460 re-created.
2461
2462 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2463 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2464 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2465 support is not required. This saves 1.5-2K of code, which is important in
2466 some applications.
2467
2468 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2469 so as not to refer to the tables, even though these functions will never be
2470 called when UTF-8 support is disabled. Otherwise there are problems with a
2471 shared library.
2472
2473 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2474
2475 (a) It was defining its arguments as char * instead of void *.
2476
2477 (b) It was assuming that all moves were upwards in memory; this was true
2478 a long time ago when I wrote it, but is no longer the case.
2479
2480 The emulated memove() is provided for those environments that have neither
2481 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2482 is clearly not the case, as these two bugs were recently reported.
2483
2484 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2485 and Detrail to create the HTML documentation, the .txt form of the man
2486 pages, and it removes trailing spaces from listed files. It also creates
2487 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2488 case, it wraps all the #defines with #ifndefs. This script should be run
2489 before "make dist".
2490
2491 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2492 with Unicode property support.
2493
2494 (a) For a maximizing quantifier, if the two different cases of the
2495 character were of different lengths in their UTF-8 codings (there are
2496 some cases like this - I found 11), and the matching function had to
2497 back up over a mixture of the two cases, it incorrectly assumed they
2498 were both the same length.
2499
2500 (b) When PCRE was configured to use the heap rather than the stack for
2501 recursion during matching, it was not correctly preserving the data for
2502 the other case of a UTF-8 character when checking ahead for a match
2503 while processing a minimizing repeat. If the check also involved
2504 matching a wide character, but failed, corruption could cause an
2505 erroneous result when trying to check for a repeat of the original
2506 character.
2507
2508 13. Some tidying changes to the testing mechanism:
2509
2510 (a) The RunTest script now detects the internal link size and whether there
2511 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2512 values substituted by "configure". (The RunGrepTest script already did
2513 this for UTF-8.) The configure.ac script no longer substitutes the
2514 relevant variables.
2515
2516 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2517 with length and offset values. This means that the output is different
2518 for different internal link sizes. Test 2 is skipped for link sizes
2519 other than 2 because of this, bypassing the problem. Unfortunately,
2520 there was also a test in test 3 (the locale tests) that used /B and
2521 failed for link sizes other than 2. Rather than cut the whole test out,
2522 I have added a new /Z option to pcretest that replaces the length and
2523 offset values with spaces. This is now used to make test 3 independent
2524 of link size. (Test 2 will be tidied up later.)
2525
2526 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2527 segmentation fault instead of returning the appropriate error message.
2528
2529 15. In multiline mode when the newline sequence was set to "any", the pattern
2530 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2531 This doesn't seem right; it now treats the CRLF combination as the line
2532 ending, and so does not match in that case. It's only a pattern such as ^$
2533 that would hit this one: something like ^ABC$ would have failed after \r
2534 and then tried again after \r\n.
2535
2536 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2537 in an attempt to make files that differ only in their line terminators
2538 compare equal. This works on Linux.
2539
2540 17. Under certain error circumstances pcregrep might try to free random memory
2541 as it exited. This is now fixed, thanks to valgrind.
2542
2543 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2544 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2545 was because its rules for how to advance for /g after matching an empty
2546 string at the end of a line did not allow for this case. They now check for
2547 it specially.
2548
2549 20. pcretest is supposed to handle patterns and data of any length, by
2550 extending its buffers when necessary. It was getting this wrong when the
2551 buffer for a data line had to be extended.
2552
2553 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2554 CRLF as a newline sequence.
2555
2556 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2557 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2558 I have nevertheless tidied it up.
2559
2560 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2561
2562 24. Added a man page for pcre-config.
2563
2564
2565 Version 7.0 19-Dec-06
2566 ---------------------
2567
2568 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2569 moving to gcc 4.1.1.
2570
2571 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2572 sys/time.h, which is documented as needed for this function. It doesn't
2573 seem to matter on Linux, but it showed up on some releases of OS X.
2574
2575 3. It seems that there are systems where bytes whose values are greater than
2576 127 match isprint() in the "C" locale. The "C" locale should be the
2577 default when a C program starts up. In most systems, only ASCII printing
2578 characters match isprint(). This difference caused the output from pcretest
2579 to vary, making some of the tests fail. I have changed pcretest so that:
2580
2581 (a) When it is outputting text in the compiled version of a pattern, bytes
2582 other than 32-126 are always shown as hex escapes.
2583
2584 (b) When it is outputting text that is a matched part of a subject string,
2585 it does the same, unless a different locale has been set for the match
2586 (using the /L modifier). In this case, it uses isprint() to decide.
2587
2588 4. Fixed a major bug that caused incorrect computation of the amount of memory
2589 required for a compiled pattern when options that changed within the
2590 pattern affected the logic of the preliminary scan that determines the
2591 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2592 that the computed length was too small. The symptoms of this bug were
2593 either the PCRE error "internal error: code overflow" from pcre_compile(),
2594 or a glibc crash with a message such as "pcretest: free(): invalid next
2595 size (fast)". Examples of patterns that provoked this bug (shown in
2596 pcretest format) are:
2597
2598 /(?-x: )/x
2599 /(?x)(?-x: \s*#\s*)/
2600 /((?i)[\x{c0}])/8
2601 /(?i:[\x{c0}])/8
2602
2603 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2604 is now done differently.
2605
2606 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2607 wrapper classes; (b) implement a new function in the C++ scanner that is
2608 more efficient than the old way of doing things because it avoids levels of
2609 recursion in the regex matching; (c) add a paragraph to the documentation
2610 for the FullMatch() function.
2611
2612 6. The escape sequence \n was being treated as whatever was defined as
2613 "newline". Not only was this contrary to the documentation, which states
2614 that \n is character 10 (hex 0A), but it also went horribly wrong when
2615 "newline" was defined as CRLF. This has been fixed.
2616
2617 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2618 was being set to -1 for the "end of line" case (supposedly a value that no
2619 character can have). Though this value is never used (the check for end of
2620 line is "zero bytes in current character"), it caused compiler complaints.
2621 I've changed it to 0xffffffff.
2622
2623 8. In pcre_version.c, the version string was being built by a sequence of
2624 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2625 string (as it is for production releases) called a macro with an empty
2626 argument. The C standard says the result of this is undefined. The gcc
2627 compiler treats it as an empty string (which was what was wanted) but it is
2628 reported that Visual C gives an error. The source has been hacked around to
2629 avoid this problem.
2630
2631 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2632 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2633 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2634 of them did).
2635
2636 10. Originally, pcretest opened its input and output without "b"; then I was
2637 told that "b" was needed in some environments, so it was added for release
2638 5.0 to both the input and output. (It makes no difference on Unix-like
2639 systems.) Later I was told that it is wrong for the input on Windows. I've
2640 now abstracted the modes into two macros, to make it easier to fiddle with
2641 them, and removed "b" from the input mode under Windows.
2642
2643 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2644
2645 12. Added -help and --help to pcretest as an official way of being reminded
2646 of the options.
2647
2648 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2649 and pcrecpp.cc because they annoy compilers at high warning levels.
2650
2651 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2652
2653 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2654 scripts are not C programs :-) and which was not noticed because it works
2655 on Linux.
2656
2657 16. pcretest is supposed to handle any length of pattern and data line (as one
2658 line or as a continued sequence of lines) by extending its input buffer if
2659 necessary. This feature was broken for very long pattern lines, leading to
2660 a string of junk being passed to pcre_compile() if the pattern was longer
2661 than about 50K.
2662
2663 17. I have done a major re-factoring of the way pcre_compile() computes the
2664 amount of memory needed for a compiled pattern. Previously, there was code
2665 that made a preliminary scan of the pattern in order to do this. That was
2666 OK when PCRE was new, but as the facilities have expanded, it has become
2667 harder and harder to keep it in step with the real compile phase, and there
2668 have been a number of bugs (see for example, 4 above). I have now found a
2669 cunning way of running the real compile function in a "fake" mode that
2670 enables it to compute how much memory it would need, while actually only
2671 ever using a few hundred bytes of working memory and without too many
2672 tests of the mode. This should make future maintenance and development
2673 easier. A side effect of this work is that the limit of 200 on the nesting
2674 depth of parentheses has been removed (though this was never a serious
2675 limitation, I suspect). However, there is a downside: pcre_compile() now
2676 runs more slowly than before (30% or more, depending on the pattern). I
2677 hope this isn't a big issue. There is no effect on runtime performance.
2678
2679 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2680 newline (only possible for the last line of a file) and it was a
2681 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2682
2683 19. Added additional timing features to pcretest. (1) The -tm option now times
2684 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2685 separate command line item, by a number that specifies the number of
2686 repeats to use when timing. The default is 50000; this gives better
2687 precision, but takes uncomfortably long for very large patterns.
2688
2689 20. Extended pcre_study() to be more clever in cases where a branch of a
2690 subpattern has no definite first character. For example, (a*|b*)[cd] would
2691 previously give no result from pcre_study(). Now it recognizes that the
2692 first character must be a, b, c, or d.
2693
2694 21. There was an incorrect error "recursive call could loop indefinitely" if
2695 a subpattern (or the entire pattern) that was being tested for matching an
2696 empty string contained only one non-empty item after a nested subpattern.
2697 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2698 incorrectly, because the \d was being skipped in the check.
2699
2700 22. The pcretest program now has a new pattern option /B and a command line
2701 option -b, which is equivalent to adding /B to every pattern. This causes
2702 it to show the compiled bytecode, without the additional information that
2703 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2704 is the same as /B/I).
2705
2706 23. A new optimization is now able automatically to treat some sequences such
2707 as a*b as a*+b. More specifically, if something simple (such as a character
2708 or a simple class like \d) has an unlimited quantifier, and is followed by
2709 something that cannot possibly match the quantified thing, the quantifier
2710 is automatically "possessified".
2711
2712 24. A recursive reference to a subpattern whose number was greater than 39
2713 went wrong under certain circumstances in UTF-8 mode. This bug could also
2714 have affected the operation of pcre_study().
2715
2716 25. Realized that a little bit of performance could be had by replacing
2717 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2718
2719 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2720
2721 27. Possessive quantifiers such as a++ were previously implemented by turning
2722 them into atomic groups such as ($>a+). Now they have their own opcodes,
2723 which improves performance. This includes the automatically created ones
2724 from 23 above.
2725
2726 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2727 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2728 the first matched character to be a colon. This applied both to named and
2729 numbered groups.
2730
2731 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2732
2733 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2734 building PCRE on Windows easier, so I have included it in the distribution.
2735
2736 31. There is now a check in pcretest against a ridiculously large number being
2737 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2738 loop, the loop is abandoned.
2739
2740 32. Forward references to subpatterns in conditions such as (?(2)...) where
2741 subpattern 2 is defined later cause pcre_compile() to search forwards in
2742 the pattern for the relevant set of parentheses. This search went wrong
2743 when there were unescaped parentheses in a character class, parentheses
2744 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2745
2746 33. "Subroutine" calls and backreferences were previously restricted to
2747 referencing subpatterns earlier in the regex. This restriction has now
2748 been removed.
2749
2750 34. Added a number of extra features that are going to be in Perl 5.10. On the
2751 whole, these are just syntactic alternatives for features that PCRE had
2752 previously implemented using the Python syntax or my own invention. The
2753 other formats are all retained for compatibility.
2754
2755 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2756 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2757 also .NET compatible.
2758
2759 (b) A recursion or subroutine call to a named group can now be defined as
2760 (?&name) as well as (?P>name).
2761
2762 (c) A backreference to a named group can now be defined as \k<name> or
2763 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2764 5.10, are also .NET compatible.
2765
2766 (d) A conditional reference to a named group can now use the syntax
2767 (?(<name>) or (?('name') as well as (?(name).
2768
2769 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2770 groups (named and numbered) that are never evaluated inline, but can be
2771 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2772 is always false. There may be only one alternative in such a group.
2773
2774 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2775 as the simple (?(R). The condition is true only if the most recent
2776 recursion is that of the given number or name. It does not search out
2777 through the entire recursion stack.
2778
2779 (g) The escape \gN or \g{N} has been added, where N is a positive or
2780 negative number, specifying an absolute or relative reference.
2781
2782 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2783 some "unreachable code" warnings.
2784
2785 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2786 things, this adds five new scripts.
2787
2788 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2789 There were also incompatibilities regarding the handling of \Q..\E inside
2790 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2791 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2792
2793 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2794 matches an empty string, and forcibly breaks the loop. There were bugs in
2795 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2796 against aaaa the result was just "a" rather than "aaaa", for example. Two
2797 separate and independent bugs (that affected different cases) have been
2798 fixed.
2799
2800 39. Refactored the code to abolish the use of different opcodes for small
2801 capturing bracket numbers. This is a tidy that I avoided doing when I
2802 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2803 The new approach is not only tidier, it makes it possible to reduce the
2804 memory needed to fix the previous bug (38).
2805
2806 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2807 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2808 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2809 mode.
2810
2811 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2812 report.
2813
2814 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2815 copy construction and assignment in the C++ wrapper.
2816
2817 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2818 couple of bugs that could have given wrong results in the "--newline=crlf"
2819 case.
2820
2821 44. Added a number of casts and did some reorganization of signed/unsigned int
2822 variables following suggestions from Dair Grant. Also renamed the variable
2823 "this" as "item" because it is a C++ keyword.
2824
2825 45. Arranged for dftables to add
2826
2827 #include "pcre_internal.h"
2828
2829 to pcre_chartables.c because without it, gcc 4.x may remove the array
2830 definition from the final binary if PCRE is built into a static library and
2831 dead code stripping is activated.
2832
2833 46. For an unanchored pattern, if a match attempt fails at the start of a
2834 newline sequence, and the newline setting is CRLF or ANY, and the next two
2835 characters are CRLF, advance by two characters instead of one.
2836
2837
2838 Version 6.7 04-Jul-06
2839 ---------------------
2840
2841 1. In order to handle tests when input lines are enormously long, pcretest has
2842 been re-factored so that it automatically extends its buffers when
2843 necessary. The code is crude, but this _is_ just a test program. The
2844 default size has been increased from 32K to 50K.
2845
2846 2. The code in pcre_study() was using the value of the re argument before
2847 testing it for NULL. (Of course, in any sensible call of the function, it
2848 won't be NULL.)
2849
2850 3. The memmove() emulation function in pcre_internal.h, which is used on
2851 systems that lack both memmove() and bcopy() - that is, hardly ever -
2852 was missing a "static" storage class specifier.
2853
2854 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2855 containing an extended class (one that cannot be represented by a bitmap
2856 because it contains high-valued characters or Unicode property items, e.g.
2857 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2858 pattern, but PCRE should not loop if you do not (it no longer does).
2859 [Detail: two cases were found: (a) a repeated subpattern containing an
2860 extended class; (b) a recursive reference to a subpattern that followed a
2861 previous extended class. It wasn't skipping over the extended class
2862 correctly when UTF-8 mode was not set.]
2863
2864 5. A negated single-character class was not being recognized as fixed-length
2865 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2866 compile error "lookbehind assertion is not fixed length".
2867
2868 6. The RunPerlTest auxiliary script was showing an unexpected difference
2869 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2870 write a Perl script that can interpret lines of an input file either as
2871 byte characters or as UTF-8, which is what "perltest" was being required to
2872 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2873 can't do is switch easily at run time between having the "use utf8;" pragma
2874 or not. In the end, I fudged it by using the RunPerlTest script to insert
2875 "use utf8;" explicitly for the UTF-8 tests.
2876
2877 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2878 the end of the subject string, contrary to the documentation and to what
2879 Perl does. This was true of both matching functions. Now it matches only at
2880 the start of the subject and immediately after *internal* newlines.
2881
2882 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2883 a pointer to an int instead of a pointer to an unsigned long int. This
2884 caused problems on 64-bit systems.
2885
2886 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2887 instance of the 'standard' template library not being so standard".
2888
2889 10. There was no check on the number of named subpatterns nor the maximum
2890 length of a subpattern name. The product of these values is used to compute
2891 the size of the memory block for a compiled pattern. By supplying a very
2892 long subpattern name and a large number of named subpatterns, the size
2893 computation could be caused to overflow. This is now prevented by limiting
2894 the length of names to 32 characters, and the number of named subpatterns
2895 to 10,000.
2896
2897 11. Subpatterns that are repeated with specific counts have to be replicated in
2898 the compiled pattern. The size of memory for this was computed from the
2899 length of the subpattern and the repeat count. The latter is limited to
2900 65535, but there was no limit on the former, meaning that integer overflow
2901 could in principle occur. The compiled length of a repeated subpattern is
2902 now limited to 30,000 bytes in order to prevent this.
2903
2904 12. Added the optional facility to have named substrings with the same name.
2905
2906 13. Added the ability to use a named substring as a condition, using the
2907 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2908 are numbers (not recommended). Forward references are permitted.
2909
2910 14. Added forward references in named backreferences (if you see what I mean).
2911
2912 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2913 pattern could run off the end of the subject. For example, the pattern
2914 "(?s)(.{1,5})"8 did this with the subject "ab".
2915
2916 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2917 PCRE_CASELESS was set when matching characters that were quantified with ?
2918 or *.
2919
2920 17. A character class other than a single negated character that had a minimum
2921 but no maximum quantifier - for example [ab]{6,} - was not handled
2922 correctly by pce_dfa_exec(). It would match only one character.
2923
2924 18. A valid (though odd) pattern that looked like a POSIX character
2925 class but used an invalid character after [ (for example [[,abc,]]) caused
2926 pcre_compile() to give the error "Failed: internal error: code overflow" or
2927 in some cases to crash with a glibc free() error. This could even happen if
2928 the pattern terminated after [[ but there just happened to be a sequence of
2929 letters, a binary zero, and a closing ] in the memory that followed.
2930
2931 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2932 over the years. Originally (before any Unicode support), just the bottom 8
2933 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2934 output from "man perlunicode" includes this:
2935
2936 The regular expression compiler produces polymorphic opcodes. That
2937 is, the pattern adapts to the data and automatically switches to
2938 the Unicode character scheme when presented with Unicode data--or
2939 instead uses a traditional byte scheme when presented with byte
2940 data.
2941
2942 Sadly, a wide octal escape does not cause a switch, and in a string with
2943 no other multibyte characters, these octal escapes are treated as before.
2944 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2945 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2946 Unicode string.
2947
2948 I have not perpetrated such confusion in PCRE. Up till now, it took just
2949 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2950 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2951 translate to the appropriate multibyte character.
2952
2953 29. Applied some refactoring to reduce the number of warnings from Microsoft
2954 and Borland compilers. This has included removing the fudge introduced
2955 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2956 a warning about an unused variable.
2957
2958 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2959 characters since release 4.0, because Perl (from release 5.004) does not.
2960 [Or at least, is documented not to: some releases seem to be in conflict
2961 with the documentation.] However, when a pattern was studied with
2962 pcre_study() and all its branches started with \s, PCRE still included VT
2963 as a possible starting character. Of course, this did no harm; it just
2964 caused an unnecessary match attempt.
2965
2966 22. Removed a now-redundant internal flag bit that recorded the fact that case
2967 dependency changed within the pattern. This was once needed for "required
2968 byte" processing, but is no longer used. This recovers a now-scarce options
2969 bit. Also moved the least significant internal flag bit to the most-
2970 significant bit of the word, which was not previously used (hangover from
2971 the days when it was an int rather than a uint) to free up another bit for
2972 the future.
2973
2974 23. Added support for CRLF line endings as well as CR and LF. As well as the
2975 default being selectable at build time, it can now be changed at runtime
2976 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2977 specify that it is scanning data with non-default line endings.
2978
2979 24. Changed the definition of CXXLINK to make it agree with the definition of
2980 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2981
2982 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2983 recursions. This makes a big different to stack usage for some patterns.
2984
2985 26. If a subpattern containing a named recursion or subroutine reference such
2986 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2987 the space required for the compiled pattern went wrong and gave too small a
2988 value. Depending on the environment, this could lead to "Failed: internal
2989 error: code overflow at offset 49" or "glibc detected double free or
2990 corruption" errors.
2991
2992 27. Applied patches from Google (a) to support the new newline modes and (b) to
2993 advance over multibyte UTF-8 characters in GlobalReplace.
2994
2995 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2996 difference for some implementation of PCRE in some Windows version.
2997
2998 29. Added some extra testing facilities to pcretest:
2999
3000 \q<number> in a data line sets the "match limit" value
3001 \Q<number> in a data line sets the "match recursion limt" value
3002 -S <number> sets the stack size, where <number> is in megabytes
3003
3004 The -S option isn't available for Windows.
3005
3006
3007 Version 6.6 06-Feb-06
3008 ---------------------
3009
3010 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3011 in pcreposix.h. I have copied the definition from pcre.h.
3012
3013 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3014 because pcre.h is no longer a built file.
3015
3016 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3017 not normally included in the compiled code.
3018
3019
3020 Version 6.5 01-Feb-06
3021 ---------------------
3022
3023 1. When using the partial match feature with pcre_dfa_exec(), it was not
3024 anchoring the second and subsequent partial matches at the new starting
3025 point. This could lead to incorrect results. For example, with the pattern
3026 /1234/, partially matching against "123" and then "a4" gave a match.
3027
3028 2. Changes to pcregrep:
3029
3030 (a) All non-match returns from pcre_exec() were being treated as failures
3031 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3032 error message is output. Some extra information is given for the
3033 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3034 probably the only errors that are likely to be caused by users (by
3035 specifying a regex that has nested indefinite repeats, for instance).
3036 If there are more than 20 of these errors, pcregrep is abandoned.
3037
3038 (b) A binary zero was treated as data while matching, but terminated the
3039 output line if it was written out. This has been fixed: binary zeroes
3040 are now no different to any other data bytes.
3041
3042 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3043 used to set a locale for matching. The --locale=xxxx long option has
3044 been added (no short equivalent) to specify a locale explicitly on the
3045 pcregrep command, overriding the environment variables.
3046
3047 (d) When -B was used with -n, some line numbers in the output were one less
3048 than they should have been.
3049
3050 (e) Added the -o (--only-matching) option.
3051
3052 (f) If -A or -C was used with -c (count only), some lines of context were
3053 accidentally printed for the final match.
3054
3055 (g) Added the -H (--with-filename) option.
3056
3057 (h) The combination of options -rh failed to suppress file names for files
3058 that were found from directory arguments.
3059
3060 (i) Added the -D (--devices) and -d (--directories) options.
3061
3062 (j) Added the -F (--fixed-strings) option.
3063
3064 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3065
3066 (l) Added the --colo(u)r option.
3067
3068 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3069 is not present by default.
3070
3071 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3072 items such as (?R) or (?1), when the recursion could match a number of
3073 alternatives. If it matched one of the alternatives, but subsequently,
3074 outside the recursion, there was a failure, the code tried to back up into
3075 the recursion. However, because of the way PCRE is implemented, this is not
3076 possible, and the result was an incorrect result from the match.
3077
3078 In order to prevent this happening, the specification of recursion has
3079 been changed so that all such subpatterns are automatically treated as
3080 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3081
3082 4. I had overlooked the fact that, in some locales, there are characters for
3083 which isalpha() is true but neither isupper() nor islower() are true. In
3084 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3085 and ordfeminine) are like this. This affected the treatment of \w and \W
3086 when they appeared in character classes, but not when they appeared outside
3087 a character class. The bit map for "word" characters is now created
3088 separately from the results of isalnum() instead of just taking it from the
3089 upper, lower, and digit maps. (Plus the underscore character, of course.)
3090
3091 5. The above bug also affected the handling of POSIX character classes such as
3092 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3093 permanent tables. Instead, the bit maps for such a class were previously
3094 created as the appropriate unions of the upper, lower, and digit bitmaps.
3095 Now they are created by subtraction from the [[:word:]] class, which has
3096 its own bitmap.
3097
3098 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3099 It is created by subtracting the vertical space characters (\x09, \x0a,
3100 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3101 subtraction was done in the overall bitmap for a character class, meaning
3102 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3103 be recognized. This bug has been fixed.
3104
3105 7. Patches from the folks at Google:
3106
3107 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3108 real life, but is still worth protecting against".
3109
3110 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3111 regular expressions".
3112
3113 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3114 have it.
3115
3116 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3117 "configure" and the latter not, in order to fix a problem somebody had
3118 with compiling the Arg class on HP-UX.
3119
3120 (e) Improve the error-handling of the C++ wrapper a little bit.
3121
3122 (f) New tests for checking recursion limiting.
3123
3124 8. The pcre_memmove() function, which is used only if the environment does not
3125 have a standard memmove() function (and is therefore rarely compiled),
3126 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3127 returning a result (though PCRE never actually uses the result).
3128
3129 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3130 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3131 returned instead of calling malloc() with an overflowing number that would
3132 most likely cause subsequent chaos.
3133
3134 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3135
3136 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3137 with this option is matched, the nmatch and pmatch options of regexec() are
3138 ignored.
3139
3140 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3141 provided in case anyone wants to the the POSIX interface with UTF-8
3142 strings.
3143
3144 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3145 C++ linking (needed for some HP-UX environments).
3146
3147 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3148 (unused parameter) and in the pcre_printint() function (omitted "default"
3149 switch label when the default is to do nothing).
3150
3151 15. Added some code to make it possible, when PCRE is compiled as a C++
3152 library, to replace subject pointers for pcre_exec() with a smart pointer
3153 class, thus making it possible to process discontinuous strings.
3154
3155 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3156 much the same function. They were added by different people who were trying
3157 to make PCRE easy to compile on non-Unix systems. It has been suggested
3158 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3159 for compiling on Windows systems. I have therefore replaced it with
3160 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3161 defaults to "extern" for C or "extern C" for C++, which works fine on
3162 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3163 SCOPE with something explicit in config.h. In addition:
3164
3165 (a) pcreposix.h still had just "extern" instead of either of these macros;
3166 I have replaced it with PCRE_DATA_SCOPE.
3167
3168 (b) Functions such as _pcre_xclass(), which are internal to the library,
3169 but external in the C sense, all had PCRE_EXPORT in their definitions.
3170 This is apparently wrong for the Windows case, so I have removed it.
3171 (It makes no difference on Unix-like systems.)
3172
3173 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3174 of recursive calls to match(). This is different to MATCH_LIMIT because
3175 that limits the total number of calls to match(), not all of which increase
3176 the depth of recursion. Limiting the recursion depth limits the amount of
3177 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3178 when PCRE is compiled, and changed at run time. A patch from Google adds
3179 this functionality to the C++ interface.
3180
3181 18. Changes to the handling of Unicode character properties:
3182
3183 (a) Updated the table to Unicode 4.1.0.
3184
3185 (b) Recognize characters that are not in the table as "Cn" (undefined).
3186
3187 (c) I revised the way the table is implemented to a much improved format
3188 which includes recognition of ranges. It now supports the ranges that
3189 are defined in UnicodeData.txt, and it also amalgamates other
3190 characters into ranges. This has reduced the number of entries in the
3191 table from around 16,000 to around 3,000, thus reducing its size
3192 considerably. I realized I did not need to use a tree structure after
3193 all - a binary chop search is just as efficient. Having reduced the
3194 number of entries, I extended their size from 6 bytes to 8 bytes to
3195 allow for more data.
3196
3197 (d) Added support for Unicode script names via properties such as \p{Han}.
3198
3199 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3200 matching that character.
3201
3202 20. When matching a repeated Unicode property with a minimum greater than zero,
3203 (for example \pL{2,}), PCRE could look past the end of the subject if it
3204 reached it while seeking the minimum number of characters. This could
3205 happen only if some of the characters were more than one byte long, because
3206 there is a check for at least the minimum number of bytes.
3207
3208 21. Refactored the implementation of \p and \P so as to be more general, to
3209 allow for more different types of property in future. This has changed the
3210 compiled form incompatibly. Anybody with saved compiled patterns that use
3211 \p or \P will have to recompile them.
3212
3213 22. Added "Any" and "L&" to the supported property types.
3214
3215 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3216 but give a compile time error if the value is greater than 0xff.
3217
3218 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3219 accidentally not being installed or uninstalled.
3220
3221 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3222 made were to insert the current release number. This seemed silly, because
3223 it made things harder for people building PCRE on systems that don't run
3224 "configure". I have turned pcre.h into a distributed file, no longer built
3225 by "configure", with the version identification directly included. There is
3226 no longer a pcre.h.in file.
3227
3228 However, this change necessitated a change to the pcre-config script as
3229 well. It is built from pcre-config.in, and one of the substitutions was the
3230 release number. I have updated configure.ac so that ./configure now finds
3231 the release number by grepping pcre.h.
3232
3233 26. Added the ability to run the tests under valgrind.
3234
3235
3236 Version 6.4 05-Sep-05
3237 ---------------------
3238
3239 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3240 "--" to be printed when multiple files were scanned, even when none of the
3241 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3242 consider it to be a bug, and have restored the previous behaviour.
3243
3244 2. A couple of code tidies to get rid of compiler warnings.
3245
3246 3. The pcretest program used to cheat by referring to symbols in the library
3247 whose names begin with _pcre_. These are internal symbols that are not
3248 really supposed to be visible externally, and in some environments it is
3249 possible to suppress them. The cheating is now confined to including
3250 certain files from the library's source, which is a bit cleaner.
3251
3252 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3253 file's purpose clearer.
3254
3255 5. Reorganized pcre_ucp_findchar().
3256
3257
3258 Version 6.3 15-Aug-05
3259 ---------------------
3260
3261 1. The file libpcre.pc.in did not have general read permission in the tarball.
3262
3263 2. There were some problems when building without C++ support:
3264
3265 (a) If C++ support was not built, "make install" and "make test" still
3266 tried to test it.
3267
3268 (b) There were problems when the value of CXX was explicitly set. Some
3269 changes have been made to try to fix these, and ...
3270
3271 (c) --disable-cpp can now be used to explicitly disable C++ support.
3272
3273 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3274 backslash in a target when C++ was disabled. This confuses some
3275 versions of "make", apparently. Using an intermediate variable solves
3276 this. (Same for CPP_LOBJ.)
3277
3278 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3279 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3280 necessary on certain architectures.
3281
3282 4. Added a setting of -export-symbols-regex to the link command to remove
3283 those symbols that are exported in the C sense, but actually are local
3284 within the library, and not documented. Their names all begin with
3285 "_pcre_". This is not a perfect job, because (a) we have to except some
3286 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3287 available (and never for static libraries). I have made a note to try to
3288 find a way round (a) in the future.
3289
3290
3291 Version 6.2 01-Aug-05
3292 ---------------------
3293
3294 1. There was no test for integer overflow of quantifier values. A construction
3295 such as {1111111111111111} would give undefined results. What is worse, if
3296 a minimum quantifier for a parenthesized subpattern overflowed and became
3297 negative, the calculation of the memory size went wrong. This could have
3298 led to memory overwriting.
3299
3300 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3301
3302 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3303 operating environments where this matters.
3304
3305 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3306 PCRE options from within the C++ wrapper.
3307
3308 5. Named capturing subpatterns were not being correctly counted when a pattern
3309 was compiled. This caused two problems: (a) If there were more than 100
3310 such subpatterns, the calculation of the memory needed for the whole
3311 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3312 back references of the form \12, where the number was greater than 9, were
3313 not recognized as back references, even though there were sufficient
3314 previous subpatterns.
3315
3316 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3317 versions of gcc, e.g. 2.95.4.
3318
3319
3320 Version 6.1 21-Jun-05
3321 ---------------------
3322
3323 1. There was one reference to the variable "posix" in pcretest.c that was not
3324 surrounded by "#if !defined NOPOSIX".
3325
3326 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3327 the cross-check on the old pcre_info() function, for the benefit of the
3328 cut-down version of PCRE that is currently imported into Exim.
3329
3330 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3331 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3332 patterns that start (?i) but I don't think that matters. The use of (?i) is
3333 just an example; this all applies to the other options as well.
3334
3335 4. Since libtool seems to echo the compile commands it is issuing, the output
3336 from "make" can be reduced a bit by putting "@" in front of each libtool
3337 compile command.
3338
3339 5. Patch from the folks at Google for configure.in to be a bit more thorough
3340 in checking for a suitable C++ installation before trying to compile the
3341 C++ stuff. This should fix a reported problem when a compiler was present,
3342 but no suitable headers.
3343
3344 6. The man pages all had just "PCRE" as their title. I have changed them to
3345 be the relevant file name. I have also arranged that these names are
3346 retained in the file doc/pcre.txt, which is a concatenation in text format
3347 of all the man pages except the little individual ones for each function.
3348
3349 7. The NON-UNIX-USE file had not been updated for the different set of source
3350 files that come with release 6. I also added a few comments about the C++
3351 wrapper.
3352
3353
3354 Version 6.0 07-Jun-05
3355 ---------------------
3356
3357 1. Some minor internal re-organization to help with my DFA experiments.
3358
3359 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3360 didn't matter for the library itself when fully configured, but did matter
3361 when compiling without UCP support, or within Exim, where the ucp files are
3362 not imported.
3363
3364 3. Refactoring of the library code to split up the various functions into
3365 different source modules. The addition of the new DFA matching code (see
3366 below) to a single monolithic source would have made it really too
3367 unwieldy, quite apart from causing all the code to be include in a
3368 statically linked application, when only some functions are used. This is
3369 relevant even without the DFA addition now that patterns can be compiled in
3370 one application and matched in another.
3371
3372 The downside of splitting up is that there have to be some external
3373 functions and data tables that are used internally in different modules of
3374 the library but which are not part of the API. These have all had their
3375 names changed to start with "_pcre_" so that they are unlikely to clash
3376 with other external names.
3377
3378 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3379 a different (DFA) algorithm. Although it is slower than the original
3380 function, it does have some advantages for certain types of matching
3381 problem.
3382
3383 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3384 including restarting after a partial match.
3385
3386 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3387 defined when compiling for Windows was sent to me. I have put it into the
3388 code, though I have no means of testing or verifying it.
3389
3390 7. Added the pcre_refcount() auxiliary function.
3391
3392 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3393 match before or at the first newline in the subject string. In pcretest,
3394 the /f option on a pattern can be used to set this.
3395
3396 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3397 would behave wrongly. This has been present in PCRE since release 4.0.
3398
3399 10. A number of changes to the pcregrep command:
3400
3401 (a) Refactored how -x works; insert ^(...)$ instead of setting
3402 PCRE_ANCHORED and checking the length, in preparation for adding
3403 something similar for -w.
3404
3405 (b) Added the -w (match as a word) option.
3406
3407 (c) Refactored the way lines are read and buffered so as to have more
3408 than one at a time available.
3409
3410 (d) Implemented a pcregrep test script.
3411
3412 (e) Added the -M (multiline match) option. This allows patterns to match
3413 over several lines of the subject. The buffering ensures that at least
3414 8K, or the rest of the document (whichever is the shorter) is available
3415 for matching (and similarly the previous 8K for lookbehind assertions).
3416
3417 (f) Changed the --help output so that it now says
3418
3419 -w, --word-regex(p)
3420
3421 instead of two lines, one with "regex" and the other with "regexp"
3422 because that confused at least one person since the short forms are the
3423 same. (This required a bit of code, as the output is generated
3424 automatically from a table. It wasn't just a text change.)
3425
3426 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3427 option but starts with a hyphen. Could be a pattern or a path name
3428 starting with a hyphen, for instance.
3429
3430 (h) "-" can be given as a file name to represent stdin.
3431
3432 (i) When file names are being printed, "(standard input)" is used for
3433 the standard input, for compatibility with GNU grep. Previously
3434 "<stdin>" was used.
3435
3436 (j) The option --label=xxx can be used to supply a name to be used for
3437 stdin when file names are being printed. There is no short form.
3438
3439 (k) Re-factored the options decoding logic because we are going to add
3440 two more options that take data. Such options can now be given in four
3441 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3442
3443 (l) Added the -A, -B, and -C options for requesting that lines of context
3444 around matches be printed.
3445
3446 (m) Added the -L option to print the names of files that do not contain
3447 any matching lines, that is, the complement of -l.
3448
3449 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3450 continue to scan other files.
3451
3452 (o) The -s option was incorrectly implemented. For compatibility with other
3453 greps, it now suppresses the error message for a non-existent or non-
3454 accessible file (but not the return code). There is a new option called
3455 -q that suppresses the output of matching lines, which was what -s was
3456 previously doing.
3457
3458 (p) Added --include and --exclude options to specify files for inclusion
3459 and exclusion when recursing.
3460
3461 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3462 Hopefully, it now does.
3463
3464 12. Missing cast in pcre_study().
3465
3466 13. Added an "uninstall" target to the makefile.
3467
3468 14. Replaced "extern" in the function prototypes in Makefile.in with
3469 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3470 world, but is set differently for Windows.
3471
3472 15. Added a second compiling function called pcre_compile2(). The only
3473 difference is that it has an extra argument, which is a pointer to an
3474 integer error code. When there is a compile-time failure, this is set
3475 non-zero, in addition to the error test pointer being set to point to an
3476 error message. The new argument may be NULL if no error number is required
3477 (but then you may as well call pcre_compile(), which is now just a
3478 wrapper). This facility is provided because some applications need a
3479 numeric error indication, but it has also enabled me to tidy up the way
3480 compile-time errors are handled in the POSIX wrapper.
3481
3482 16. Added VPATH=.libs to the makefile; this should help when building with one
3483 prefix path and installing with another. (Or so I'm told by someone who
3484 knows more about this stuff than I do.)
3485
3486 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3487 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3488 match everything, including newlines. This is not POSIX-compatible, but
3489 somebody wanted the feature. From pcretest it can be activated by using
3490 both the P and the s flags.
3491
3492 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3493
3494 19. libpcre.pc was being incorrectly installed as executable.
3495
3496 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3497 it now also looks for '\r' so that it will work unmodified on Windows.
3498
3499 21. Added Google's contributed C++ wrapper to the distribution.
3500
3501 22. Added some untidy missing memory free() calls in pcretest, to keep
3502 Electric Fence happy when testing.
3503
3504
3505
3506 Version 5.0 13-Sep-04
3507 ---------------------
3508
3509 1. Internal change: literal characters are no longer packed up into items
3510 containing multiple characters in a single byte-string. Each character
3511 is now matched using a separate opcode. However, there may be more than one
3512 byte in the character in UTF-8 mode.
3513
3514 2. The pcre_callout_block structure has two new fields: pattern_position and
3515 next_item_length. These contain the offset in the pattern to the next match
3516 item, and its length, respectively.
3517
3518 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3519 insertion of callouts before each pattern item. Added the /C option to
3520 pcretest to make use of this.
3521
3522 4. On the advice of a Windows user, the lines
3523
3524 #if defined(_WIN32) || defined(WIN32)
3525 _setmode( _fileno( stdout ), 0x8000 );
3526 #endif /* defined(_WIN32) || defined(WIN32) */
3527
3528 have been added to the source of pcretest. This apparently does useful
3529 magic in relation to line terminators.
3530
3531 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3532 for the benefit of those environments where the "b" makes a difference.
3533
3534 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3535 to know about it. I have put a hack into configure.in that adds in code
3536 to set GCC=yes if CC=icc. This seems to end up at a point in the
3537 generated configure script that is early enough to affect the setting of
3538 compiler options, which is what is needed, but I have no means of testing
3539 whether it really works. (The user who reported this had patched the
3540 generated configure script, which of course I cannot do.)
3541
3542 LATER: After change 22 below (new libtool files), the configure script
3543 seems to know about icc (and also ecc). Therefore, I have commented out
3544 this hack in configure.in.
3545
3546 7. Added support for pkg-config (2 patches were sent in).
3547
3548 8. Negated POSIX character classes that used a combination of internal tables
3549 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3550 [[:^ascii]]. Typically, they would match almost any characters. The other
3551 POSIX classes were not broken in this way.
3552
3553 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3554 to find the match, as PCRE was deluded into thinking that the match had to
3555 start at the start point or following a newline. The same bug applied to
3556 patterns with negative forward assertions or any backward assertions
3557 preceding ".*" at the start, unless the pattern required a fixed first
3558 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3559
3560 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3561 starting at the last subject character, bytes beyond the end of the subject
3562 string were read.
3563
3564 11. Renamed the variable "class" as "classbits" to make life easier for C++
3565 users. (Previously there was a macro definition, but it apparently wasn't
3566 enough.)
3567
3568 12. Added the new field "tables" to the extra data so that tables can be passed
3569 in at exec time, or the internal tables can be re-selected. This allows
3570 a compiled regex to be saved and re-used at a later time by a different
3571 program that might have everything at different addresses.
3572
3573 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3574 -R library as well as a -L library.
3575
3576 14. The debugging options of pcretest (-d on the command line or D on a
3577 pattern) showed incorrect output for anything following an extended class
3578 that contained multibyte characters and which was followed by a quantifier.
3579
3580 15. Added optional support for general category Unicode character properties
3581 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3582 support. It adds about 90K to the size of the library. The meanings of the
3583 inbuilt class escapes such as \d and \s have NOT been changed.
3584
3585 16. Updated pcredemo.c to include calls to free() to release the memory for the
3586 compiled pattern.
3587
3588 17. The generated file chartables.c was being created in the source directory
3589 instead of in the building directory. This caused the build to fail if the
3590 source directory was different from the building directory, and was
3591 read-only.
3592
3593 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3594 file. No doubt somebody will tell me if they don't make sense... Also added
3595 Dan Mooney's comments about building on OpenVMS.
3596
3597 19. Added support for partial matching via the PCRE_PARTIAL option for
3598 pcre_exec() and the \P data escape in pcretest.
3599
3600 20. Extended pcretest with 3 new pattern features:
3601
3602 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3603 write the compiled pattern to the file whose name is "rest-of-line".
3604 This is a straight binary dump of the data, with the saved pointer to
3605 the character tables forced to be NULL. The study data, if any, is
3606 written too. After writing, pcretest reads a new pattern.
3607
3608 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3609 compiled pattern from the given file. There must not be any
3610 occurrences of "<" in the file name (pretty unlikely); if there are,
3611 pcretest will instead treat the initial "<" as a pattern delimiter.
3612 After reading in the pattern, pcretest goes on to read data lines as
3613 usual.
3614
3615 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3616 and 16-bit fields in a compiled pattern, to simulate a pattern that
3617 was compiled on a host of opposite endianness.
3618
3619 21. The pcre-exec() function can now cope with patterns that were compiled on
3620 hosts of opposite endianness, with this restriction:
3621
3622 As for any compiled expression that is saved and used later, the tables
3623 pointer field cannot be preserved; the extra_data field in the arguments
3624 to pcre_exec() should be used to pass in a tables address if a value
3625 other than the default internal tables were used at compile time.
3626
3627 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3628 now diagnosed as an error. Previously, most of the time, a negative number
3629 would have been treated as zero, but if in addition "ovector" was passed as
3630 NULL, a crash could occur.
3631
3632 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3633 new versions from the libtool 1.5 distribution (the last one is a copy of
3634 a file called libtool.m4). This seems to have fixed the need to patch
3635 "configure" to support Darwin 1.3 (which I used to do). However, I still
3636 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3637 workstation).
3638
3639 24. Changed the PCRE licence to be the more standard "BSD" licence.
3640
3641
3642 Version 4.5 01-Dec-03
3643 ---------------------
3644
3645 1. There has been some re-arrangement of the code for the match() function so
3646 that it can be compiled in a version that does not call itself recursively.
3647 Instead, it keeps those local variables that need separate instances for
3648 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3649 needs to "recurse". Keeping track of where control must go is done by means
3650 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3651 hide most of the details from the main code, and operates only if
3652 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3653 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3654 operating.
3655
3656 To make it easier for callers to provide specially tailored get/free
3657 functions for this usage, two new functions, pcre_stack_malloc, and
3658 pcre_stack_free, are used. They are always called in strict stacking order,
3659 and the size of block requested is always the same.
3660
3661 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3662 PCRE has been compiled to use the stack or the heap for recursion. The
3663 -C option of pcretest uses this to show which version is compiled.
3664
3665 A new data escape \S, is added to pcretest; it causes the amounts of store
3666 obtained and freed by both kinds of malloc/free at match time to be added
3667 to the output.
3668
3669 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3670 what's available on my current Linux desktop machine.
3671
3672 3. When matching a UTF-8 string, the test for a valid string at the start has
3673 been extended. If start_offset is not zero, PCRE now checks that it points
3674 to a byte that is the start of a UTF-8 character. If not, it returns
3675 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3676 this is necessary because there may be backward assertions in the pattern.
3677 When matching the same subject several times, it may save resources to use
3678 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3679
3680 4. The code for checking the validity of UTF-8 strings has been tightened so
3681 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3682 containing "overlong sequences".
3683
3684 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3685 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3686 should have been "&", but it just so happened that all the cases this let
3687 through by mistake were picked up later in the function.
3688
3689 6. I had used a variable called "isblank" - this is a C99 function, causing
3690 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3691
3692 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3693 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3694 so the version is shown; (c) stop "make test" echoing "./RunTest".
3695
3696 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3697
3698 9. The prototype for memmove() for systems that don't have it was using
3699 size_t, but the inclusion of the header that defines size_t was later. I've
3700 moved the #includes for the C headers earlier to avoid this.
3701
3702 10. Added some adjustments to the code to make it easier to compiler on certain
3703 special systems:
3704
3705 (a) Some "const" qualifiers were missing.
3706 (b) Added the macro EXPORT before all exported functions; by default this
3707 is defined to be empty.
3708 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3709 that it reads its output file name as an argument instead of writing
3710 to the standard output and assuming this can be redirected.
3711
3712 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3713 class containing characters with values greater than 255, PCRE compilation
3714 went into a loop.
3715
3716 12. A recursive reference to a subpattern that was within another subpattern
3717 that had a minimum quantifier of zero caused PCRE to crash. For example,
3718 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3719 recursion. If the recursively-called subpattern itself had a zero repeat,
3720 that was OK.
3721
3722 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3723 buffer into which it was copied (for escape processing) was still set at
3724 1024, so long lines caused crashes.
3725
3726 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3727 "internal error: code overflow...". This applied to any character class
3728 that was followed by a possessive quantifier.
3729
3730 15. Modified the Makefile to add libpcre.la as a prerequisite for
3731 libpcreposix.la because I was told this is needed for a parallel build to
3732 work.
3733
3734 16. If a pattern that contained .* following optional items at the start was
3735 studied, the wrong optimizing data was generated, leading to matching
3736 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3737 matching string must start with a or b or c. The correct conclusion for
3738 this pattern is that a match can start with any character.
3739
3740
3741 Version 4.4 13-Aug-03
3742 ---------------------
3743
3744 1. In UTF-8 mode, a character class containing characters with values between
3745 127 and 255 was not handled correctly if the compiled pattern was studied.
3746 In fixing this, I have also improved the studying algorithm for such
3747 classes (slightly).
3748
3749 2. Three internal functions had redundant arguments passed to them. Removal
3750 might give a very teeny performance improvement.
3751
3752 3. Documentation bug: the value of the capture_top field in a callout is *one
3753 more than* the number of the hightest numbered captured substring.
3754
3755 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3756 in incorrectly linking with a previously installed version. They now link
3757 explicitly with libpcre.la.
3758
3759 5. configure.in no longer needs to recognize Cygwin specially.
3760
3761 6. A problem in pcre.in for Windows platforms is fixed.
3762
3763 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3764 pcretest, it used to include the size of the study block as part of its
3765 output. Unfortunately, the structure contains a field that has a different
3766 size on different hardware architectures. This meant that the tests that
3767 showed this size failed. As the block is currently always of a fixed size,
3768 this information isn't actually particularly useful in pcretest output, so
3769 I have just removed it.
3770
3771 8. Three pre-processor statements accidentally did not start in column 1.
3772 Sadly, there are *still* compilers around that complain, even though
3773 standard C has not required this for well over a decade. Sigh.
3774
3775 9. In pcretest, the code for checking callouts passed small integers in the
3776 callout_data field, which is a void * field. However, some picky compilers
3777 complained about the casts involved for this on 64-bit systems. Now
3778 pcretest passes the address of the small integer instead, which should get
3779 rid of the warnings.
3780
3781 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3782 both compile and run time, and gives an error if an invalid UTF-8 sequence
3783 is found. There is a option for disabling this check in cases where the
3784 string is known to be correct and/or the maximum performance is wanted.
3785
3786 11. In response to a bug report, I changed one line in Makefile.in from
3787
3788 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3789 to
3790 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3791
3792 to look similar to other lines, but I have no way of telling whether this
3793 is the right thing to do, as I do not use Windows. No doubt I'll get told
3794 if it's wrong...
3795
3796
3797 Version 4.3 21-May-03
3798 ---------------------
3799
3800 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3801 Makefile.
3802
3803 2. Some refactoring to improve the quality of the code:
3804
3805 (i) The utf8_table... variables are now declared "const".
3806
3807 (ii) The code for \cx, which used the "case flipping" table to upper case
3808 lower case letters, now just substracts 32. This is ASCII-specific,
3809 but the whole concept of \cx is ASCII-specific, so it seems
3810 reasonable.
3811
3812 (iii) PCRE was using its character types table to recognize decimal and
3813 hexadecimal digits in the pattern. This is silly, because it handles
3814 only 0-9, a-f, and A-F, but the character types table is locale-
3815 specific, which means strange things might happen. A private
3816 table is now used for this - though it costs 256 bytes, a table is
3817 much faster than multiple explicit tests. Of course, the standard
3818 character types table is still used for matching digits in subject
3819 strings against \d.
3820
3821 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3822 ending in _t are). So I've renamed it as ESC_tee.
3823
3824 3. The first argument for regexec() in the POSIX wrapper should have been
3825 defined as "const".
3826
3827 4. Changed pcretest to use malloc() for its buffers so that they can be
3828 Electric Fenced for debugging.
3829
3830 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3831 to read one or more bytes before the start of the subject string. Often this
3832 had no effect on PCRE's behaviour, but in some circumstances it could
3833 provoke a segmentation fault.
3834
3835 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3836 to try to read one or more bytes before the start of the subject string.
3837
3838 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3839 UTF-8 support could misbehave in various ways if the subject string
3840 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3841 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3842 back over UTF-8 characters.)
3843
3844
3845 Version 4.2 14-Apr-03
3846 ---------------------
3847
3848 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3849
3850 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3851 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3852 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3853 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3854 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3855 and BUILD_EXEEXT
3856 Note: automatic setting of the BUILD variables is not yet working
3857 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3858 compile-time but not at link-time
3859 [LINK]: use for linking executables only
3860 make different versions for Windows and non-Windows
3861 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3862 libraries
3863 [LINK_FOR_BUILD]: new variable
3864 [OBJEXT]: use throughout
3865 [EXEEXT]: use throughout
3866 <winshared>: new target
3867 <wininstall>: new target
3868 <dftables.o>: use native compiler
3869 <dftables>: use native linker
3870 <install>: handle Windows platform correctly
3871 <clean>: ditto
3872 <check>: ditto
3873 copy DLL to top builddir before testing
3874
3875 As part of these changes, -no-undefined was removed again. This was reported
3876 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3877 in any case.
3878
3879 3. Some tidies to get rid of compiler warnings:
3880
3881 . In the match_data structure, match_limit was an unsigned long int, whereas
3882 match_call_count was an int. I've made them both unsigned long ints.
3883
3884 . In pcretest the fact that a const uschar * doesn't automatically cast to
3885 a void * provoked a warning.
3886
3887 . Turning on some more compiler warnings threw up some "shadow" variables
3888 and a few more missing casts.
3889
3890 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3891 option, a class that contained a single character with a value between 128
3892 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3893
3894 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3895 option, a class that contained several characters, but with at least one
3896 whose value was between 128 and 255 caused PCRE to crash.
3897
3898
3899 Version 4.1 12-Mar-03
3900 ---------------------
3901
3902 1. Compiling with gcc -pedantic found a couple of places where casts were
3903 needed, and a string in dftables.c that was longer than standard compilers are
3904 required to support.
3905
3906 2. Compiling with Sun's compiler found a few more places where the code could
3907 be tidied up in order to avoid warnings.
3908
3909 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3910 first of these names is deprecated in the latest Autoconf in favour of the name
3911 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3912 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3913 analogy I have changed it to CFLAGS_FOR_BUILD.
3914
3915 4. Added -no-undefined to the linking command in the Makefile, because this is
3916 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3917 linking step for the pcreposix library.
3918
3919 5. PCRE was failing to diagnose the case of two named groups with the same
3920 name.
3921
3922 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3923 literal character that is needed in the subject for a match, and scans along to
3924 ensure that it is present before embarking on the full matching process. This
3925 saves time in cases of nested unlimited repeats that are never going to match.
3926 Problem: the scan can take a lot of time if the subject is very long (e.g.
3927 megabytes), thus penalizing straightforward matches. It is now done only if the
3928 amount of subject to be scanned is less than 1000 bytes.
3929
3930 7. A lesser problem with the same optimization is that it was recording the
3931 first character of an anchored pattern as "needed", thus provoking a search
3932 right along the subject, even when the first match of the pattern was going to
3933 fail. The "needed" character is now not set for anchored patterns, unless it
3934 follows something in the pattern that is of non-fixed length. Thus, it still
3935 fulfils its original purpose of finding quick non-matches in cases of nested
3936 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3937
3938
3939 Version 4.0 17-Feb-03
3940 ---------------------
3941
3942 1. If a comment in an extended regex that started immediately after a meta-item
3943 extended to the end of string, PCRE compiled incorrect data. This could lead to
3944 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3945
3946 2. Moved to autoconf 2.53 and libtool 1.4.2.
3947
3948 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3949 the special perltest8 script is no longer needed - all the tests can be run
3950 from a single perltest script.
3951
3952 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3953 by \s. It has now been removed in PCRE. This means it isn't recognized as
3954 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3955 class [:space:] *does* include VT, thereby creating a mess.
3956
3957 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3958 space and tab.
3959
3960 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3961 its new features into the main test script, reducing the number of scripts.
3962
3963 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3964 were backward compatible, and made the (?i) apply to the whole pattern, as if
3965 /i were given. Now it behaves more logically, and applies the option setting
3966 only to what follows. PCRE has been changed to follow suit. However, if it
3967 finds options settings right at the start of the pattern, it extracts them into
3968 the global options, as before. Thus, they show up in the info data.
3969
3970 8. Added support for the \Q...\E escape sequence. Characters in between are
3971 treated as literals. This is slightly different from Perl in that $ and @ are
3972 also handled as literals inside the quotes. In Perl, they will cause variable
3973 interpolation. Note the following examples:
3974
3975 Pattern PCRE matches Perl matches
3976
3977 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3978 \Qabc\$xyz\E abc\$xyz abc\$xyz
3979 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3980
3981 For compatibility with Perl, \Q...\E sequences are recognized inside character
3982 classes as well as outside them.
3983
3984 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3985 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3986 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3987 signed/unsigned warnings.
3988
3989 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3990 option for pcretest, so I've replaced it by a simple function that does just
3991 that job.
3992
3993 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3994 "pcregrep -".
3995
3996 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3997 Java package. This provides some syntactic sugar for simple cases of what my
3998 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3999 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4000 item, you can use this simplified notation. Note that only makes sense with
4001 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4002 greediness, whatever the setting of the PCRE_UNGREEDY option.
4003
4004 13. A change of greediness default within a pattern was not taking effect at
4005 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4006 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4007 was abstracted outside.
4008
4009 14. PCRE now supports the \G assertion. It is true when the current matching
4010 position is at the start point of the match. This differs from \A when the
4011 starting offset is non-zero. Used with the /g option of pcretest (or similar
4012 code), it works in the same way as it does for Perl's /g option. If all
4013 alternatives of a regex begin with \G, the expression is anchored to the start
4014 match position, and the "anchored" flag is set in the compiled expression.
4015
4016 15. Some bugs concerning the handling of certain option changes within patterns
4017 have been fixed. These applied to options other than (?ims). For example,
4018 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4019 been the other way round. Some of this was related to change 7 above.
4020
4021 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4022 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4023 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4024 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4025
4026 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4027 mode. Unlike ".", it always matches newline, whatever the setting of
4028 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4029 assertions. Perl allows it, but it doesn't (in general) work because it can't
4030 calculate the length of the lookbehind. At least, that's the case for Perl
4031 5.8.0 - I've been told they are going to document that it doesn't work in
4032 future.
4033
4034 18. Added an error diagnosis for escapes that PCRE does not support: these are
4035 \L, \l, \N, \P, \p, \U, \u, and \X.
4036
4037 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4038 reading past the end of the pattern in cases such as /[abcd/.
4039
4040 20. PCRE was getting more memory than necessary for patterns with classes that
4041 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4042
4043 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4044 compiling PCRE for use with Virtual Pascal.
4045
4046 22. Small fix to the Makefile to make it work properly if the build is done
4047 outside the source tree.
4048
4049 23. Added a new extension: a condition to go with recursion. If a conditional
4050 subpattern starts with (?(R) the "true" branch is used if recursion has
4051 happened, whereas the "false" branch is used only at the top level.
4052
4053 24. When there was a very long string of literal characters (over 255 bytes
4054 without UTF support, over 250 bytes with UTF support), the computation of how
4055 much memory was required could be incorrect, leading to segfaults or other
4056 strange effects.
4057
4058 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4059 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4060 there was a subsequent back reference to those brackets. This meant that, for
4061 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4062 possible to check for precisely this case. All we can do is abandon the
4063 optimization if .* occurs inside capturing brackets when there are any back
4064 references whatsoever. (See below for a better fix that came later.)
4065
4066 26. The handling of the optimization for finding the first character of a
4067 non-anchored pattern, and for finding a character that is required later in the
4068 match were failing in some cases. This didn't break the matching; it just
4069 failed to optimize when it could. The way this is done has been re-implemented.
4070
4071 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4072
4073 28. Added a new feature that provides some of the functionality that Perl
4074 provides with (?{...}). The facility is termed a "callout". The way it is done
4075 in PCRE is for the caller to provide an optional function, by setting
4076 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4077 global variable. By default it is unset, which disables all calling out. To get
4078 the function called, the regex must include (?C) at appropriate points. This
4079 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4080 This provides a means of identifying different callout points. When PCRE
4081 reaches such a point in the regex, if pcre_callout has been set, the external
4082 function is called. It is provided with data in a structure called
4083 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4084 matching continues; if it returns a non-zero value, the match at the current
4085 point fails. However, backtracking will occur if possible. [This was changed
4086 later and other features added - see item 49 below.]
4087
4088 29. pcretest is upgraded to test the callout functionality. It provides a
4089 callout function that displays information. By default, it shows the start of
4090 the match and the current position in the text. There are some new data escapes
4091 to vary what happens:
4092
4093 \C+ in addition, show current contents of captured substrings
4094 \C- do not supply a callout function
4095 \C!n return 1 when callout number n is reached
4096 \C!n!m return 1 when callout number n is reached for the mth time
4097
4098 30. If pcregrep was called with the -l option and just a single file name, it
4099 output "<stdin>" if a match was found, instead of the file name.
4100
4101 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4102 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4103 pcre_exec(). This saves a malloc/free per call. The default value of
4104 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4105 when configuring.
4106
4107 32. The default maximum size of a compiled pattern is 64K. There have been a
4108 few cases of people hitting this limit. The code now uses macros to handle the
4109 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4110 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4111 configuring. Tests 2 and 5 work only with 2-byte links because they output
4112 debugging information about compiled patterns.
4113
4114 33. Internal code re-arrangements:
4115
4116 (a) Moved the debugging function for printing out a compiled regex into
4117 its own source file (printint.c) and used #include to pull it into
4118 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4119 separate copies.
4120
4121 (b) Defined the list of op-code names for debugging as a macro in
4122 internal.h so that it is next to the definition of the opcodes.
4123
4124 (c) Defined a table of op-code lengths for simpler skipping along compiled
4125 code. This is again a macro in internal.h so that it is next to the
4126 definition of the opcodes.
4127
4128 34. Added support for recursive calls to individual subpatterns, along the
4129 lines of Robin Houston's patch (but implemented somewhat differently).
4130
4131 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4132 allow it to read and process whole directories in Win32. This code was
4133 contributed by Lionel Fourquaux; it has not been tested by me.
4134
4135 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4136 used to name a group. Names consist of alphanumerics and underscores, and must
4137 be unique. Back references use the syntax (?P=name) and recursive calls use
4138 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4139 numbers. The function pcre_fullinfo() can be used after compilation to extract
4140 a name/number map. There are three relevant calls:
4141
4142 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4143 PCRE_INFO_NAMECOUNT yields the number of entries
4144 PCRE_INFO_NAMETABLE yields a pointer to the map.
4145
4146 The map is a vector of fixed-size entries. The size of each entry depends on
4147 the length of the longest name used. The first two bytes of each entry are the
4148 group number, most significant byte first. There follows the corresponding
4149 name, zero terminated. The names are in alphabetical order.
4150
4151 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4152 case instead of 255. Making it the same both with and without UTF-8 support
4153 means that the same test output works with both.
4154
4155 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4156 calling malloc() with a zero argument.
4157
4158 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4159 optimization. I've improved things by keeping a bitmap of backreferences with
4160 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4161 fact referenced, the optimization can be applied. It is unlikely that a
4162 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4163 the match to follow \n) will appear inside brackets with a number greater than
4164 31, but if it does, any back reference > 31 suppresses the optimization.
4165
4166 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4167 of disabling numbered capturing parentheses. Any opening parenthesis that is
4168 not followed by ? behaves as if it were followed by ?: but named parentheses
4169 can still be used for capturing (and they will acquire numbers in the usual
4170 way).
4171
4172 41. Redesigned the return codes from the match() function into yes/no/error so
4173 that errors can be passed back from deep inside the nested calls. A malloc
4174 failure while inside a recursive subpattern call now causes the
4175 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4176
4177 42. It is now possible to set a limit on the number of times the match()
4178 function is called in a call to pcre_exec(). This facility makes it possible to
4179 limit the amount of recursion and backtracking, though not in a directly
4180 obvious way, because the match() function is used in a number of different
4181 circumstances. The count starts from zero for each position in the subject
4182 string (for non-anchored patterns). The default limit is, for compatibility, a
4183 large number, namely 10 000 000. You can change this in two ways:
4184
4185 (a) When configuring PCRE before making, you can use --with-match-limit=n
4186 to set a default value for the compiled library.
4187
4188 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4189 a different value is set. See 45 below.
4190
4191 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4192
4193 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4194 of things that can be changed at compile time. The first argument specifies
4195 what is wanted and the second points to where the information is to be placed.
4196 The current list of available information is:
4197
4198 PCRE_CONFIG_UTF8
4199
4200 The output is an integer that is set to one if UTF-8 support is available;
4201 otherwise it is set to zero.
4202
4203 PCRE_CONFIG_NEWLINE
4204
4205 The output is an integer that it set to the value of the code that is used for
4206 newline. It is either LF (10) or CR (13).
4207
4208 PCRE_CONFIG_LINK_SIZE
4209
4210 The output is an integer that contains the number of bytes used for internal
4211 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4212
4213 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4214
4215 The output is an integer that contains the threshold above which the POSIX
4216 interface uses malloc() for output vectors. See item 31 above.
4217
4218 PCRE_CONFIG_MATCH_LIMIT
4219
4220 The output is an unsigned integer that contains the default limit of the number
4221 of match() calls in a pcre_exec() execution. See 42 above.
4222
4223 44. pcretest has been upgraded by the addition of the -C option. This causes it
4224 to extract all the available output from the new pcre_config() function, and to
4225 output it. The program then exits immediately.
4226
4227 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4228 order to support additional features. One way would have been to define
4229 pcre_exec2() (for example) with extra arguments, but this would not have been
4230 extensible, and would also have required all calls to the original function to
4231 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4232 is used for passing in "extra" data from pcre_study().
4233
4234 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4235 contains the following fields:
4236
4237 flags a bitmap indicating which of the following fields are set
4238 study_data opaque data from pcre_study()
4239 match_limit a way of specifying a limit on match() calls for a specific
4240 call to pcre_exec()
4241 callout_data data for callouts (see 49 below)
4242
4243 The flag bits are also defined in pcre.h, and are
4244
4245 PCRE_EXTRA_STUDY_DATA
4246 PCRE_EXTRA_MATCH_LIMIT
4247 PCRE_EXTRA_CALLOUT_DATA
4248
4249 The pcre_study() function now returns one of these new pcre_extra blocks, with
4250 the actual study data pointed to by the study_data field, and the
4251 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4252 before. That is, this change is entirely upwards-compatible and requires no
4253 change to existing code.
4254
4255 If you want to pass in additional data to pcre_exec(), you can either place it
4256 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4257 block.
4258
4259 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4260 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4261 times with different match limits, until it finds the minimum value needed for
4262 pcre_exec() to complete. The value is then output. This can be instructive; for
4263 most simple matches the number is quite small, but for pathological cases it
4264 gets very large very quickly.
4265
4266 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4267 returns the size of the data block pointed to by the study_data field in a
4268 pcre_extra block, that is, the value that was passed as the argument to
4269 pcre_malloc() when PCRE was getting memory in which to place the information
4270 created by pcre_study(). The fourth argument should point to a size_t variable.
4271 pcretest has been extended so that this information is shown after a successful
4272 pcre_study() call when information about the compiled regex is being displayed.
4273
4274 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4275 because what follows is always an absolute path. (Later: it turns out that this
4276 is more than cosmetic for MinGW, because it doesn't like empty path
4277 components.)
4278
4279 49. Some changes have been made to the callout feature (see 28 above):
4280
4281 (i) A callout function now has three choices for what it returns:
4282
4283 0 => success, carry on matching
4284 > 0 => failure at this point, but backtrack if possible
4285 < 0 => serious error, return this value from pcre_exec()
4286
4287 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4288 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4289 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4290 use by callout functions. It will never be used by PCRE itself.
4291
4292 (ii) The pcre_extra structure (see 45 above) has a void * field called
4293 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4294 pcre_callout_block structure has a field of the same name. The contents of
4295 the field passed in the pcre_extra structure are passed to the callout
4296 function in the corresponding field in the callout block. This makes it
4297 easier to use the same callout-containing regex from multiple threads. For
4298 testing, the pcretest program has a new data escape
4299
4300 \C*n pass the number n (may be negative) as callout_data
4301
4302 If the callout function in pcretest receives a non-zero value as
4303 callout_data, it returns that value.
4304
4305 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4306 there were some redundant $(CFLAGS) in commands that are now specified as
4307 $(LINK), which already includes $(CFLAGS).
4308
4309 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4310 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4311 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4312 one-byte characters throughout. Note that case-insensitive matching applies
4313 only to characters whose values are less than 256. PCRE doesn't support the
4314 notion of cases for higher-valued characters.
4315
4316 (i) A character class whose characters are all within 0-255 is handled as
4317 a bit map, and the map is inverted for negative classes. Previously, a
4318 character > 255 always failed to match such a class; however it should
4319 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4320
4321 (ii) A negated character class with a single character < 255 is coded as
4322 "not this character" (OP_NOT). This wasn't working properly when the test
4323 character was multibyte, either singly or repeated.
4324
4325 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4326 mode, for example: \x{100}{2,3}.
4327
4328 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4329 singly or repeated) now correctly test multibyte characters. However,
4330 PCRE doesn't recognize any characters with values greater than 255 as
4331 digits, spaces, or word characters. Such characters always match \D, \S,
4332 and \W, and never match \d, \s, or \w.
4333
4334 (v) Classes may now contain characters and character ranges with values
4335 greater than 255. For example: [ab\x{100}-\x{400}].
4336
4337 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4338 PCRE in UTF-8 mode.
4339
4340 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4341 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4342 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4343 value.)
4344
4345 53. The single man page has become too large. I have therefore split it up into
4346 a number of separate man pages. These also give rise to individual HTML pages;
4347 these are now put in a separate directory, and there is an index.html page that
4348 lists them all. Some hyperlinking between the pages has been installed.
4349
4350 54. Added convenience functions for handling named capturing parentheses.
4351
4352 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4353 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4354 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4355 are faulted.
4356
4357 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4358 calling configure. These values are used when compiling the dftables.c program
4359 which is run to generate the source of the default character tables. They
4360 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4361 you will need to set these values.
4362
4363 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4364
4365
4366 Version 3.9 02-Jan-02
4367 ---------------------
4368
4369 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4370
4371 2. If --disable-static was given, the building process failed when trying to
4372 build pcretest and pcregrep. (For some reason it was using libtool to compile
4373 them, which is not right, as they aren't part of the library.)
4374
4375
4376 Version 3.8 18-Dec-01
4377 ---------------------
4378
4379 1. The experimental UTF-8 code was completely screwed up. It was packing the
4380 bytes in the wrong order. How dumb can you get?
4381
4382
4383 Version 3.7 29-Oct-01
4384 ---------------------
4385
4386 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4387 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4388 this didn't happen under Solaris 8, where I normally test things.
4389
4390 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4391 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4392 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4393 this fix an hour or so after the initial 3.7 release.)
4394
4395
4396 Version 3.6 23-Oct-01
4397 ---------------------
4398
4399 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4400 offsets passed as NULL with zero offset count.
4401
4402 2. The config.guess and config.sub files had not been updated when I moved to
4403 the latest autoconf.
4404
4405
4406 Version 3.5 15-Aug-01
4407 ---------------------
4408
4409 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4410 had been forgotten.
4411
4412 2. By using declared but undefined structures, we can avoid using "void"
4413 definitions in pcre.h while keeping the internal definitions of the structures
4414 private.
4415
4416 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4417 user point of view, this means that both static and shared libraries are built
4418 by default, but this can be individually controlled. More of the work of
4419 handling this static/shared cases is now inside libtool instead of PCRE's make
4420 file.
4421
4422 4. The pcretest utility is now installed along with pcregrep because it is
4423 useful for users (to test regexs) and by doing this, it automatically gets
4424 relinked by libtool. The documentation has been turned into a man page, so
4425 there are now .1, .txt, and .html versions in /doc.
4426
4427 5. Upgrades to pcregrep:
4428 (i) Added long-form option names like gnu grep.
4429 (ii) Added --help to list all options with an explanatory phrase.
4430 (iii) Added -r, --recursive to recurse into sub-directories.
4431 (iv) Added -f, --file to read patterns from a file.
4432
4433 6. pcre_exec() was referring to its "code" argument before testing that
4434 argument for NULL (and giving an error if it was NULL).
4435
4436 7. Upgraded Makefile.in to allow for compiling in a different directory from
4437 the source directory.
4438
4439 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4440 options bits, the pointer it was passed was to an int instead of to an unsigned
4441 long int. This mattered only on 64-bit systems.
4442
4443 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4444 generated) instead of pcre.in, which it its source. Also made the same change
4445 in several of the .c files.
4446
4447 10. A new release of gcc defines printf() as a macro, which broke pcretest
4448 because it had an ifdef in the middle of a string argument for printf(). Fixed
4449 by using separate calls to printf().
4450
4451 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4452 script, to force use of CR or LF instead of \n in the source. On non-Unix
4453 systems, the value can be set in config.h.
4454
4455 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4456 absolute limit. Changed the text of the error message to make this clear, and
4457 likewise updated the man page.
4458
4459 13. The limit of 99 on the number of capturing subpatterns has been removed.
4460 The new limit is 65535, which I hope will not be a "real" limit.
4461
4462
4463 Version 3.4 22-Aug-00
4464 ---------------------
4465
4466 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4467
4468 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4469
4470
4471 Version 3.3 01-Aug-00
4472 ---------------------
4473
4474 1. If an octal character was given, but the value was greater than \377, it
4475 was not getting masked to the least significant bits, as documented. This could
4476 lead to crashes in some systems.
4477
4478 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4479 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4480
4481 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4482 These just pass their arguments on to (pcre_free)(), but they are provided
4483 because some uses of PCRE bind it to non-C systems that can call its functions,
4484 but cannot call free() or pcre_free() directly.
4485
4486 4. Add "make test" as a synonym for "make check". Corrected some comments in
4487 the Makefile.
4488
4489 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4490 Makefile.
4491
4492 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4493 command called pgrep for grepping around the active processes.
4494
4495 7. Added the beginnings of support for UTF-8 character strings.
4496
4497 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4498 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4499 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4500 out for the ar command.)
4501
4502
4503 Version 3.2 12-May-00
4504 ---------------------
4505
4506 This is purely a bug fixing release.
4507
4508 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4509 of ZA. This was just one example of several cases that could provoke this bug,
4510 which was introduced by change 9 of version 2.00. The code for breaking
4511 infinite loops after an iteration that matches an empty string was't working
4512 correctly.
4513
4514 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4515 when matched against abbab (for example). After matching an empty string, it
4516 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4517 caused it to match further down the string than it should.
4518
4519 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4520 was there because it doesn't seem to be needed, and it causes trouble on some
4521 systems, as it is not a Standard C header. It has been removed.
4522
4523 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4524 were reported on the Macintosh. The changes were from
4525
4526 while ((c = *(++ptr)) != 0 && c != '\n');
4527 to
4528 while ((c = *(++ptr)) != 0 && c != '\n') ;
4529
4530 Totally extraordinary, but if that's what it takes...
4531
4532 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4533 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4534 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4535 assumes the way PCRE uses memmove() (always moving upwards).
4536
4537 6. PCRE is being used in one environment where strchr() is not available. There
4538 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4539 faster code anyway.
4540
4541
4542 Version 3.1 09-Feb-00
4543 ---------------------
4544
4545 The only change in this release is the fixing of some bugs in Makefile.in for
4546 the "install" target:
4547
4548 (1) It was failing to install pcreposix.h.
4549
4550 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4551
4552
4553 Version 3.0 01-Feb-00
4554 ---------------------
4555
4556 1. Add support for the /+ modifier to perltest (to output $` like it does in
4557 pcretest).
4558
4559 2. Add support for the /g modifier to perltest.
4560
4561 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4562 matches null strings.
4563
4564 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4565 pattern. Perl treats empty patterns specially - it reuses the most recent
4566 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4567 effect.
4568
4569 5. The POSIX interface was broken in that it was just handing over the POSIX
4570 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4571 required a bigger vector, with some working space on the end. This means that
4572 the POSIX wrapper now has to get and free some memory, and copy the results.
4573
4574 6. Added some simple autoconf support, placing the test data and the
4575 documentation in separate directories, re-organizing some of the
4576 information files, and making it build pcre-config (a GNU standard). Also added
4577 libtool support for building PCRE as a shared library, which is now the
4578 default.
4579
4580 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4581 09 are not valid octal constants. Single digits will be used for minor values
4582 less than 10.
4583
4584 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4585 existing programs that set these in the POSIX interface can use PCRE without
4586 modification.
4587
4588 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4589 return all that pcre_info() returns, plus additional data. The pcre_info()
4590 function is retained for compatibility, but is considered to be obsolete.
4591
4592 10. Added experimental recursion feature (?R) to handle one common case that
4593 Perl 5.6 will be able to do with (?p{...}).
4594
4595 11. Added support for POSIX character classes like [:alpha:], which Perl is
4596 adopting.
4597
4598
4599 Version 2.08 31-Aug-99
4600 ----------------------
4601
4602 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4603 trying to match at the startoffset position, but instead was moving forward to
4604 the next newline as if a previous match had failed.
4605
4606 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4607 and could get into a loop if a null string was matched other than at the start
4608 of the subject.
4609
4610 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4611 be distinguished at compile time, and for completeness also added PCRE_DATE.
4612
4613 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4614 in GnuWin32 environments.
4615
4616
4617 Version 2.07 29-Jul-99
4618 ----------------------
4619
4620 1. The documentation is now supplied in plain text form and HTML as well as in
4621 the form of man page sources.
4622
4623 2. C++ compilers don't like assigning (void *) values to other pointer types.
4624 In particular this affects malloc(). Although there is no problem in Standard
4625 C, I've put in casts to keep C++ compilers happy.
4626
4627 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4628 should be (const char *).
4629
4630 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4631 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4632 However, I haven't made this a standard facility. The documentation doesn't
4633 mention it, and the Makefile doesn't support it.
4634
4635 5. The Makefile now contains an "install" target, with editable destinations at
4636 the top of the file. The pcretest program is not installed.
4637
4638 6. pgrep -V now gives the PCRE version number and date.
4639
4640 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4641 causing the entire string to be ignored, instead of just the last character.
4642
4643 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4644 non-matching string, it can take a very, very long time, even for strings of
4645 quite modest length, because of the nested recursion. PCRE now does better in
4646 some of these cases. It does this by remembering the last required literal
4647 character in the pattern, and pre-searching the subject to ensure it is present
4648 before running the real match. In other words, it applies a heuristic to detect
4649 some types of certain failure quickly, and in the above example, if presented
4650 with a string that has no trailing " it gives "no match" very quickly.
4651
4652 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4653 other alternatives are tried instead.
4654
4655
4656 Version 2.06 09-Jun-99
4657 ----------------------
4658
4659 1. Change pcretest's output for amount of store used to show just the code
4660 space, because the remainder (the data block) varies in size between 32-bit and
4661 64-bit systems.
4662
4663 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4664 start matching at. This allows lookbehinds to work when searching for multiple
4665 occurrences in a string.
4666
4667 3. Added additional options to pcretest for testing multiple occurrences:
4668
4669 /+ outputs the rest of the string that follows a match
4670 /g loops for multiple occurrences, using the new startoffset argument
4671 /G loops for multiple occurrences by passing an incremented pointer
4672
4673 4. PCRE wasn't doing the "first character" optimization for patterns starting
4674 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4675 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4676 the letter 'x'. On long subject strings, this gives a significant speed-up.
4677
4678
4679 Version 2.05 21-Apr-99
4680 ----------------------
4681
4682 1. Changed the type of magic_number from int to long int so that it works
4683 properly on 16-bit systems.
4684
4685 2. Fixed a bug which caused patterns starting with .* not to work correctly
4686 when the subject string contained newline characters. PCRE was assuming
4687 anchoring for such patterns in all cases, which is not correct because .* will
4688 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4689 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4690 must be retried after every newline in the subject.
4691
4692
4693 Version 2.04 18-Feb-99
4694 ----------------------
4695
4696 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4697 computation of the store needed to hold the pattern was incorrect (too large).
4698 If such patterns were nested a few deep, this could multiply and become a real
4699 problem.
4700
4701 2. Added /M option to pcretest to show the memory requirement of a specific
4702 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4703
4704 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4705 compiled in such a way that the backtracking after subsequent failure was
4706 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4707 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4708
4709
4710 Version 2.03 02-Feb-99
4711 ----------------------
4712
4713 1. Fixed typo and small mistake in man page.
4714
4715 2. Added 4th condition (GPL supersedes if conflict) and created separate
4716 LICENCE file containing the conditions.
4717
4718 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4719 Perl, that is the internal \ allows the delimiter to be included in the
4720 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4721 the final delimiter, add \ to the end of the pattern (to test the error).
4722
4723 4. Added the convenience functions for extracting substrings after a successful
4724 match. Updated pcretest to make it able to test these functions.
4725
4726
4727 Version 2.02 14-Jan-99
4728 ----------------------
4729
4730 1. Initialized the working variables associated with each extraction so that
4731 their saving and restoring doesn't refer to uninitialized store.
4732
4733 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4734 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4735 fix the problem.
4736
4737 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4738 calls, and wasn't printing the correct value for compiling calls. Increased the
4739 default value of LOOPREPEAT, and the number of significant figures in the
4740 times.
4741
4742 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4743
4744 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4745 a building problem on Windows NT with a FAT file system.
4746
4747
4748 Version 2.01 21-Oct-98
4749 ----------------------
4750
4751 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4752 to character tables built by pcre_maketables() in the current locale. If NULL
4753 is passed, the default tables are used.
4754
4755
4756 Version 2.00 24-Sep-98
4757 ----------------------
4758
4759 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4760 it any more.
4761
4762 2. Allow quantification of (?>) groups, and make it work correctly.
4763
4764 3. The first character computation wasn't working for (?>) groups.
4765
4766 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4767 end of the subject) and add 5.005's \z, which really does match only at the
4768 very end of the subject.
4769
4770 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4771
4772 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4773 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4774 localized options. All options to pcre_study() were also removed.
4775
4776 7. Add other new features from 5.005:
4777
4778 $(?<= positive lookbehind
4779 $(?<! negative lookbehind
4780 (?imsx-imsx) added the unsetting capability
4781 such a setting is global if at outer level; local otherwise
4782 (?imsx-imsx:) non-capturing groups with option setting
4783 (?(cond)re|re) conditional pattern matching
4784
4785 A backreference to itself in a repeated group matches the previous
4786 captured string.
4787
4788 8. General tidying up of studying (both automatic and via "study")
4789 consequential on the addition of new assertions.
4790
4791 9. As in 5.005, unlimited repeated groups that could match an empty substring
4792 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4793 runtime if any iteration does actually match an empty substring.
4794
4795 10. Include the RunTest script in the distribution.
4796
4797 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4798 discrepancies, some of which were old and were also with respect to 5.004. They
4799 have now been fixed.
4800
4801
4802 Version 1.09 28-Apr-98
4803 ----------------------
4804
4805 1. A negated single character class followed by a quantifier with a minimum
4806 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4807 program crashes, or just wrong answers. This did not apply to negated classes
4808 containing more than one character, or to minima other than one.
4809
4810
4811 Version 1.08 27-Mar-98
4812 ----------------------
4813
4814 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4815
4816 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4817 latter must appear before anything that relies on it in the pattern.
4818
4819
4820 Version 1.07 16-Feb-98
4821 ----------------------
4822
4823 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4824 repeat of a potentially empty string).
4825
4826
4827 Version 1.06 23-Jan-98
4828 ----------------------
4829
4830 1. Added Markus Oberhumer's little patches for C++.
4831
4832 2. Literal strings longer than 255 characters were broken.
4833
4834
4835 Version 1.05 23-Dec-97
4836 ----------------------
4837
4838 1. Negated character classes containing more than one character were failing if
4839 PCRE_CASELESS was set at run time.
4840
4841
4842 Version 1.04 19-Dec-97
4843 ----------------------
4844
4845 1. Corrected the man page, where some "const" qualifiers had been omitted.
4846
4847 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4848 input syntax.
4849
4850 3. Fixed memory leak which occurred when a regex with back references was
4851 matched with an offsets vector that wasn't big enough. The temporary memory
4852 that is used in this case wasn't being freed if the match failed.
4853
4854 4. Tidied pcretest to ensure it frees memory that it gets.
4855
4856 5. Temporary memory was being obtained in the case where the passed offsets
4857 vector was exactly big enough.
4858
4859 6. Corrected definition of offsetof() from change 5 below.
4860
4861 7. I had screwed up change 6 below and broken the rules for the use of
4862 setjmp(). Now fixed.
4863
4864
4865 Version 1.03 18-Dec-97
4866 ----------------------
4867
4868 1. A erroneous regex with a missing opening parenthesis was correctly
4869 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4870 on some systems.
4871
4872 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4873 it was reported that one broken compiler failed on the former because "code" is
4874 also an independent variable.
4875
4876 3. The erroneous regex a[]b caused an array overrun reference.
4877
4878 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4879 fail on data ending with that character. (It was going on too far, and checking
4880 the next character, typically a binary zero.) This was specific to the
4881 optimized code for single-character negative classes.
4882
4883 5. Added a contributed patch from the TIN world which does the following:
4884
4885 + Add an undef for memmove, in case the the system defines a macro for it.
4886
4887 + Add a definition of offsetof(), in case there isn't one. (I don't know
4888 the reason behind this - offsetof() is part of the ANSI standard - but
4889 it does no harm).
4890
4891 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4892 most of the places where whitespace preceded '#'. I have given up and
4893 allowed the remaining 2 cases to be at the margin.
4894
4895 + Rename some variables in pcre to eliminate shadowing. This seems very
4896 pedantic, but does no harm, of course.
4897
4898 6. Moved the call to setjmp() into its own function, to get rid of warnings
4899 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4900
4901 7. Constructs such as \d{8,} were compiling into the equivalent of
4902 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4903 outcome, but in this particular case used more store than had been allocated,
4904 which caused the bug to be discovered because it threw up an internal error.
4905
4906 8. The debugging code in both pcre and pcretest for outputting the compiled
4907 form of a regex was going wrong in the case of back references followed by
4908 curly-bracketed repeats.
4909
4910
4911 Version 1.02 12-Dec-97
4912 ----------------------
4913
4914 1. Typos in pcre.3 and comments in the source fixed.
4915
4916 2. Applied a contributed patch to get rid of places where it used to remove
4917 'const' from variables, and fixed some signed/unsigned and uninitialized
4918 variable warnings.
4919
4920 3. Added the "runtest" target to Makefile.
4921
4922 4. Set default compiler flag to -O2 rather than just -O.
4923
4924
4925 Version 1.01 19-Nov-97
4926 ----------------------
4927
4928 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4929 like /([ab]*)*/, that is, for classes with more than one character in them.
4930
4931 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4932 as /((?>a*))*/ (a PCRE_EXTRA facility).
4933
4934
4935 Version 1.00 18-Nov-97
4936 ----------------------
4937
4938 1. Added compile-time macros to support systems such as SunOS4 which don't have
4939 memmove() or strerror() but have other things that can be used instead.
4940
4941 2. Arranged that "make clean" removes the executables.
4942
4943
4944 Version 0.99 27-Oct-97
4945 ----------------------
4946
4947 1. Fixed bug in code for optimizing classes with only one character. It was
4948 initializing a 32-byte map regardless, which could cause it to run off the end
4949 of the memory it had got.
4950
4951 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4952
4953
4954 Version 0.98 22-Oct-97
4955 ----------------------
4956
4957 1. Fixed bug in code for handling temporary memory usage when there are more
4958 back references than supplied space in the ovector. This could cause segfaults.
4959
4960
4961 Version 0.97 21-Oct-97
4962 ----------------------
4963
4964 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4965
4966 2. Optimized negated single characters not to use a bit map.
4967
4968 3. Brought error texts together as macro definitions; clarified some of them;
4969 fixed one that was wrong - it said "range out of order" when it meant "invalid
4970 escape sequence".
4971
4972 4. Changed some char * arguments to const char *.
4973
4974 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4975
4976 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4977 pcretest.
4978
4979
4980 Version 0.96 16-Oct-97
4981 ----------------------
4982
4983 1. Added a simple "pgrep" utility to the distribution.
4984
4985 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4986 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4987 where "ddd" means "one or more decimal digits".
4988
4989 3. Fixed serious bug. If a pattern had a back reference, but the call to
4990 pcre_exec() didn't supply a large enough ovector to record the related
4991 identifying subpattern, the match always failed. PCRE now remembers the number
4992 of the largest back reference, and gets some temporary memory in which to save
4993 the offsets during matching if necessary, in order to ensure that
4994 backreferences always work.
4995
4996 4. Increased the compatibility with Perl in a number of ways:
4997
4998 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4999 to request this handling. The option can be set at compile or exec time.
5000
5001 (b) $ matches before a terminating newline by default; an option
5002 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5003 mode). The option can be set at compile or exec time.
5004
5005 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5006 the same as Perl's. If the decimal number it represents is less than 10
5007 or there aren't that many previous left capturing parentheses, an octal
5008 escape is read. Inside a character class, it's always an octal escape,
5009 even if it is a single digit.
5010
5011 (d) An escaped but undefined alphabetic character is taken as a literal,
5012 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5013 escapes.
5014
5015 (e) {0} is now permitted. (The previous item is removed from the compiled
5016 pattern).
5017
5018 5. Changed all the names of code files so that the basic parts are no longer
5019 than 10 characters, and abolished the teeny "globals.c" file.
5020
5021 6. Changed the handling of character classes; they are now done with a 32-byte
5022 bit map always.
5023
5024 7. Added the -d and /D options to pcretest to make it possible to look at the
5025 internals of compilation without having to recompile pcre.
5026
5027
5028 Version 0.95 23-Sep-97
5029 ----------------------
5030
5031 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5032 \x20 at the start of a run of normal characters. These were being treated as
5033 real characters, instead of the source characters being re-checked.
5034
5035
5036 Version 0.94 18-Sep-97
5037 ----------------------
5038
5039 1. The functions are now thread-safe, with the caveat that the global variables
5040 containing pointers to malloc() and free() or alternative functions are the
5041 same for all threads.
5042
5043 2. Get pcre_study() to generate a bitmap of initial characters for non-
5044 anchored patterns when this is possible, and use it if passed to pcre_exec().
5045
5046
5047 Version 0.93 15-Sep-97
5048 ----------------------
5049
5050 1. /(b)|(:+)/ was computing an incorrect first character.
5051
5052 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5053 but not actually doing anything yet.
5054
5055 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5056 as Perl does (e.g. [-az] or [az-]).
5057
5058 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5059 all possible positions.
5060
5061 5. Split up into different modules to avoid including unneeded functions in a
5062 compiled binary. However, compile and exec are still in one module. The "study"
5063 function is split off.
5064
5065 6. The character tables are now in a separate module whose source is generated
5066 by an auxiliary program - but can then be edited by hand if required. There are
5067 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5068 toupper() in the code.
5069
5070 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5071 make them global. Abolish the function for setting them, as the caller can now
5072 set them directly.
5073
5074
5075 Version 0.92 11-Sep-97
5076 ----------------------
5077
5078 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5079 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5080
5081 2. Caseless matching was not working in character classes if the characters in
5082 the pattern were in upper case.
5083
5084 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5085
5086 4. Make PCRE_ANCHORED public and accept as a compile option.
5087
5088 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5089 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5090 pass them.
5091
5092 6. Give an error if bad option bits passed at compile or run time.
5093
5094 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5095 pcretest to cause it to pass that flag.
5096
5097 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5098 options, and the first character, if set.
5099
5100 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5101
5102
5103 Version 0.91 10-Sep-97
5104 ----------------------
5105
5106 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5107 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5108
5109 2. PCRE was looping on encountering an indefinitely repeated back reference to
5110 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5111 Perl does - treats the match as successful.
5112
5113 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5