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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1309 - (show annotations)
Fri Apr 5 15:35:59 2013 UTC (6 years, 6 months ago) by ph10
File size: 245924 byte(s)
Implement PCRE_NEVER_UTF
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 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
136 blocking (*UTF) etc.
137
138
139 Version 8.32 30-November-2012
140 -----------------------------
141
142 1. Improved JIT compiler optimizations for first character search and single
143 character iterators.
144
145 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
146 Patch by Daniel Richard G.
147
148 3. Single character iterator optimizations in the JIT compiler.
149
150 4. Improved JIT compiler optimizations for character ranges.
151
152 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
153 Reported by Giuseppe D'Angelo.
154
155 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
156 of a line, was being set incorrectly in cases where .* appeared inside
157 atomic brackets at the start of a pattern, or where there was a subsequent
158 *PRUNE or *SKIP.
159
160 7. Improved instruction cache flush for POWER/PowerPC.
161 Patch by Daniel Richard G.
162
163 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
164 grep:
165
166 (a) There is now no limit to the number of patterns to be matched.
167
168 (b) An error is given if a pattern is too long.
169
170 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
171 are now supported.
172
173 (d) --exclude-from and --include-from (multiple use) have been added.
174
175 (e) Exclusions and inclusions now apply to all files and directories, not
176 just to those obtained from scanning a directory recursively.
177
178 (f) Multiple uses of -f and --file-list are now supported.
179
180 (g) In a Windows environment, the default for -d has been changed from
181 "read" (the GNU grep default) to "skip", because otherwise the presence
182 of a directory in the file list provokes an error.
183
184 (h) The documentation has been revised and clarified in places.
185
186 9. Improve the matching speed of capturing brackets.
187
188 10. Changed the meaning of \X so that it now matches a Unicode extended
189 grapheme cluster.
190
191 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
192 out POSIX threads when JIT support is configured.
193
194 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
195
196 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
197 structure could go wrong in environments where size_t is not the same size
198 as int.
199
200 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
201 set.
202
203 15. The EBCDIC support had decayed; later updates to the code had included
204 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
205 general tidy up of EBCDIC-related issues, and the documentation was also
206 not quite right. There is now a test that can be run on ASCII systems to
207 check some of the EBCDIC-related things (but is it not a full test).
208
209 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
210 in a small tidy to the code.
211
212 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
213
214 18. If the --only-matching (-o) option in pcregrep is specified multiple
215 times, each one causes appropriate output. For example, -o1 -o2 outputs the
216 substrings matched by the 1st and 2nd capturing parentheses. A separating
217 string can be specified by --om-separator (default empty).
218
219 19. Improving the first n character searches.
220
221 20. Turn case lists for horizontal and vertical white space into macros so that
222 they are defined only once.
223
224 21. This set of changes together give more compatible Unicode case-folding
225 behaviour for characters that have more than one other case when UCP
226 support is available.
227
228 (a) The Unicode property table now has offsets into a new table of sets of
229 three or more characters that are case-equivalent. The MultiStage2.py
230 script that generates these tables (the pcre_ucd.c file) now scans
231 CaseFolding.txt instead of UnicodeData.txt for character case
232 information.
233
234 (b) The code for adding characters or ranges of characters to a character
235 class has been abstracted into a generalized function that also handles
236 case-independence. In UTF-mode with UCP support, this uses the new data
237 to handle characters with more than one other case.
238
239 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
240 whose other case is greater than 256 are now correctly matched
241 caselessly. Previously, the high codepoint matched the low one, but not
242 vice versa.
243
244 (d) The processing of \h, \H, \v, and \ in character classes now makes use
245 of the new class addition function, using character lists defined as
246 macros alongside the case definitions of 20 above.
247
248 (e) Caseless back references now work with characters that have more than
249 one other case.
250
251 (f) General caseless matching of characters with more than one other case
252 is supported.
253
254 22. Unicode character properties were updated from Unicode 6.2.0
255
256 23. Improved CMake support under Windows. Patch by Daniel Richard G.
257
258 24. Add support for 32-bit character strings, and UTF-32
259
260 25. Major JIT compiler update (code refactoring and bugfixing).
261 Experimental Sparc 32 support is added.
262
263 26. Applied a modified version of Daniel Richard G's patch to create
264 pcre.h.generic and config.h.generic by "make" instead of in the
265 PrepareRelease script.
266
267 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
268 pcre_compile.c when checking for a zero character.
269
270 28. Introducing a native interface for JIT. Through this interface, the compiled
271 machine code can be directly executed. The purpose of this interface is to
272 provide fast pattern matching, so several sanity checks are not performed.
273 However, feature tests are still performed. The new interface provides
274 1.4x speedup compared to the old one.
275
276 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
277 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
278 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
279
280 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
281 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
282 the "old" RFC 2279). Instead, it ended up passing a negative length to
283 pcre_exec().
284
285 31. Add support for GCC's visibility feature to hide internal functions.
286
287 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
288 "unknown -C option" after outputting 0 or 1.
289
290 33. There is now support for generating a code coverage report for the test
291 suite in environments where gcc is the compiler and lcov is installed. This
292 is mainly for the benefit of the developers.
293
294 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
295 unaddressable using valgrind annotations, allowing valgrind to detect
296 invalid memory accesses. This is mainly for the benefit of the developers.
297
298 25. (*UTF) can now be used to start a pattern in any of the three libraries.
299
300 26. Give configure error if --enable-cpp but no C++ compiler found.
301
302
303 Version 8.31 06-July-2012
304 -------------------------
305
306 1. Fixing a wrong JIT test case and some compiler warnings.
307
308 2. Removed a bashism from the RunTest script.
309
310 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
311 to unsigned type, result still unsigned" that was given by an MS compiler
312 on encountering the code "-sizeof(xxx)".
313
314 4. Partial matching support is added to the JIT compiler.
315
316 5. Fixed several bugs concerned with partial matching of items that consist
317 of more than one character:
318
319 (a) /^(..)\1/ did not partially match "aba" because checking references was
320 done on an "all or nothing" basis. This also applied to repeated
321 references.
322
323 (b) \R did not give a hard partial match if \r was found at the end of the
324 subject.
325
326 (c) \X did not give a hard partial match after matching one or more
327 characters at the end of the subject.
328
329 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
330 a partial match for the string "\r".
331
332 (e) When newline was set to CRLF, the metacharacter "." did not recognize
333 a partial match for a CR character at the end of the subject string.
334
335 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
336 running pcretest, the text "(JIT)" added to the output whenever JIT is
337 actually used to run the match.
338
339 7. Individual JIT compile options can be set in pcretest by following -s+[+]
340 or /S+[+] with a digit between 1 and 7.
341
342 8. OP_NOT now supports any UTF character not just single-byte ones.
343
344 9. (*MARK) control verb is now supported by the JIT compiler.
345
346 10. The command "./RunTest list" lists the available tests without actually
347 running any of them. (Because I keep forgetting what they all are.)
348
349 11. Add PCRE_INFO_MAXLOOKBEHIND.
350
351 12. Applied a (slightly modified) user-supplied patch that improves performance
352 when the heap is used for recursion (compiled with --disable-stack-for-
353 recursion). Instead of malloc and free for each heap frame each time a
354 logical recursion happens, frames are retained on a chain and re-used where
355 possible. This sometimes gives as much as 30% improvement.
356
357 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
358 call.
359
360 14. As documented, (*COMMIT) is now confined to within a positive assertion.
361
362 15. It is now possible to link pcretest with libedit as an alternative to
363 libreadline.
364
365 16. (*COMMIT) control verb is now supported by the JIT compiler.
366
367 17. The Unicode data tables have been updated to Unicode 6.1.0.
368
369 18. Added --file-list option to pcregrep.
370
371 19. Added binary file support to pcregrep, including the -a, --binary-files,
372 -I, and --text options.
373
374 20. The madvise function is renamed for posix_madvise for QNX compatibility
375 reasons. Fixed by Giuseppe D'Angelo.
376
377 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
378
379 22. Changed the option for creating symbolic links for 16-bit man pages from
380 -s to -sf so that re-installing does not cause issues.
381
382 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
383
384 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
385 matches in certain environments (the workspace was not being correctly
386 retained). Also added to pcre_dfa_exec() a simple plausibility check on
387 some of the workspace data at the beginning of a restart.
388
389 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
390 was not doing so when it should - probably a typo introduced by SVN 528
391 (change 8.10/14).
392
393 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
394 \w+ when the character tables indicated that \x{c4} was a word character.
395 There were several related cases, all because the tests for doing a table
396 lookup were testing for characters less than 127 instead of 255.
397
398 27. If a pattern contains capturing parentheses that are not used in a match,
399 their slots in the ovector are set to -1. For those that are higher than
400 any matched groups, this happens at the end of processing. In the case when
401 there were back references that the ovector was too small to contain
402 (causing temporary malloc'd memory to be used during matching), and the
403 highest capturing number was not used, memory off the end of the ovector
404 was incorrectly being set to -1. (It was using the size of the temporary
405 memory instead of the true size.)
406
407 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
408 ovector size, it uses memory at the end of the block that it has got.
409
410 29. Check for an overlong MARK name and give an error at compile time. The
411 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
412
413 30. JIT compiler update.
414
415 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
416 Rill for the patch.
417
418 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
419
420 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
421
422 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
423 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
424 was enabled.
425
426 35. Improve JIT code generation for greedy plus quantifier.
427
428 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
429 1 to "aa" instead of to an empty string. The bug affected repeated groups
430 that could potentially match an empty string.
431
432 37. Optimizing single character iterators in JIT.
433
434 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
435 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
436 codepoints that are too big for the mode are faulted, and in a UTF mode,
437 disallowed codepoints are also faulted.
438
439 39. If PCRE was compiled with UTF support, in three places in the DFA
440 matcher there was code that should only have been obeyed in UTF mode, but
441 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
442 processing when bytes with values greater than 127 were present. In 16-bit
443 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
444 both cases the values are those that cannot be the first data item in a UTF
445 character. The three items that might have provoked this were recursions,
446 possessively repeated groups, and atomic groups.
447
448 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
449 and pcregrep, because some OS require shared objects to be explicitly
450 passed to ld, causing the link step to fail if they are not.
451
452 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
453 mode, patterns that started with \h* or \R* might be incorrectly matched.
454
455
456 Version 8.30 04-February-2012
457 -----------------------------
458
459 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
460 name is defined in ctype.h.
461
462 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
463 only in quite long subpatterns.
464
465 3. Removed the function pcre_info(), which has been obsolete and deprecated
466 since it was replaced by pcre_fullinfo() in February 2000.
467
468 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
469 match a (*MARK), and the match failed at the start of the subject, a
470 reference to memory before the start of the subject could occur. This bug
471 was introduced by fix 17 of release 8.21.
472
473 5. A reference to an unset group with zero minimum repetition was giving
474 totally wrong answers (in non-JavaScript-compatibility mode). For example,
475 /(another)?(\1?)test/ matched against "hello world test". This bug was
476 introduced in release 8.13.
477
478 6. Add support for 16-bit character strings (a large amount of work involving
479 many changes and refactorings).
480
481 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
482 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
483 from a file.
484
485 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
486 rounding is not applied in this particular case).
487
488 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
489 if they appear, or are escaped, in patterns.
490
491 10. Get rid of a number of -Wunused-but-set-variable warnings.
492
493 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
494 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
495 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
496 also returns the mark "x". This bug applied to capturing parentheses,
497 non-capturing parentheses, and atomic parentheses. It also applied to some
498 assertions.
499
500 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
501 information out of configure.ac instead of relying on pcre.h.generic, which
502 is not stored in the repository.
503
504 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
505 -lreadline.
506
507 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
508
509 15. Applied Graycode's patch to put the top-level frame on the stack rather
510 than the heap when not using the stack for recursion. This gives a
511 performance improvement in many cases when recursion is not deep.
512
513 16. Experimental code added to "pcretest -C" to output the stack frame size.
514
515
516 Version 8.21 12-Dec-2011
517 ------------------------
518
519 1. Updating the JIT compiler.
520
521 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
522 are added as well.
523
524 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
525 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
526 calling _pcre_jit_exec. Some extra comments are added.
527
528 4. (*MARK) settings inside atomic groups that do not contain any capturing
529 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
530 was introduced by change 18 for 8.20.
531
532 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
533 ECMA-262 standard.
534
535 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
536 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
537 This bug was probably introduced by change 9 of 8.13.
538
539 7. While fixing 6 above, I noticed that a number of other items were being
540 incorrectly rejected as "not fixed length". This arose partly because newer
541 opcodes had not been added to the fixed-length checking code. I have (a)
542 corrected the bug and added tests for these items, and (b) arranged for an
543 error to occur if an unknown opcode is encountered while checking for fixed
544 length instead of just assuming "not fixed length". The items that were
545 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
546 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
547 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
548
549 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
550 being incorrectly compiled and would have given unpredicatble results.
551
552 9. A possessively repeated subpattern with minimum repeat count greater than
553 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
554 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
555 the first (A) could occur when it should not.
556
557 10. Add a cast and remove a redundant test from the code.
558
559 11. JIT should use pcre_malloc/pcre_free for allocation.
560
561 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
562 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
563 is anything other than /usr, -L is still shown).
564
565 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
566
567 14. Perl does not support \N without a following name in a [] class; PCRE now
568 also gives an error.
569
570 15. If a forward reference was repeated with an upper limit of around 2000,
571 it caused the error "internal error: overran compiling workspace". The
572 maximum number of forward references (including repeats) was limited by the
573 internal workspace, and dependent on the LINK_SIZE. The code has been
574 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
575 the default depends on LINK_SIZE. There is a new upper limit (for safety)
576 of around 200,000 forward references. While doing this, I also speeded up
577 the filling in of repeated forward references.
578
579 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
580 incorrectly expecting the subject to contain another "a" after the start.
581
582 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
583 in the match, the SKIP should be ignored. This was not happening; instead
584 the SKIP was being treated as NOMATCH. For patterns such as
585 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
586 tested.
587
588 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
589 now much more compatible with Perl, in particular in cases where the result
590 is a non-match for a non-anchored pattern. For example, if
591 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
592 "m", where previously it did not return a name. A side effect of this
593 change is that for partial matches, the last encountered mark name is
594 returned, as for non matches. A number of tests that were previously not
595 Perl-compatible have been moved into the Perl-compatible test files. The
596 refactoring has had the pleasing side effect of removing one argument from
597 the match() function, thus reducing its stack requirements.
598
599 19. If the /S+ option was used in pcretest to study a pattern using JIT,
600 subsequent uses of /S (without +) incorrectly behaved like /S+.
601
602 21. Retrieve executable code size support for the JIT compiler and fixing
603 some warnings.
604
605 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
606 not work when the shorter character appeared right at the end of the
607 subject string.
608
609 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
610 systems.
611
612 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
613 output it when the /M option is used in pcretest.
614
615 25. The CheckMan script was not being included in the distribution. Also, added
616 an explicit "perl" to run Perl scripts from the PrepareRelease script
617 because this is reportedly needed in Windows.
618
619 26. If study data was being save in a file and studying had not found a set of
620 "starts with" bytes for the pattern, the data written to the file (though
621 never used) was taken from uninitialized memory and so caused valgrind to
622 complain.
623
624 27. Updated RunTest.bat as provided by Sheri Pierce.
625
626 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
627
628 29. Computation of memory usage for the table of capturing group names was
629 giving an unnecessarily large value.
630
631
632 Version 8.20 21-Oct-2011
633 ------------------------
634
635 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
636 a POSIX class. After further experiments with Perl, which convinced me that
637 Perl has bugs and confusions, a closing square bracket is no longer allowed
638 in a POSIX name. This bug also affected patterns with classes that started
639 with full stops.
640
641 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
642 captured substring, but while checking the failing first alternative,
643 substring 1 is temporarily captured. If the output vector supplied to
644 pcre_exec() was not big enough for this capture, the yield of the function
645 was still zero ("insufficient space for captured substrings"). This cannot
646 be totally fixed without adding another stack variable, which seems a lot
647 of expense for a edge case. However, I have improved the situation in cases
648 such as /(a)(b)x|abc/ matched against "abc", where the return code
649 indicates that fewer than the maximum number of slots in the ovector have
650 been set.
651
652 3. Related to (2) above: when there are more back references in a pattern than
653 slots in the output vector, pcre_exec() uses temporary memory during
654 matching, and copies in the captures as far as possible afterwards. It was
655 using the entire output vector, but this conflicts with the specification
656 that only 2/3 is used for passing back captured substrings. Now it uses
657 only the first 2/3, for compatibility. This is, of course, another edge
658 case.
659
660 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
661 main code base, and can be used by building with --enable-jit. When this is
662 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
663 runtime --no-jit option is given.
664
665 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
666 ovector, the return from the function was zero, implying that there were
667 other matches that did not fit. The correct "exactly full" value is now
668 returned.
669
670 6. If a subpattern that was called recursively or as a subroutine contained
671 (*PRUNE) or any other control that caused it to give a non-standard return,
672 invalid errors such as "Error -26 (nested recursion at the same subject
673 position)" or even infinite loops could occur.
674
675 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
676 computing the minimum length on reaching *ACCEPT, and so ended up with the
677 wrong value of 1 rather than 0. Further investigation indicates that
678 computing a minimum subject length in the presence of *ACCEPT is difficult
679 (think back references, subroutine calls), and so I have changed the code
680 so that no minimum is registered for a pattern that contains *ACCEPT.
681
682 8. If (*THEN) was present in the first (true) branch of a conditional group,
683 it was not handled as intended. [But see 16 below.]
684
685 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
686 Sheri Pierce.
687
688 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
689 the first byte in a match must be "a".
690
691 11. Change 17 for 8.13 increased the recursion depth for patterns like
692 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
693 pattern contains any instances of (*THEN). If it does not, the old
694 optimizations are restored. It would be nice to do this on a per-group
695 basis, but at the moment that is not feasible.
696
697 12. In some environments, the output of pcretest -C is CRLF terminated. This
698 broke RunTest's code that checks for the link size. A single white space
699 character after the value is now allowed for.
700
701 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
702 For "fr", it uses the Windows-specific input and output files.
703
704 14. If (*THEN) appeared in a group that was called recursively or as a
705 subroutine, it did not work as intended. [But see next item.]
706
707 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
708 pattern fragments (but not containing any | characters). If A and B are
709 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
710 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
711 D. In other words, Perl considers parentheses that do not contain any |
712 characters to be part of a surrounding alternative, whereas PCRE was
713 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
714 differently. PCRE now behaves in the same way as Perl, except in the case
715 of subroutine/recursion calls such as (?1) which have in any case always
716 been different (but PCRE had them first :-).
717
718 16. Related to 15 above: Perl does not treat the | in a conditional group as
719 creating alternatives. Such a group is treated in the same way as an
720 ordinary group without any | characters when processing (*THEN). PCRE has
721 been changed to match Perl's behaviour.
722
723 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
724 RunGrepTest script failed.
725
726 18. Change 22 for version 13 caused atomic groups to use more stack. This is
727 inevitable for groups that contain captures, but it can lead to a lot of
728 stack use in large patterns. The old behaviour has been restored for atomic
729 groups that do not contain any capturing parentheses.
730
731 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
732 suppress the check for a minimum subject length at run time. (If it was
733 given to pcre_exec() or pcre_dfa_exec() it did work.)
734
735 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
736 fail to work when decoding hex characters in data strings in EBCDIC
737 environments.
738
739 21. It appears that in at least one Mac OS environment, the isxdigit() function
740 is implemented as a macro that evaluates to its argument more than once,
741 contravening the C 90 Standard (I haven't checked a later standard). There
742 was an instance in pcretest which caused it to go wrong when processing
743 \x{...} escapes in subject strings. The has been rewritten to avoid using
744 things like p++ in the argument of isxdigit().
745
746
747 Version 8.13 16-Aug-2011
748 ------------------------
749
750 1. The Unicode data tables have been updated to Unicode 6.0.0.
751
752 2. Two minor typos in pcre_internal.h have been fixed.
753
754 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
755 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
756 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
757
758 4. There were a number of related bugs in the code for matching backrefences
759 caselessly in UTF-8 mode when codes for the characters concerned were
760 different numbers of bytes. For example, U+023A and U+2C65 are an upper
761 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
762 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
763 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
764 2-byte code at the end of the subject (it thought there wasn't enough data
765 left).
766
767 5. Comprehensive information about what went wrong is now returned by
768 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
769 as the output vector has at least 2 elements. The offset of the start of
770 the failing character and a reason code are placed in the vector.
771
772 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
773 now returned is for the first byte of the failing character, instead of the
774 last byte inspected. This is an incompatible change, but I hope it is small
775 enough not to be a problem. It makes the returned offset consistent with
776 pcre_exec() and pcre_dfa_exec().
777
778 7. pcretest now gives a text phrase as well as the error number when
779 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
780 failure, the offset and reason code are output.
781
782 8. When \R was used with a maximizing quantifier it failed to skip backwards
783 over a \r\n pair if the subsequent match failed. Instead, it just skipped
784 back over a single character (\n). This seems wrong (because it treated the
785 two characters as a single entity when going forwards), conflicts with the
786 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
787 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
788 has been changed.
789
790 9. Some internal refactoring has changed the processing so that the handling
791 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
792 time (the PCRE_DOTALL option was changed this way some time ago: version
793 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
794 which was always a bit of a fudge. It also means that there is one less
795 argument for the match() function, which reduces its stack requirements
796 slightly. This change also fixes an incompatibility with Perl: the pattern
797 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
798
799 10. More internal refactoring has drastically reduced the number of recursive
800 calls to match() for possessively repeated groups such as (abc)++ when
801 using pcre_exec().
802
803 11. While implementing 10, a number of bugs in the handling of groups were
804 discovered and fixed:
805
806 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
807 (a|)*(?1) gave a compile-time internal error.
808 ((a|)+)+ did not notice that the outer group could match an empty string.
809 (^a|^)+ was not marked as anchored.
810 (.*a|.*)+ was not marked as matching at start or after a newline.
811
812 12. Yet more internal refactoring has removed another argument from the match()
813 function. Special calls to this function are now indicated by setting a
814 value in a variable in the "match data" data block.
815
816 13. Be more explicit in pcre_study() instead of relying on "default" for
817 opcodes that mean there is no starting character; this means that when new
818 ones are added and accidentally left out of pcre_study(), testing should
819 pick them up.
820
821 14. The -s option of pcretest has been documented for ages as being an old
822 synonym of -m (show memory usage). I have changed it to mean "force study
823 for every regex", that is, assume /S for every regex. This is similar to -i
824 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
825 using it. It makes it easier to run collections of tests with and without
826 study enabled, and thereby test pcre_study() more easily. All the standard
827 tests are now run with and without -s (but some patterns can be marked as
828 "never study" - see 20 below).
829
830 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
831 restoration of the capturing data to the outer values was not happening
832 correctly.
833
834 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
835 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
836 pattern had matched an empty string, and so incorrectly returned a no
837 match.
838
839 17. There was optimizing code for the last branch of non-capturing parentheses,
840 and also for the obeyed branch of a conditional subexpression, which used
841 tail recursion to cut down on stack usage. Unfortunately, now that there is
842 the possibility of (*THEN) occurring in these branches, tail recursion is
843 no longer possible because the return has to be checked for (*THEN). These
844 two optimizations have therefore been removed. [But see 8.20/11 above.]
845
846 18. If a pattern containing \R was studied, it was assumed that \R always
847 matched two bytes, thus causing the minimum subject length to be
848 incorrectly computed because \R can also match just one byte.
849
850 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
851 was incorrectly computed.
852
853 20. If /S is present twice on a test pattern in pcretest input, it now
854 *disables* studying, thereby overriding the use of -s on the command line
855 (see 14 above). This is necessary for one or two tests to keep the output
856 identical in both cases.
857
858 21. When (*ACCEPT) was used in an assertion that matched an empty string and
859 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
860
861 22. When an atomic group that contained a capturing parenthesis was
862 successfully matched, but the branch in which it appeared failed, the
863 capturing was not being forgotten if a higher numbered group was later
864 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
865 group 1 to "a", when in fact it should be unset. This applied to multi-
866 branched capturing and non-capturing groups, repeated or not, and also to
867 positive assertions (capturing in negative assertions does not happen
868 in PCRE) and also to nested atomic groups.
869
870 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
871 subject after a captured substring, to make it easier to tell which of a
872 number of identical substrings has been captured.
873
874 24. The way atomic groups are processed by pcre_exec() has been changed so that
875 if they are repeated, backtracking one repetition now resets captured
876 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
877 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
878 "aaa". Previously, it would have been "a". As part of this code
879 refactoring, the way recursive calls are handled has also been changed.
880
881 25. If an assertion condition captured any substrings, they were not passed
882 back unless some other capturing happened later. For example, if
883 (?(?=(a))a) was matched against "a", no capturing was returned.
884
885 26. When studying a pattern that contained subroutine calls or assertions,
886 the code for finding the minimum length of a possible match was handling
887 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
888 group 1 called group 2 while simultaneously a separate group 2 called group
889 1). A stack overflow occurred in this case. I have fixed this by limiting
890 the recursion depth to 10.
891
892 27. Updated RunTest.bat in the distribution to the version supplied by Tom
893 Fortmann. This supports explicit test numbers on the command line, and has
894 argument validation and error reporting.
895
896 28. An instance of \X with an unlimited repeat could fail if at any point the
897 first character it looked at was a mark character.
898
899 29. Some minor code refactoring concerning Unicode properties and scripts
900 should reduce the stack requirement of match() slightly.
901
902 30. Added the '=' option to pcretest to check the setting of unused capturing
903 slots at the end of the pattern, which are documented as being -1, but are
904 not included in the return count.
905
906 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
907 compiled something random. Now it gives a compile-time error (as does
908 Perl).
909
910 32. A *MARK encountered during the processing of a positive assertion is now
911 recorded and passed back (compatible with Perl).
912
913 33. If --only-matching or --colour was set on a pcregrep call whose pattern
914 had alternative anchored branches, the search for a second match in a line
915 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
916 matched the line "0102" twice. The same bug affected patterns that started
917 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
918 twice.
919
920 34. Previously, PCRE did not allow quantification of assertions. However, Perl
921 does, and because of capturing effects, quantifying parenthesized
922 assertions may at times be useful. Quantifiers are now allowed for
923 parenthesized assertions.
924
925 35. A minor code tidy in pcre_compile() when checking options for \R usage.
926
927 36. \g was being checked for fancy things in a character class, when it should
928 just be a literal "g".
929
930 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
931 appearance of a nested POSIX class supersedes an apparent external class.
932 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
933 unescaped square brackets may also appear as part of class names. For
934 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
935 more like Perl. (But see 8.20/1 above.)
936
937 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
938 was because it thought it was \N{name}, which is not supported).
939
940 39. Add minix to OS list not supporting the -S option in pcretest.
941
942 40. PCRE tries to detect cases of infinite recursion at compile time, but it
943 cannot analyze patterns in sufficient detail to catch mutual recursions
944 such as ((?1))((?2)). There is now a runtime test that gives an error if a
945 subgroup is called recursively as a subpattern for a second time at the
946 same position in the subject string. In previous releases this might have
947 been caught by the recursion limit, or it might have run out of stack.
948
949 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
950 happen only once. PCRE was, however incorrectly giving a compile time error
951 "recursive call could loop indefinitely" because it cannot analyze the
952 pattern in sufficient detail. The compile time test no longer happens when
953 PCRE is compiling a conditional subpattern, but actual runaway loops are
954 now caught at runtime (see 40 above).
955
956 42. It seems that Perl allows any characters other than a closing parenthesis
957 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
958 has been changed to be the same.
959
960 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
961 as not to get warnings when autogen.sh is called. Also changed
962 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
963
964 44. To help people who use pcregrep to scan files containing exceedingly long
965 lines, the following changes have been made:
966
967 (a) The default value of the buffer size parameter has been increased from
968 8K to 20K. (The actual buffer used is three times this size.)
969
970 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
971 PCRE is built.
972
973 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
974 to be set at run time.
975
976 (d) Numerical values in pcregrep options can be followed by K or M, for
977 example --buffer-size=50K.
978
979 (e) If a line being scanned overflows pcregrep's buffer, an error is now
980 given and the return code is set to 2.
981
982 45. Add a pointer to the latest mark to the callout data block.
983
984 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
985 partial match of an empty string instead of no match. This was specific to
986 the use of ".".
987
988 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
989 complete match instead of a partial match. This bug was dependent on both
990 the PCRE_UTF8 and PCRE_DOTALL options being set.
991
992 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
993 starting byte set, because \b was not being ignored.
994
995
996 Version 8.12 15-Jan-2011
997 ------------------------
998
999 1. Fixed some typos in the markup of the man pages, and wrote a script that
1000 checks for such things as part of the documentation building process.
1001
1002 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1003 --match-limit and --recursion-limit options (added for 8.11). In
1004 particular, this made one of the standard tests fail. (The integer value
1005 went into the wrong half of a long int.)
1006
1007 3. If the --colour option was given to pcregrep with -v (invert match), it
1008 did strange things, either producing crazy output, or crashing. It should,
1009 of course, ignore a request for colour when reporting lines that do not
1010 match.
1011
1012 4. Another pcregrep bug caused similar problems if --colour was specified with
1013 -M (multiline) and the pattern match finished with a line ending.
1014
1015 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1016 matched in multiline mode, the following line was shown as part of the
1017 match. This seems wrong, so I have changed it.
1018
1019 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1020 the check for further matches in the same line (so they could be coloured)
1021 to overrun the end of the current line. If another match was found, it was
1022 incorrectly shown (and then shown again when found in the next line).
1023
1024 7. If pcregrep was compiled under Windows, there was a reference to the
1025 function pcregrep_exit() before it was defined. I am assuming this was
1026 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1027 reported by a user. I've moved the definition above the reference.
1028
1029
1030 Version 8.11 10-Dec-2010
1031 ------------------------
1032
1033 1. (*THEN) was not working properly if there were untried alternatives prior
1034 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1035 backtracked to try for "b" instead of moving to the next alternative branch
1036 at the same level (in this case, to look for "c"). The Perl documentation
1037 is clear that when (*THEN) is backtracked onto, it goes to the "next
1038 alternative in the innermost enclosing group".
1039
1040 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1041 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1042 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1043 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1044 (*THEN).
1045
1046 3. If \s appeared in a character class, it removed the VT character from
1047 the class, even if it had been included by some previous item, for example
1048 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1049 of \s, but is part of the POSIX "space" class.)
1050
1051 4. A partial match never returns an empty string (because you can always
1052 match an empty string at the end of the subject); however the checking for
1053 an empty string was starting at the "start of match" point. This has been
1054 changed to the "earliest inspected character" point, because the returned
1055 data for a partial match starts at this character. This means that, for
1056 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1057 (previously it gave "no match").
1058
1059 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1060 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1061 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1062 has an implication that the given string is incomplete (because a partial
1063 match is preferred over a full match). For this reason, these items now
1064 give a partial match in this situation. [Aside: previously, the one case
1065 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1066 match rather than a full match, which was wrong by the old rules, but is
1067 now correct.]
1068
1069 6. There was a bug in the handling of #-introduced comments, recognized when
1070 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1071 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1072 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1073 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1074 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1075 places in pcre_compile().
1076
1077 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1078 comments when looking ahead for named forward references to subpatterns,
1079 the only newline sequence it recognized was NL. It now handles newlines
1080 according to the set newline convention.
1081
1082 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1083 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1084 cater for a lack of strerror(). These oversights have been fixed.
1085
1086 9. Added --match-limit and --recursion-limit to pcregrep.
1087
1088 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1089
1090 11. When the -o option was used, pcregrep was setting a return code of 1, even
1091 when matches were found, and --line-buffered was not being honoured.
1092
1093 12. Added an optional parentheses number to the -o and --only-matching options
1094 of pcregrep.
1095
1096 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1097 can match an empty string. The code to do it in pcretest and pcredemo
1098 needed fixing:
1099
1100 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1101 only one byte after an empty string match just before CRLF (this case
1102 just got forgotten; "any" and "anycrlf" were OK).
1103
1104 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1105 mode when an empty string match preceded an ASCII character followed by
1106 a non-ASCII character. (The code for advancing by one character rather
1107 than one byte was nonsense.)
1108
1109 (c) The pcredemo.c sample program did not have any code at all to handle
1110 the cases when CRLF is a valid newline sequence.
1111
1112 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1113 as a starting offset was within the subject string. There is now a new
1114 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1115 negative or greater than the length of the string. In order to test this,
1116 pcretest is extended to allow the setting of negative starting offsets.
1117
1118 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1119 starting offset points to the beginning of a UTF-8 character was
1120 unnecessarily clumsy. I tidied it up.
1121
1122 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1123 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1124
1125 17. Nobody had reported that the --include_dir option, which was added in
1126 release 7.7 should have been called --include-dir (hyphen, not underscore)
1127 for compatibility with GNU grep. I have changed it to --include-dir, but
1128 left --include_dir as an undocumented synonym, and the same for
1129 --exclude-dir, though that is not available in GNU grep, at least as of
1130 release 2.5.4.
1131
1132 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1133 characters from a string of bytes) have been redefined so as not to use
1134 loops, in order to improve performance in some environments. At the same
1135 time, I abstracted some of the common code into auxiliary macros to save
1136 repetition (this should not affect the compiled code).
1137
1138 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1139 compile-time error is now given if \c is not followed by an ASCII
1140 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1141 different, and any byte value is allowed.)
1142
1143 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1144 START_OPTIMIZE option, which is now allowed at compile time - but just
1145 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1146 to pcregrep and other applications that have no direct access to PCRE
1147 options. The new /Y option in pcretest sets this option when calling
1148 pcre_compile().
1149
1150 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1151 back references. Groups containing recursive back references were forced to
1152 be atomic by that change, but in the case of named groups, the amount of
1153 memory required was incorrectly computed, leading to "Failed: internal
1154 error: code overflow". This has been fixed.
1155
1156 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1157 pcretest.c, to avoid build problems in some Borland environments.
1158
1159
1160 Version 8.10 25-Jun-2010
1161 ------------------------
1162
1163 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1164 THEN.
1165
1166 2. (*ACCEPT) was not working when inside an atomic group.
1167
1168 3. Inside a character class, \B is treated as a literal by default, but
1169 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1170 causes the error). The code is unchanged, but I tidied the documentation.
1171
1172 4. Inside a character class, PCRE always treated \R and \X as literals,
1173 whereas Perl faults them if its -w option is set. I have changed PCRE so
1174 that it faults them when PCRE_EXTRA is set.
1175
1176 5. Added support for \N, which always matches any character other than
1177 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1178
1179 6. When compiling pcregrep with newer versions of gcc which may have
1180 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1181 declared with attribute warn_unused_result" were given. Just casting the
1182 result to (void) does not stop the warnings; a more elaborate fudge is
1183 needed. I've used a macro to implement this.
1184
1185 7. Minor change to pcretest.c to avoid a compiler warning.
1186
1187 8. Added four artifical Unicode properties to help with an option to make
1188 \s etc use properties (see next item). The new properties are: Xan
1189 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1190
1191 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1192 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1193 this option. Modified pcretest to add /W to test this facility. Added
1194 REG_UCP to make it available via the POSIX interface.
1195
1196 10. Added --line-buffered to pcregrep.
1197
1198 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1199 studied, and the match started with a letter with a code point greater than
1200 127 whose first byte was different to the first byte of the other case of
1201 the letter, the other case of this starting letter was not recognized
1202 (#976).
1203
1204 12. If a pattern that was studied started with a repeated Unicode property
1205 test, for example, \p{Nd}+, there was the theoretical possibility of
1206 setting up an incorrect bitmap of starting bytes, but fortunately it could
1207 not have actually happened in practice until change 8 above was made (it
1208 added property types that matched character-matching opcodes).
1209
1210 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1211 possible starting bytes for non-anchored patterns.
1212
1213 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1214 \R, and also a number of cases that involve Unicode properties, both
1215 explicit and implicit when PCRE_UCP is set.
1216
1217 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1218 input, it could crash or give wrong results if characters with values
1219 greater than 0xc0 were present in the subject string. (Detail: it assumed
1220 UTF-8 input when processing these items.)
1221
1222 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1223 size_t is 64-bit (#991).
1224
1225 17. Added a check for running out of memory when PCRE is compiled with
1226 --disable-stack-for-recursion (#990).
1227
1228 18. If the last data line in a file for pcretest does not have a newline on
1229 the end, a newline was missing in the output.
1230
1231 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1232 less than 128) in its various bitmaps. However, there is a facility for
1233 generating tables according to the current locale when PCRE is compiled. It
1234 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1235 characters, are recognized by isspace() and therefore were getting set in
1236 these tables, and indeed these tables seem to approximate to ISO 8859. This
1237 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1238 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1239 which of course cannot start UTF-8 characters. I have changed the code so
1240 that only real ASCII characters (less than 128) and the correct starting
1241 bytes for UTF-8 encodings are set for characters greater than 127 when in
1242 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1243 altogether.)
1244
1245 20. Added the /T option to pcretest so as to be able to run tests with non-
1246 standard character tables, thus making it possible to include the tests
1247 used for 19 above in the standard set of tests.
1248
1249 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1250 reference to a subpattern the other side of a comment that contains an
1251 opening parenthesis caused either an internal compiling error, or a
1252 reference to the wrong subpattern.
1253
1254
1255 Version 8.02 19-Mar-2010
1256 ------------------------
1257
1258 1. The Unicode data tables have been updated to Unicode 5.2.0.
1259
1260 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1261 configured.
1262
1263 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1264 original author of that file, following a query about its status.
1265
1266 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1267 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1268
1269 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1270 quantifier applied to a forward-referencing subroutine call, could compile
1271 incorrect code or give the error "internal error: previously-checked
1272 referenced subpattern not found".
1273
1274 6. Both MS Visual Studio and Symbian OS have problems with initializing
1275 variables to point to external functions. For these systems, therefore,
1276 pcre_malloc etc. are now initialized to local functions that call the
1277 relevant global functions.
1278
1279 7. There were two entries missing in the vectors called coptable and poptable
1280 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1281 I've fixed the data, and added a kludgy way of testing at compile time that
1282 the lengths are correct (equal to the number of opcodes).
1283
1284 8. Following on from 7, I added a similar kludge to check the length of the
1285 eint vector in pcreposix.c.
1286
1287 9. Error texts for pcre_compile() are held as one long string to avoid too
1288 much relocation at load time. To find a text, the string is searched,
1289 counting zeros. There was no check for running off the end of the string,
1290 which could happen if a new error number was added without updating the
1291 string.
1292
1293 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1294
1295 11. \K was not working if it appeared in an atomic group or in a group that
1296 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1297 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1298 the assertion is positive, but not if it is negative.
1299
1300 12. Change 11 fortuitously reduced the size of the stack frame used in the
1301 "match()" function of pcre_exec.c by one pointer. Forthcoming
1302 implementation of support for (*MARK) will need an extra pointer on the
1303 stack; I have reserved it now, so that the stack frame size does not
1304 decrease.
1305
1306 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1307 item in branch that calls a recursion is a subroutine call - as in the
1308 second branch in the above example - was incorrectly given the compile-
1309 time error "recursive call could loop indefinitely" because pcre_compile()
1310 was not correctly checking the subroutine for matching a non-empty string.
1311
1312 14. The checks for overrunning compiling workspace could trigger after an
1313 overrun had occurred. This is a "should never occur" error, but it can be
1314 triggered by pathological patterns such as hundreds of nested parentheses.
1315 The checks now trigger 100 bytes before the end of the workspace.
1316
1317 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1318
1319
1320 Version 8.01 19-Jan-2010
1321 ------------------------
1322
1323 1. If a pattern contained a conditional subpattern with only one branch (in
1324 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1325 computed the wrong minimum data length (which is of course zero for such
1326 subpatterns). This could cause incorrect "no match" results.
1327
1328 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1329 the pattern is reset in the first branch, pcre_compile() failed with
1330 "internal error: code overflow at offset...". This happened only when
1331 the reset was to the original external option setting. (An optimization
1332 abstracts leading options settings into an external setting, which was the
1333 cause of this.)
1334
1335 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1336 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1337 assertion pattern did not match (meaning that the assertion was true), it
1338 was incorrectly treated as false if the SKIP had been reached during the
1339 matching. This also applied to assertions used as conditions.
1340
1341 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1342 assertion subpattern, including such a pattern used as a condition,
1343 unpredictable results occurred, instead of the error return
1344 PCRE_ERROR_DFA_UITEM.
1345
1346 5. The C++ GlobalReplace function was not working like Perl for the special
1347 situation when an empty string is matched. It now does the fancy magic
1348 stuff that is necessary.
1349
1350 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1351 removed. (These were left over from very, very early versions of PCRE.)
1352
1353 7. Some cosmetic changes to the code to make life easier when compiling it
1354 as part of something else:
1355
1356 (a) Change DEBUG to PCRE_DEBUG.
1357
1358 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1359 called "current" as "current_branch", to prevent a collision with the
1360 Linux macro when compiled as a kernel module.
1361
1362 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1363 prevent a collision with the Linux macro when compiled as a kernel
1364 module.
1365
1366 8. In pcre_compile() there are some checks for integer overflows that used to
1367 cast potentially large values to (double). This has been changed to that
1368 when building, a check for int64_t is made, and if it is found, it is used
1369 instead, thus avoiding the use of floating point arithmetic. (There is no
1370 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1371 double.
1372
1373 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1374 2005 (difference between two addresses compared to an unsigned value).
1375
1376 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1377 custom one, because of the following reported problem in Windows:
1378
1379 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1380 under Win32.
1381 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1382 therefore missing the function definition.
1383 - The compiler thus generates a "C" signature for the test function.
1384 - The linker fails to find the "C" function.
1385 - PCRE fails to configure if asked to do so against libbz2.
1386
1387 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1388 messages were output:
1389
1390 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1391 rerunning libtoolize, to keep the correct libtool macros in-tree.
1392 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1393
1394 I have done both of these things.
1395
1396 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1397 most of the time, it *can* run out if it is given a pattern that contains a
1398 runaway infinite recursion. I updated the discussion in the pcrestack man
1399 page.
1400
1401 13. Now that we have gone to the x.xx style of version numbers, the minor
1402 version may start with zero. Using 08 or 09 is a bad idea because users
1403 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1404 interpreted as invalid octal numbers. I've updated the previous comment in
1405 configure.ac, and also added a check that gives an error if 08 or 09 are
1406 used.
1407
1408 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1409 causing partial matching to fail when the end of the subject matched \W
1410 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1411
1412 15. There were some discrepancies between the declarations in pcre_internal.h
1413 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1414 their definitions. The declarations used "const uschar *" and the
1415 definitions used USPTR. Even though USPTR is normally defined as "const
1416 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1417 reported that: "This difference in casting confuses some C++ compilers, for
1418 example, SunCC recognizes above declarations as different functions and
1419 generates broken code for hbpcre." I have changed the declarations to use
1420 USPTR.
1421
1422 16. GNU libtool is named differently on some systems. The autogen.sh script now
1423 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1424 (FreeBSD).
1425
1426 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1427 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1428 comment: "Figure out how to create a longlong from a string: strtoll and
1429 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1430 instance, but it only takes 2 args instead of 3!"
1431
1432 18. A subtle bug concerned with back references has been fixed by a change of
1433 specification, with a corresponding code fix. A pattern such as
1434 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1435 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1436 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1437 same bug. Such groups have to be quantified to be useful, or contained
1438 inside another quantified group. (If there's no repetition, the reference
1439 can never match.) The problem arises because, having left the group and
1440 moved on to the rest of the pattern, a later failure that backtracks into
1441 the group uses the captured value from the final iteration of the group
1442 rather than the correct earlier one. I have fixed this in PCRE by forcing
1443 any group that contains a reference to itself to be an atomic group; that
1444 is, there cannot be any backtracking into it once it has completed. This is
1445 similar to recursive and subroutine calls.
1446
1447
1448 Version 8.00 19-Oct-09
1449 ----------------------
1450
1451 1. The table for translating pcre_compile() error codes into POSIX error codes
1452 was out-of-date, and there was no check on the pcre_compile() error code
1453 being within the table. This could lead to an OK return being given in
1454 error.
1455
1456 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1457 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1458 in a Windows environment.
1459
1460 3. The pcregrep --count option prints the count for each file even when it is
1461 zero, as does GNU grep. However, pcregrep was also printing all files when
1462 --files-with-matches was added. Now, when both options are given, it prints
1463 counts only for those files that have at least one match. (GNU grep just
1464 prints the file name in this circumstance, but including the count seems
1465 more useful - otherwise, why use --count?) Also ensured that the
1466 combination -clh just lists non-zero counts, with no names.
1467
1468 4. The long form of the pcregrep -F option was incorrectly implemented as
1469 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1470 but it seems right to fix it, and I didn't think it was worth preserving
1471 the old behaviour.
1472
1473 5. The command line items --regex=pattern and --regexp=pattern were not
1474 recognized by pcregrep, which required --regex pattern or --regexp pattern
1475 (with a space rather than an '='). The man page documented the '=' forms,
1476 which are compatible with GNU grep; these now work.
1477
1478 6. No libpcreposix.pc file was created for pkg-config; there was just
1479 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1480
1481 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1482 when UCP support is not needed, by modifying the Python script that
1483 generates it from Unicode data files. This should not matter if the module
1484 is correctly used as a library, but I received one complaint about 50K of
1485 unwanted data. My guess is that the person linked everything into his
1486 program rather than using a library. Anyway, it does no harm.
1487
1488 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1489 was a minimum greater than 1 for a wide character in a possessive
1490 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1491 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1492 character. Chaos in the form of incorrect output or a compiling loop could
1493 result.
1494
1495 9. The restrictions on what a pattern can contain when partial matching is
1496 requested for pcre_exec() have been removed. All patterns can now be
1497 partially matched by this function. In addition, if there are at least two
1498 slots in the offset vector, the offset of the earliest inspected character
1499 for the match and the offset of the end of the subject are set in them when
1500 PCRE_ERROR_PARTIAL is returned.
1501
1502 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1503 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1504 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1505 and may be more useful for multi-segment matching.
1506
1507 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1508 used to be given if ever the end of the subject was reached; now it is
1509 given only if matching could not proceed because another character was
1510 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1511 string "Z", which now yields "no match" instead of "partial match". In the
1512 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1513 final character ended with (*FAIL).
1514
1515 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1516 if the pattern had a "must contain" character that was already found in the
1517 earlier partial match, unless partial matching was again requested. For
1518 example, with the pattern /dog.(body)?/, the "must contain" character is
1519 "g". If the first part-match was for the string "dog", restarting with
1520 "sbody" failed. This bug has been fixed.
1521
1522 13. The string returned by pcre_dfa_exec() after a partial match has been
1523 changed so that it starts at the first inspected character rather than the
1524 first character of the match. This makes a difference only if the pattern
1525 starts with a lookbehind assertion or \b or \B (\K is not supported by
1526 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1527 matching functions compatible, and I think it's the right thing to do.
1528
1529 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1530 so that the demonstration program is easily available in environments where
1531 PCRE has not been installed from source.
1532
1533 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1534 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1535 library.
1536
1537 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1538 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1539 is not the first non-POSIX option to be added. Clearly some people find
1540 these options useful.
1541
1542 17. If a caller to the POSIX matching function regexec() passes a non-zero
1543 value for nmatch with a NULL value for pmatch, the value of
1544 nmatch is forced to zero.
1545
1546 18. RunGrepTest did not have a test for the availability of the -u option of
1547 the diff command, as RunTest does. It now checks in the same way as
1548 RunTest, and also checks for the -b option.
1549
1550 19. If an odd number of negated classes containing just a single character
1551 interposed, within parentheses, between a forward reference to a named
1552 subpattern and the definition of the subpattern, compilation crashed with
1553 an internal error, complaining that it could not find the referenced
1554 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1555 [The bug was that it was starting one character too far in when skipping
1556 over the character class, thus treating the ] as data rather than
1557 terminating the class. This meant it could skip too much.]
1558
1559 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1560 /g option in pcretest when the pattern contains \K, which makes it possible
1561 to have an empty string match not at the start, even when the pattern is
1562 anchored. Updated pcretest and pcredemo to use this option.
1563
1564 21. If the maximum number of capturing subpatterns in a recursion was greater
1565 than the maximum at the outer level, the higher number was returned, but
1566 with unset values at the outer level. The correct (outer level) value is
1567 now given.
1568
1569 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1570 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1571 make it do so. The string so far is captured, making this feature
1572 compatible with Perl.
1573
1574 23. The tests have been re-organized, adding tests 11 and 12, to make it
1575 possible to check the Perl 5.10 features against Perl 5.10.
1576
1577 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1578 pattern matches a fixed length string. PCRE did not allow this; now it
1579 does. Neither allows recursion.
1580
1581 25. I finally figured out how to implement a request to provide the minimum
1582 length of subject string that was needed in order to match a given pattern.
1583 (It was back references and recursion that I had previously got hung up
1584 on.) This code has now been added to pcre_study(); it finds a lower bound
1585 to the length of subject needed. It is not necessarily the greatest lower
1586 bound, but using it to avoid searching strings that are too short does give
1587 some useful speed-ups. The value is available to calling programs via
1588 pcre_fullinfo().
1589
1590 26. While implementing 25, I discovered to my embarrassment that pcretest had
1591 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1592 study optimizations had never been tested with that matching function.
1593 Oops. What is worse, even when it was passed study data, there was a bug in
1594 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1595 were also very few tests of studied patterns with pcre_dfa_exec().
1596
1597 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1598 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1599 on the other side of the coin, they are no longer allowed to have different
1600 names, because these cannot be distinguished in PCRE, and this has caused
1601 confusion. (This is a difference from Perl.)
1602
1603 28. When duplicate subpattern names are present (necessarily with different
1604 numbers, as required by 27 above), and a test is made by name in a
1605 conditional pattern, either for a subpattern having been matched, or for
1606 recursion in such a pattern, all the associated numbered subpatterns are
1607 tested, and the overall condition is true if the condition is true for any
1608 one of them. This is the way Perl works, and is also more like the way
1609 testing by number works.
1610
1611
1612 Version 7.9 11-Apr-09
1613 ---------------------
1614
1615 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1616 (pcretest), all targets were linked against these libraries. This included
1617 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1618 libraries. This caused unwanted dependencies to be created. This problem
1619 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1620 pcretest is linked with readline.
1621
1622 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1623 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1624 moved outside it again, because FALSE and TRUE are already defined in AIX,
1625 but BOOL is not.
1626
1627 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1628 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1629
1630 4. The pcregrep documentation said spaces were inserted as well as colons (or
1631 hyphens) following file names and line numbers when outputting matching
1632 lines. This is not true; no spaces are inserted. I have also clarified the
1633 wording for the --colour (or --color) option.
1634
1635 5. In pcregrep, when --colour was used with -o, the list of matching strings
1636 was not coloured; this is different to GNU grep, so I have changed it to be
1637 the same.
1638
1639 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1640 each matching line was coloured. Now it goes on to look for further matches
1641 of any of the test patterns, which is the same behaviour as GNU grep.
1642
1643 7. A pattern that could match an empty string could cause pcregrep to loop; it
1644 doesn't make sense to accept an empty string match in pcregrep, so I have
1645 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1646 seems to be how GNU grep behaves.
1647
1648 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1649 start or after a newline", because the conditional assertion was not being
1650 correctly handled. The rule now is that both the assertion and what follows
1651 in the first alternative must satisfy the test.
1652
1653 9. If auto-callout was enabled in a pattern with a conditional group whose
1654 condition was an assertion, PCRE could crash during matching, both with
1655 pcre_exec() and pcre_dfa_exec().
1656
1657 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1658 used for matching.
1659
1660 11. Unicode property support in character classes was not working for
1661 characters (bytes) greater than 127 when not in UTF-8 mode.
1662
1663 12. Added the -M command line option to pcretest.
1664
1665 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1666
1667 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1668
1669 16. Added comments and documentation about mis-use of no_arg in the C++
1670 wrapper.
1671
1672 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1673 from Martin Jerabek that uses macro names for all relevant character and
1674 string constants.
1675
1676 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1677 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1678 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1679 these, but not everybody uses configure.
1680
1681 19. A conditional group that had only one branch was not being correctly
1682 recognized as an item that could match an empty string. This meant that an
1683 enclosing group might also not be so recognized, causing infinite looping
1684 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1685 with the subject "ab", where knowledge that the repeated group can match
1686 nothing is needed in order to break the loop.
1687
1688 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1689 exec(), but without supplying a callout function, matching went wrong.
1690
1691 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1692 leak if the size of the offset vector was greater than 30. When the vector
1693 is smaller, the saved offsets during recursion go onto a local stack
1694 vector, but for larger vectors malloc() is used. It was failing to free
1695 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1696 error, in fact).
1697
1698 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1699 heapframe that is used only when UTF-8 support is enabled. This caused no
1700 problem, but was untidy.
1701
1702 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1703 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1704 included within another project.
1705
1706 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1707 slightly modified by me:
1708
1709 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1710 not building pcregrep.
1711
1712 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1713 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1714
1715 25. Forward references, both numeric and by name, in patterns that made use of
1716 duplicate group numbers, could behave incorrectly or give incorrect errors,
1717 because when scanning forward to find the reference group, PCRE was not
1718 taking into account the duplicate group numbers. A pattern such as
1719 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1720
1721 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1722 the feature of a custom pointer more persuasive (as requested by a user).
1723
1724 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1725 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1726 pre-defined.
1727
1728 28. Added support for (*UTF8) at the start of a pattern.
1729
1730 29. Arrange for flags added by the "release type" setting in CMake to be shown
1731 in the configuration summary.
1732
1733
1734 Version 7.8 05-Sep-08
1735 ---------------------
1736
1737 1. Replaced UCP searching code with optimized version as implemented for Ad
1738 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1739 stage table and inline lookup instead of a function, giving speed ups of 2
1740 to 5 times on some simple patterns that I tested. Permission was given to
1741 distribute the MultiStage2.py script that generates the tables (it's not in
1742 the tarball, but is in the Subversion repository).
1743
1744 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1745 scripts.
1746
1747 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1748 a group with a zero qualifier. The result of the study could be incorrect,
1749 or the function might crash, depending on the pattern.
1750
1751 4. Caseless matching was not working for non-ASCII characters in back
1752 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1753 It now works when Unicode Property Support is available.
1754
1755 5. In pcretest, an escape such as \x{de} in the data was always generating
1756 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1757 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1758 truncation.
1759
1760 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1761
1762 7. Added two (int) casts to pcregrep when printing the difference of two
1763 pointers, in case they are 64-bit values.
1764
1765 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1766 test 2 if it fails.
1767
1768 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1769 and a #define of that name to empty if it is not externally set. This is to
1770 allow users of MSVC to set it if necessary.
1771
1772 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1773 the convenience functions in the pcre_get.c source file.
1774
1775 11. An option change at the start of a pattern that had top-level alternatives
1776 could cause overwriting and/or a crash. This command provoked a crash in
1777 some environments:
1778
1779 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1780
1781 This potential security problem was recorded as CVE-2008-2371.
1782
1783 12. For a pattern where the match had to start at the beginning or immediately
1784 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1785 pcre_dfa_exec() could read past the end of the passed subject if there was
1786 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1787 pcretest so that it places the subject at the end of its malloc-ed buffer.
1788
1789 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1790 exec() might read past the end of the data buffer in UTF-8 mode.
1791
1792 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1793 the data contained the byte 0x85 as part of a UTF-8 character within its
1794 first line. This applied both to normal and DFA matching.
1795
1796 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1797 /^[^d]*?$/8 failed to match "abc".
1798
1799 16. Added a missing copyright notice to pcrecpp_internal.h.
1800
1801 17. Make it more clear in the documentation that values returned from
1802 pcre_exec() in ovector are byte offsets, not character counts.
1803
1804 18. Tidied a few places to stop certain compilers from issuing warnings.
1805
1806 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1807 supplied by Stefan Weber. I made a further small update for 7.8 because
1808 there is a change of source arrangements: the pcre_searchfuncs.c module is
1809 replaced by pcre_ucd.c.
1810
1811
1812 Version 7.7 07-May-08
1813 ---------------------
1814
1815 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1816 a string to a long long, pretend we don't even have a long long." This is
1817 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1818
1819 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1820 pre-7.6 versions, which defined a global no_arg variable instead of putting
1821 it in the RE class. (See also #8 below.)
1822
1823 3. Remove a line of dead code, identified by coverity and reported by Nuno
1824 Lopes.
1825
1826 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1827
1828 (1) The include/exclude patterns were being applied to the whole pathnames
1829 of files, instead of just to the final components.
1830
1831 (2) If there was more than one level of directory, the subdirectories were
1832 skipped unless they satisfied the include/exclude conditions. This is
1833 inconsistent with GNU grep (and could even be seen as contrary to the
1834 pcregrep specification - which I improved to make it absolutely clear).
1835 The action now is always to scan all levels of directory, and just
1836 apply the include/exclude patterns to regular files.
1837
1838 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1839 --exclude_dir in the tests to avoid scanning .svn directories.
1840
1841 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1842 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1843 doesn't support NULs in patterns.
1844
1845 7. Added some missing "const"s to declarations of static tables in
1846 pcre_compile.c and pcre_dfa_exec.c.
1847
1848 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1849 caused by fix #2 above. (Subsequently also a second patch to fix the
1850 first patch. And a third patch - this was a messy problem.)
1851
1852 9. Applied Craig's patch to remove the use of push_back().
1853
1854 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1855 matching function regexec().
1856
1857 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1858 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1859 references. PCRE supports relative numbers with this syntax (I don't think
1860 Oniguruma does).
1861
1862 12. Previously, a group with a zero repeat such as (...){0} was completely
1863 omitted from the compiled regex. However, this means that if the group
1864 was called as a subroutine from elsewhere in the pattern, things went wrong
1865 (an internal error was given). Such groups are now left in the compiled
1866 pattern, with a new opcode that causes them to be skipped at execution
1867 time.
1868
1869 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1870 to the way PCRE behaves:
1871
1872 (a) A lone ] character is dis-allowed (Perl treats it as data).
1873
1874 (b) A back reference to an unmatched subpattern matches an empty string
1875 (Perl fails the current match path).
1876
1877 (c) A data ] in a character class must be notated as \] because if the
1878 first data character in a class is ], it defines an empty class. (In
1879 Perl it is not possible to have an empty class.) The empty class []
1880 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1881 The negative empty class [^] matches any one character, independently
1882 of the DOTALL setting.
1883
1884 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1885 non-existent subpattern following a character class starting with ']' and
1886 containing () gave an internal compiling error instead of "reference to
1887 non-existent subpattern". Fortunately, when the pattern did exist, the
1888 compiled code was correct. (When scanning forwards to check for the
1889 existencd of the subpattern, it was treating the data ']' as terminating
1890 the class, so got the count wrong. When actually compiling, the reference
1891 was subsequently set up correctly.)
1892
1893 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1894 it was being rejected as not supported by pcre_dfa_exec(), even though
1895 other assertions are supported. I have made pcre_dfa_exec() support
1896 (*FAIL).
1897
1898 16. The implementation of 13c above involved the invention of a new opcode,
1899 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1900 cannot be changed at match time, I realized I could make a small
1901 improvement to matching performance by compiling OP_ALLANY instead of
1902 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1903 on the OP_ANY path.
1904
1905 17. Compiling pcretest on Windows with readline support failed without the
1906 following two fixes: (1) Make the unistd.h include conditional on
1907 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1908
1909 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1910 ncurses library to be included for pcretest when ReadLine support is
1911 requested, but also to allow for it to be overridden. This patch came from
1912 Daniel Bergström.
1913
1914 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1915 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1916 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1917 spotting this.
1918
1919
1920 Version 7.6 28-Jan-08
1921 ---------------------
1922
1923 1. A character class containing a very large number of characters with
1924 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1925 overflow.
1926
1927 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1928 HAVE_LONG_LONG is not defined.
1929
1930 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1931 bring them up to date and include new features. This patch includes:
1932
1933 - Fixed PH's badly added libz and libbz2 support.
1934 - Fixed a problem with static linking.
1935 - Added pcredemo. [But later removed - see 7 below.]
1936 - Fixed dftables problem and added an option.
1937 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1938 HAVE_LONG_LONG.
1939 - Added readline support for pcretest.
1940 - Added an listing of the option settings after cmake has run.
1941
1942 4. A user submitted a patch to Makefile that makes it easy to create
1943 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1944 Makefile.am that cause it to include this special target, without
1945 affecting anything else. Note that the same mingw target plus all
1946 the other distribution libraries and programs are now supported
1947 when configuring with CMake (see 6 below) instead of with
1948 Configure/Make.
1949
1950 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1951 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1952 exported in the Windows port". It has not yet been confirmed that the patch
1953 solves the problem, but it does no harm.
1954
1955 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1956 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1957 with CMake, and also correct the comment about stack recursion.
1958
1959 7. Remove the automatic building of pcredemo from the ./configure system and
1960 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1961 of a program that users should build themselves after PCRE is installed, so
1962 building it automatically is not really right. What is more, it gave
1963 trouble in some build environments.
1964
1965 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1966
1967
1968 Version 7.5 10-Jan-08
1969 ---------------------
1970
1971 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1972 values in parens when parsing an RE using the C++ wrapper."
1973
1974 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1975 Characters greater than 255 were excluded from the class instead of being
1976 included.
1977
1978 3. The same bug as (2) above applied to negated POSIX classes such as
1979 [:^space:].
1980
1981 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1982 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1983 I have changed it.
1984
1985 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1986 first named subpattern) and a construct such as (?&a) would reference the
1987 first named subpattern whose name started with "a" (in other words, the
1988 length check was missing). Both these problems are fixed. "Subpattern name
1989 expected" is now given for (?&) (a zero-length name), and this patch also
1990 makes it give the same error for \k'' (previously it complained that that
1991 was a reference to a non-existent subpattern).
1992
1993 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1994 this is right because (?- can be followed by option settings as well as by
1995 digits. I have, however, made the messages clearer.
1996
1997 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1998 than the number used in the conditional) now cause a compile-time error.
1999 This is actually not compatible with Perl, which accepts such patterns, but
2000 treats the conditional as always being FALSE (as PCRE used to), but it
2001 seems to me that giving a diagnostic is better.
2002
2003 8. Change "alphameric" to the more common word "alphanumeric" in comments
2004 and messages.
2005
2006 9. Fix two occurrences of "backslash" in comments that should have been
2007 "backspace".
2008
2009 10. Remove two redundant lines of code that can never be obeyed (their function
2010 was moved elsewhere).
2011
2012 11. The program that makes PCRE's Unicode character property table had a bug
2013 which caused it to generate incorrect table entries for sequences of
2014 characters that have the same character type, but are in different scripts.
2015 It amalgamated them into a single range, with the script of the first of
2016 them. In other words, some characters were in the wrong script. There were
2017 thirteen such cases, affecting characters in the following ranges:
2018
2019 U+002b0 - U+002c1
2020 U+0060c - U+0060d
2021 U+0061e - U+00612
2022 U+0064b - U+0065e
2023 U+0074d - U+0076d
2024 U+01800 - U+01805
2025 U+01d00 - U+01d77
2026 U+01d9b - U+01dbf
2027 U+0200b - U+0200f
2028 U+030fc - U+030fe
2029 U+03260 - U+0327f
2030 U+0fb46 - U+0fbb1
2031 U+10450 - U+1049d
2032
2033 12. The -o option (show only the matching part of a line) for pcregrep was not
2034 compatible with GNU grep in that, if there was more than one match in a
2035 line, it showed only the first of them. It now behaves in the same way as
2036 GNU grep.
2037
2038 13. If the -o and -v options were combined for pcregrep, it printed a blank
2039 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2040 does the same. The return code can be used to tell if there were any
2041 non-matching lines.
2042
2043 14. Added --file-offsets and --line-offsets to pcregrep.
2044
2045 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2046 infinitely looping recursion. The bug was that positive lookaheads were not
2047 being skipped when checking for a possible empty match (negative lookaheads
2048 and both kinds of lookbehind were skipped).
2049
2050 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2051 inclusion of <windows.h> to before rather than after the definition of
2052 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2053
2054 17. Specifying a possessive quantifier with a specific limit for a Unicode
2055 character property caused pcre_compile() to compile bad code, which led at
2056 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2057 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2058 caused the error; without that there was no problem.
2059
2060 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2061
2062 19. Added --enable-pcretest-libreadline.
2063
2064 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2065 RE::GlobalReplace(). As a result, the number of replacements returned was
2066 double what it should be. I removed one of the increments, but Craig sent a
2067 later patch that removed the other one (the right fix) and added unit tests
2068 that check the return values (which was not done before).
2069
2070 21. Several CMake things:
2071
2072 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2073 the names libpcre and libpcreposix, not just pcre and pcreposix.
2074
2075 (2) The above change means that pcretest and pcregrep are now correctly
2076 linked with the newly-built libraries, not previously installed ones.
2077
2078 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2079
2080 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2081 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2082 UTF-8 newline character). The key issue is that the pattern starts .*;
2083 this means that the match must be either at the beginning, or after a
2084 newline. The bug was in the code for advancing after a failed match and
2085 checking that the new position followed a newline. It was not taking
2086 account of UTF-8 characters correctly.
2087
2088 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2089 character classes. PCRE was not treating the sequence [:...:] as a
2090 character class unless the ... were all letters. Perl, however, seems to
2091 allow any characters between [: and :], though of course it rejects as
2092 unknown any "names" that contain non-letters, because all the known class
2093 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2094 for example, whereas PCRE did not - it did not recognize a POSIX character
2095 class. This seemed a bit dangerous, so the code has been changed to be
2096 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2097 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2098 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2099 Perl does, and where it didn't before.
2100
2101 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2102 Windows environments %n is disabled by default.
2103
2104
2105 Version 7.4 21-Sep-07
2106 ---------------------
2107
2108 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2109 means that a class such as [\s] counted as "explicit reference to CR or
2110 LF". That isn't really right - the whole point of the change was to try to
2111 help when there was an actual mention of one of the two characters. So now
2112 the change happens only if \r or \n (or a literal CR or LF) character is
2113 encountered.
2114
2115 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2116 of both had grown to the point where there were only 3 bits left.
2117 Fortunately, there was spare space in the data structure, and so I have
2118 moved the internal flags into a new 16-bit field to free up more option
2119 bits.
2120
2121 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2122 but did not set the internal JCHANGED flag - either of these is enough to
2123 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2124 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2125 start sets both bits.
2126
2127 4. Added options (at build time, compile time, exec time) to change \R from
2128 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2129
2130 5. doc/pcresyntax.html was missing from the distribution.
2131
2132 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2133 compatibility, even though it is no longer used.
2134
2135 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2136 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2137 windows.h file is present (where different names are used). [This was
2138 reversed later after testing - see 16 below.]
2139
2140 8. Changed all #include <config.h> to #include "config.h". There were also
2141 some further <pcre.h> cases that I changed to "pcre.h".
2142
2143 9. When pcregrep was used with the --colour option, it missed the line ending
2144 sequence off the lines that it output.
2145
2146 10. It was pointed out to me that arrays of string pointers cause lots of
2147 relocations when a shared library is dynamically loaded. A technique of
2148 using a single long string with a table of offsets can drastically reduce
2149 these. I have refactored PCRE in four places to do this. The result is
2150 dramatic:
2151
2152 Originally: 290
2153 After changing UCP table: 187
2154 After changing error message table: 43
2155 After changing table of "verbs" 36
2156 After changing table of Posix names 22
2157
2158 Thanks to the folks working on Gregex for glib for this insight.
2159
2160 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2161 unicode-properties was also set.
2162
2163 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2164
2165 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2166 checked only for CRLF.
2167
2168 14. Added casts to pcretest.c to avoid compiler warnings.
2169
2170 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2171
2172 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2173 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2174 entirely. This removes changes made in 7 above.
2175
2176 17. The CMake files have been updated, and there is now more information about
2177 building with CMake in the NON-UNIX-USE document.
2178
2179
2180 Version 7.3 28-Aug-07
2181 ---------------------
2182
2183 1. In the rejigging of the build system that eventually resulted in 7.1, the
2184 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2185 brackets there is not right, since it causes compilers to look for an
2186 installed pcre.h, not the version that is in the source that is being
2187 compiled (which of course may be different). I have changed it back to:
2188
2189 #include "pcre.h"
2190
2191 I have a vague recollection that the change was concerned with compiling in
2192 different directories, but in the new build system, that is taken care of
2193 by the VPATH setting the Makefile.
2194
2195 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2196 when the subject happened to end in the byte 0x85 (e.g. if the last
2197 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2198 characters but of course it shouldn't be taken as a newline when it is part
2199 of another character. The bug was that, for an unlimited repeat of . in
2200 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2201 characters when looking for a newline.
2202
2203 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2204
2205 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2206 in debug output.
2207
2208 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2209 long printing in the pcrecpp unittest when running under MinGW.
2210
2211 6. ESC_K was left out of the EBCDIC table.
2212
2213 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2214 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2215 limit also applies to "virtual nesting" when a pattern is recursive, and in
2216 this case 1000 isn't so big. I have been able to remove this limit at the
2217 expense of backing off one optimization in certain circumstances. Normally,
2218 when pcre_exec() would call its internal match() function recursively and
2219 immediately return the result unconditionally, it uses a "tail recursion"
2220 feature to save stack. However, when a subpattern that can match an empty
2221 string has an unlimited repetition quantifier, it no longer makes this
2222 optimization. That gives it a stack frame in which to save the data for
2223 checking that an empty string has been matched. Previously this was taken
2224 from the 1000-entry workspace that had been reserved. So now there is no
2225 explicit limit, but more stack is used.
2226
2227 8. Applied Daniel's patches to solve problems with the import/export magic
2228 syntax that is required for Windows, and which was going wrong for the
2229 pcreposix and pcrecpp parts of the library. These were overlooked when this
2230 problem was solved for the main library.
2231
2232 9. There were some crude static tests to avoid integer overflow when computing
2233 the size of patterns that contain repeated groups with explicit upper
2234 limits. As the maximum quantifier is 65535, the maximum group length was
2235 set at 30,000 so that the product of these two numbers did not overflow a
2236 32-bit integer. However, it turns out that people want to use groups that
2237 are longer than 30,000 bytes (though not repeat them that many times).
2238 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2239 made it possible to implement the integer overflow checks in a much more
2240 dynamic way, which I have now done. The artificial limitation on group
2241 length has been removed - we now have only the limit on the total length of
2242 the compiled pattern, which depends on the LINK_SIZE setting.
2243
2244 10. Fixed a bug in the documentation for get/copy named substring when
2245 duplicate names are permitted. If none of the named substrings are set, the
2246 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2247 empty string.
2248
2249 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2250 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2251 because the ] is interpreted as the first data character and the
2252 terminating ] is not found. PCRE has been made compatible with Perl in this
2253 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2254 cause memory overwriting.
2255
2256 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2257 string has been matched (to stop an infinite loop). It was not recognizing
2258 a conditional subpattern that could match an empty string if that
2259 subpattern was within another subpattern. For example, it looped when
2260 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2261 condition was not nested. This bug has been fixed.
2262
2263 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2264 past the start of the subject in the presence of bytes with the top bit
2265 set, for example "\x8aBCD".
2266
2267 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2268 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2269
2270 14. Optimized (?!) to (*FAIL).
2271
2272 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2273 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2274 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2275 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2276 does: it's just the validity check that is more restrictive.
2277
2278 16. Inserted checks for integer overflows during escape sequence (backslash)
2279 processing, and also fixed erroneous offset values for syntax errors during
2280 backslash processing.
2281
2282 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2283 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2284
2285 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2286 caused an overrun.
2287
2288 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2289 something other than just ASCII characters) inside a group that had an
2290 unlimited repeat caused a loop at compile time (while checking to see
2291 whether the group could match an empty string).
2292
2293 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2294 [\P{Any}] did so. (Error in the code for printing property names.)
2295
2296 21. An orphan \E inside a character class could cause a crash.
2297
2298 22. A repeated capturing bracket such as (A)? could cause a wild memory
2299 reference during compilation.
2300
2301 23. There are several functions in pcre_compile() that scan along a compiled
2302 expression for various reasons (e.g. to see if it's fixed length for look
2303 behind). There were bugs in these functions when a repeated \p or \P was
2304 present in the pattern. These operators have additional parameters compared
2305 with \d, etc, and these were not being taken into account when moving along
2306 the compiled data. Specifically:
2307
2308 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2309 length.
2310
2311 (b) An item such as \pL+ within a repeated group could cause crashes or
2312 loops.
2313
2314 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2315 "reference to non-existent subpattern" error.
2316
2317 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2318
2319 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2320 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2321
2322 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2323
2324 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2325 character were causing crashes (broken optimization).
2326
2327 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2328 \p or \P) caused a compile-time loop.
2329
2330 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2331 break. For example, the unstudied pattern [\r\n]A does not match the string
2332 "\r\nA" because change 7.0/46 below moves the current point on by two
2333 characters after failing to match at the start. However, the pattern \nA
2334 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2335 the same is true. There doesn't seem any very clean way out of this, but
2336 what I have chosen to do makes the common cases work: PCRE now takes note
2337 of whether there can be an explicit match for \r or \n anywhere in the
2338 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2339 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2340 pattern has explicit CR or LF references.
2341
2342 29. Added (*CR) etc for changing newline setting at start of pattern.
2343
2344
2345 Version 7.2 19-Jun-07
2346 ---------------------
2347
2348 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2349 which is apparently normally available under Windows.
2350
2351 2. Re-jig the pcregrep tests with different newline settings in an attempt
2352 to make them independent of the local environment's newline setting.
2353
2354 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2355
2356 4. Some of the "internals" tests were previously cut out when the link size
2357 was not 2, because the output contained actual offsets. The recent new
2358 "Z" feature of pcretest means that these can be cut out, making the tests
2359 usable with all link sizes.
2360
2361 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2362 stack recursion. This gives a massive performance boost under BSD, but just
2363 a small improvement under Linux. However, it saves one field in the frame
2364 in all cases.
2365
2366 6. Added more features from the forthcoming Perl 5.10:
2367
2368 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2369 recursion call. It refers to the nth most recently opened parentheses.
2370
2371 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2372 to be opened parentheses.
2373
2374 (c) Conditions that refer to capturing parentheses can be specified
2375 relatively, for example, (?(-2)... or (?(+3)...
2376
2377 (d) \K resets the start of the current match so that everything before
2378 is not part of it.
2379
2380 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2381
2382 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2383 reference syntax.
2384
2385 (g) (?| introduces a group in which the numbering of parentheses in each
2386 alternative starts with the same number.
2387
2388 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2389
2390 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2391 PCRE_INFO_JCHANGED.
2392
2393 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2394 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2395 for detecting groups that can match an empty string.
2396
2397 9. A pattern with a very large number of alternatives (more than several
2398 hundred) was running out of internal workspace during the pre-compile
2399 phase, where pcre_compile() figures out how much memory will be needed. A
2400 bit of new cunning has reduced the workspace needed for groups with
2401 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2402 workspace instead of running out of the 4096 that are available.
2403
2404 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2405
2406 11. Applied patch from Google to remove an optimization that didn't quite work.
2407 The report of the bug said:
2408
2409 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2410 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2411 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2412
2413 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2414 it matched the wrong number of bytes.
2415
2416
2417 Version 7.1 24-Apr-07
2418 ---------------------
2419
2420 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2421 that is more "standard", making use of automake and other Autotools. There
2422 is some re-arrangement of the files and adjustment of comments consequent
2423 on this.
2424
2425 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2426 for recursive directory scanning broke on some systems because the files
2427 are not scanned in any specific order and on different systems the order
2428 was different. A call to "sort" has been inserted into RunGrepTest for the
2429 approprate test as a short-term fix. In the longer term there may be an
2430 alternative.
2431
2432 3. I had an email from Eric Raymond about problems translating some of PCRE's
2433 man pages to HTML (despite the fact that I distribute HTML pages, some
2434 people do their own conversions for various reasons). The problems
2435 concerned the use of low-level troff macros .br and .in. I have therefore
2436 removed all such uses from the man pages (some were redundant, some could
2437 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2438 HTML has been updated to handle .nf/.fi and to complain if it encounters
2439 .br or .in.
2440
2441 4. Updated comments in configure.ac that get placed in config.h.in and also
2442 arranged for config.h to be included in the distribution, with the name
2443 config.h.generic, for the benefit of those who have to compile without
2444 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2445
2446 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2447 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2448 makevp.bat for the current PCRE, using the additional files
2449 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2450
2451 6. A Windows user reported a minor discrepancy with test 2, which turned out
2452 to be caused by a trailing space on an input line that had got lost in his
2453 copy. The trailing space was an accident, so I've just removed it.
2454
2455 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2456 that is needed.
2457
2458 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2459 as "const" (a) because they are and (b) because it helps the PHP
2460 maintainers who have recently made a script to detect big data structures
2461 in the php code that should be moved to the .rodata section. I remembered
2462 to update Builducptable as well, so it won't revert if ucptable.h is ever
2463 re-created.
2464
2465 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2466 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2467 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2468 support is not required. This saves 1.5-2K of code, which is important in
2469 some applications.
2470
2471 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2472 so as not to refer to the tables, even though these functions will never be
2473 called when UTF-8 support is disabled. Otherwise there are problems with a
2474 shared library.
2475
2476 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2477
2478 (a) It was defining its arguments as char * instead of void *.
2479
2480 (b) It was assuming that all moves were upwards in memory; this was true
2481 a long time ago when I wrote it, but is no longer the case.
2482
2483 The emulated memove() is provided for those environments that have neither
2484 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2485 is clearly not the case, as these two bugs were recently reported.
2486
2487 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2488 and Detrail to create the HTML documentation, the .txt form of the man
2489 pages, and it removes trailing spaces from listed files. It also creates
2490 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2491 case, it wraps all the #defines with #ifndefs. This script should be run
2492 before "make dist".
2493
2494 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2495 with Unicode property support.
2496
2497 (a) For a maximizing quantifier, if the two different cases of the
2498 character were of different lengths in their UTF-8 codings (there are
2499 some cases like this - I found 11), and the matching function had to
2500 back up over a mixture of the two cases, it incorrectly assumed they
2501 were both the same length.
2502
2503 (b) When PCRE was configured to use the heap rather than the stack for
2504 recursion during matching, it was not correctly preserving the data for
2505 the other case of a UTF-8 character when checking ahead for a match
2506 while processing a minimizing repeat. If the check also involved
2507 matching a wide character, but failed, corruption could cause an
2508 erroneous result when trying to check for a repeat of the original
2509 character.
2510
2511 13. Some tidying changes to the testing mechanism:
2512
2513 (a) The RunTest script now detects the internal link size and whether there
2514 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2515 values substituted by "configure". (The RunGrepTest script already did
2516 this for UTF-8.) The configure.ac script no longer substitutes the
2517 relevant variables.
2518
2519 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2520 with length and offset values. This means that the output is different
2521 for different internal link sizes. Test 2 is skipped for link sizes
2522 other than 2 because of this, bypassing the problem. Unfortunately,
2523 there was also a test in test 3 (the locale tests) that used /B and
2524 failed for link sizes other than 2. Rather than cut the whole test out,
2525 I have added a new /Z option to pcretest that replaces the length and
2526 offset values with spaces. This is now used to make test 3 independent
2527 of link size. (Test 2 will be tidied up later.)
2528
2529 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2530 segmentation fault instead of returning the appropriate error message.
2531
2532 15. In multiline mode when the newline sequence was set to "any", the pattern
2533 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2534 This doesn't seem right; it now treats the CRLF combination as the line
2535 ending, and so does not match in that case. It's only a pattern such as ^$
2536 that would hit this one: something like ^ABC$ would have failed after \r
2537 and then tried again after \r\n.
2538
2539 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2540 in an attempt to make files that differ only in their line terminators
2541 compare equal. This works on Linux.
2542
2543 17. Under certain error circumstances pcregrep might try to free random memory
2544 as it exited. This is now fixed, thanks to valgrind.
2545
2546 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2547 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2548 was because its rules for how to advance for /g after matching an empty
2549 string at the end of a line did not allow for this case. They now check for
2550 it specially.
2551
2552 20. pcretest is supposed to handle patterns and data of any length, by
2553 extending its buffers when necessary. It was getting this wrong when the
2554 buffer for a data line had to be extended.
2555
2556 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2557 CRLF as a newline sequence.
2558
2559 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2560 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2561 I have nevertheless tidied it up.
2562
2563 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2564
2565 24. Added a man page for pcre-config.
2566
2567
2568 Version 7.0 19-Dec-06
2569 ---------------------
2570
2571 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2572 moving to gcc 4.1.1.
2573
2574 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2575 sys/time.h, which is documented as needed for this function. It doesn't
2576 seem to matter on Linux, but it showed up on some releases of OS X.
2577
2578 3. It seems that there are systems where bytes whose values are greater than
2579 127 match isprint() in the "C" locale. The "C" locale should be the
2580 default when a C program starts up. In most systems, only ASCII printing
2581 characters match isprint(). This difference caused the output from pcretest
2582 to vary, making some of the tests fail. I have changed pcretest so that:
2583
2584 (a) When it is outputting text in the compiled version of a pattern, bytes
2585 other than 32-126 are always shown as hex escapes.
2586
2587 (b) When it is outputting text that is a matched part of a subject string,
2588 it does the same, unless a different locale has been set for the match
2589 (using the /L modifier). In this case, it uses isprint() to decide.
2590
2591 4. Fixed a major bug that caused incorrect computation of the amount of memory
2592 required for a compiled pattern when options that changed within the
2593 pattern affected the logic of the preliminary scan that determines the
2594 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2595 that the computed length was too small. The symptoms of this bug were
2596 either the PCRE error "internal error: code overflow" from pcre_compile(),
2597 or a glibc crash with a message such as "pcretest: free(): invalid next
2598 size (fast)". Examples of patterns that provoked this bug (shown in
2599 pcretest format) are:
2600
2601 /(?-x: )/x
2602 /(?x)(?-x: \s*#\s*)/
2603 /((?i)[\x{c0}])/8
2604 /(?i:[\x{c0}])/8
2605
2606 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2607 is now done differently.
2608
2609 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2610 wrapper classes; (b) implement a new function in the C++ scanner that is
2611 more efficient than the old way of doing things because it avoids levels of
2612 recursion in the regex matching; (c) add a paragraph to the documentation
2613 for the FullMatch() function.
2614
2615 6. The escape sequence \n was being treated as whatever was defined as
2616 "newline". Not only was this contrary to the documentation, which states
2617 that \n is character 10 (hex 0A), but it also went horribly wrong when
2618 "newline" was defined as CRLF. This has been fixed.
2619
2620 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2621 was being set to -1 for the "end of line" case (supposedly a value that no
2622 character can have). Though this value is never used (the check for end of
2623 line is "zero bytes in current character"), it caused compiler complaints.
2624 I've changed it to 0xffffffff.
2625
2626 8. In pcre_version.c, the version string was being built by a sequence of
2627 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2628 string (as it is for production releases) called a macro with an empty
2629 argument. The C standard says the result of this is undefined. The gcc
2630 compiler treats it as an empty string (which was what was wanted) but it is
2631 reported that Visual C gives an error. The source has been hacked around to
2632 avoid this problem.
2633
2634 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2635 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2636 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2637 of them did).
2638
2639 10. Originally, pcretest opened its input and output without "b"; then I was
2640 told that "b" was needed in some environments, so it was added for release
2641 5.0 to both the input and output. (It makes no difference on Unix-like
2642 systems.) Later I was told that it is wrong for the input on Windows. I've
2643 now abstracted the modes into two macros, to make it easier to fiddle with
2644 them, and removed "b" from the input mode under Windows.
2645
2646 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2647
2648 12. Added -help and --help to pcretest as an official way of being reminded
2649 of the options.
2650
2651 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2652 and pcrecpp.cc because they annoy compilers at high warning levels.
2653
2654 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2655
2656 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2657 scripts are not C programs :-) and which was not noticed because it works
2658 on Linux.
2659
2660 16. pcretest is supposed to handle any length of pattern and data line (as one
2661 line or as a continued sequence of lines) by extending its input buffer if
2662 necessary. This feature was broken for very long pattern lines, leading to
2663 a string of junk being passed to pcre_compile() if the pattern was longer
2664 than about 50K.
2665
2666 17. I have done a major re-factoring of the way pcre_compile() computes the
2667 amount of memory needed for a compiled pattern. Previously, there was code
2668 that made a preliminary scan of the pattern in order to do this. That was
2669 OK when PCRE was new, but as the facilities have expanded, it has become
2670 harder and harder to keep it in step with the real compile phase, and there
2671 have been a number of bugs (see for example, 4 above). I have now found a
2672 cunning way of running the real compile function in a "fake" mode that
2673 enables it to compute how much memory it would need, while actually only
2674 ever using a few hundred bytes of working memory and without too many
2675 tests of the mode. This should make future maintenance and development
2676 easier. A side effect of this work is that the limit of 200 on the nesting
2677 depth of parentheses has been removed (though this was never a serious
2678 limitation, I suspect). However, there is a downside: pcre_compile() now
2679 runs more slowly than before (30% or more, depending on the pattern). I
2680 hope this isn't a big issue. There is no effect on runtime performance.
2681
2682 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2683 newline (only possible for the last line of a file) and it was a
2684 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2685
2686 19. Added additional timing features to pcretest. (1) The -tm option now times
2687 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2688 separate command line item, by a number that specifies the number of
2689 repeats to use when timing. The default is 50000; this gives better
2690 precision, but takes uncomfortably long for very large patterns.
2691
2692 20. Extended pcre_study() to be more clever in cases where a branch of a
2693 subpattern has no definite first character. For example, (a*|b*)[cd] would
2694 previously give no result from pcre_study(). Now it recognizes that the
2695 first character must be a, b, c, or d.
2696
2697 21. There was an incorrect error "recursive call could loop indefinitely" if
2698 a subpattern (or the entire pattern) that was being tested for matching an
2699 empty string contained only one non-empty item after a nested subpattern.
2700 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2701 incorrectly, because the \d was being skipped in the check.
2702
2703 22. The pcretest program now has a new pattern option /B and a command line
2704 option -b, which is equivalent to adding /B to every pattern. This causes
2705 it to show the compiled bytecode, without the additional information that
2706 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2707 is the same as /B/I).
2708
2709 23. A new optimization is now able automatically to treat some sequences such
2710 as a*b as a*+b. More specifically, if something simple (such as a character
2711 or a simple class like \d) has an unlimited quantifier, and is followed by
2712 something that cannot possibly match the quantified thing, the quantifier
2713 is automatically "possessified".
2714
2715 24. A recursive reference to a subpattern whose number was greater than 39
2716 went wrong under certain circumstances in UTF-8 mode. This bug could also
2717 have affected the operation of pcre_study().
2718
2719 25. Realized that a little bit of performance could be had by replacing
2720 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2721
2722 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2723
2724 27. Possessive quantifiers such as a++ were previously implemented by turning
2725 them into atomic groups such as ($>a+). Now they have their own opcodes,
2726 which improves performance. This includes the automatically created ones
2727 from 23 above.
2728
2729 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2730 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2731 the first matched character to be a colon. This applied both to named and
2732 numbered groups.
2733
2734 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2735
2736 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2737 building PCRE on Windows easier, so I have included it in the distribution.
2738
2739 31. There is now a check in pcretest against a ridiculously large number being
2740 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2741 loop, the loop is abandoned.
2742
2743 32. Forward references to subpatterns in conditions such as (?(2)...) where
2744 subpattern 2 is defined later cause pcre_compile() to search forwards in
2745 the pattern for the relevant set of parentheses. This search went wrong
2746 when there were unescaped parentheses in a character class, parentheses
2747 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2748
2749 33. "Subroutine" calls and backreferences were previously restricted to
2750 referencing subpatterns earlier in the regex. This restriction has now
2751 been removed.
2752
2753 34. Added a number of extra features that are going to be in Perl 5.10. On the
2754 whole, these are just syntactic alternatives for features that PCRE had
2755 previously implemented using the Python syntax or my own invention. The
2756 other formats are all retained for compatibility.
2757
2758 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2759 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2760 also .NET compatible.
2761
2762 (b) A recursion or subroutine call to a named group can now be defined as
2763 (?&name) as well as (?P>name).
2764
2765 (c) A backreference to a named group can now be defined as \k<name> or
2766 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2767 5.10, are also .NET compatible.
2768
2769 (d) A conditional reference to a named group can now use the syntax
2770 (?(<name>) or (?('name') as well as (?(name).
2771
2772 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2773 groups (named and numbered) that are never evaluated inline, but can be
2774 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2775 is always false. There may be only one alternative in such a group.
2776
2777 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2778 as the simple (?(R). The condition is true only if the most recent
2779 recursion is that of the given number or name. It does not search out
2780 through the entire recursion stack.
2781
2782 (g) The escape \gN or \g{N} has been added, where N is a positive or
2783 negative number, specifying an absolute or relative reference.
2784
2785 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2786 some "unreachable code" warnings.
2787
2788 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2789 things, this adds five new scripts.
2790
2791 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2792 There were also incompatibilities regarding the handling of \Q..\E inside
2793 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2794 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2795
2796 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2797 matches an empty string, and forcibly breaks the loop. There were bugs in
2798 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2799 against aaaa the result was just "a" rather than "aaaa", for example. Two
2800 separate and independent bugs (that affected different cases) have been
2801 fixed.
2802
2803 39. Refactored the code to abolish the use of different opcodes for small
2804 capturing bracket numbers. This is a tidy that I avoided doing when I
2805 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2806 The new approach is not only tidier, it makes it possible to reduce the
2807 memory needed to fix the previous bug (38).
2808
2809 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2810 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2811 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2812 mode.
2813
2814 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2815 report.
2816
2817 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2818 copy construction and assignment in the C++ wrapper.
2819
2820 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2821 couple of bugs that could have given wrong results in the "--newline=crlf"
2822 case.
2823
2824 44. Added a number of casts and did some reorganization of signed/unsigned int
2825 variables following suggestions from Dair Grant. Also renamed the variable
2826 "this" as "item" because it is a C++ keyword.
2827
2828 45. Arranged for dftables to add
2829
2830 #include "pcre_internal.h"
2831
2832 to pcre_chartables.c because without it, gcc 4.x may remove the array
2833 definition from the final binary if PCRE is built into a static library and
2834 dead code stripping is activated.
2835
2836 46. For an unanchored pattern, if a match attempt fails at the start of a
2837 newline sequence, and the newline setting is CRLF or ANY, and the next two
2838 characters are CRLF, advance by two characters instead of one.
2839
2840
2841 Version 6.7 04-Jul-06
2842 ---------------------
2843
2844 1. In order to handle tests when input lines are enormously long, pcretest has
2845 been re-factored so that it automatically extends its buffers when
2846 necessary. The code is crude, but this _is_ just a test program. The
2847 default size has been increased from 32K to 50K.
2848
2849 2. The code in pcre_study() was using the value of the re argument before
2850 testing it for NULL. (Of course, in any sensible call of the function, it
2851 won't be NULL.)
2852
2853 3. The memmove() emulation function in pcre_internal.h, which is used on
2854 systems that lack both memmove() and bcopy() - that is, hardly ever -
2855 was missing a "static" storage class specifier.
2856
2857 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2858 containing an extended class (one that cannot be represented by a bitmap
2859 because it contains high-valued characters or Unicode property items, e.g.
2860 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2861 pattern, but PCRE should not loop if you do not (it no longer does).
2862 [Detail: two cases were found: (a) a repeated subpattern containing an
2863 extended class; (b) a recursive reference to a subpattern that followed a
2864 previous extended class. It wasn't skipping over the extended class
2865 correctly when UTF-8 mode was not set.]
2866
2867 5. A negated single-character class was not being recognized as fixed-length
2868 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2869 compile error "lookbehind assertion is not fixed length".
2870
2871 6. The RunPerlTest auxiliary script was showing an unexpected difference
2872 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2873 write a Perl script that can interpret lines of an input file either as
2874 byte characters or as UTF-8, which is what "perltest" was being required to
2875 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2876 can't do is switch easily at run time between having the "use utf8;" pragma
2877 or not. In the end, I fudged it by using the RunPerlTest script to insert
2878 "use utf8;" explicitly for the UTF-8 tests.
2879
2880 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2881 the end of the subject string, contrary to the documentation and to what
2882 Perl does. This was true of both matching functions. Now it matches only at
2883 the start of the subject and immediately after *internal* newlines.
2884
2885 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2886 a pointer to an int instead of a pointer to an unsigned long int. This
2887 caused problems on 64-bit systems.
2888
2889 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2890 instance of the 'standard' template library not being so standard".
2891
2892 10. There was no check on the number of named subpatterns nor the maximum
2893 length of a subpattern name. The product of these values is used to compute
2894 the size of the memory block for a compiled pattern. By supplying a very
2895 long subpattern name and a large number of named subpatterns, the size
2896 computation could be caused to overflow. This is now prevented by limiting
2897 the length of names to 32 characters, and the number of named subpatterns
2898 to 10,000.
2899
2900 11. Subpatterns that are repeated with specific counts have to be replicated in
2901 the compiled pattern. The size of memory for this was computed from the
2902 length of the subpattern and the repeat count. The latter is limited to
2903 65535, but there was no limit on the former, meaning that integer overflow
2904 could in principle occur. The compiled length of a repeated subpattern is
2905 now limited to 30,000 bytes in order to prevent this.
2906
2907 12. Added the optional facility to have named substrings with the same name.
2908
2909 13. Added the ability to use a named substring as a condition, using the
2910 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2911 are numbers (not recommended). Forward references are permitted.
2912
2913 14. Added forward references in named backreferences (if you see what I mean).
2914
2915 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2916 pattern could run off the end of the subject. For example, the pattern
2917 "(?s)(.{1,5})"8 did this with the subject "ab".
2918
2919 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2920 PCRE_CASELESS was set when matching characters that were quantified with ?
2921 or *.
2922
2923 17. A character class other than a single negated character that had a minimum
2924 but no maximum quantifier - for example [ab]{6,} - was not handled
2925 correctly by pce_dfa_exec(). It would match only one character.
2926
2927 18. A valid (though odd) pattern that looked like a POSIX character
2928 class but used an invalid character after [ (for example [[,abc,]]) caused
2929 pcre_compile() to give the error "Failed: internal error: code overflow" or
2930 in some cases to crash with a glibc free() error. This could even happen if
2931 the pattern terminated after [[ but there just happened to be a sequence of
2932 letters, a binary zero, and a closing ] in the memory that followed.
2933
2934 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2935 over the years. Originally (before any Unicode support), just the bottom 8
2936 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2937 output from "man perlunicode" includes this:
2938
2939 The regular expression compiler produces polymorphic opcodes. That
2940 is, the pattern adapts to the data and automatically switches to
2941 the Unicode character scheme when presented with Unicode data--or
2942 instead uses a traditional byte scheme when presented with byte
2943 data.
2944
2945 Sadly, a wide octal escape does not cause a switch, and in a string with
2946 no other multibyte characters, these octal escapes are treated as before.
2947 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2948 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2949 Unicode string.
2950
2951 I have not perpetrated such confusion in PCRE. Up till now, it took just
2952 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2953 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2954 translate to the appropriate multibyte character.
2955
2956 29. Applied some refactoring to reduce the number of warnings from Microsoft
2957 and Borland compilers. This has included removing the fudge introduced
2958 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2959 a warning about an unused variable.
2960
2961 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2962 characters since release 4.0, because Perl (from release 5.004) does not.
2963 [Or at least, is documented not to: some releases seem to be in conflict
2964 with the documentation.] However, when a pattern was studied with
2965 pcre_study() and all its branches started with \s, PCRE still included VT
2966 as a possible starting character. Of course, this did no harm; it just
2967 caused an unnecessary match attempt.
2968
2969 22. Removed a now-redundant internal flag bit that recorded the fact that case
2970 dependency changed within the pattern. This was once needed for "required
2971 byte" processing, but is no longer used. This recovers a now-scarce options
2972 bit. Also moved the least significant internal flag bit to the most-
2973 significant bit of the word, which was not previously used (hangover from
2974 the days when it was an int rather than a uint) to free up another bit for
2975 the future.
2976
2977 23. Added support for CRLF line endings as well as CR and LF. As well as the
2978 default being selectable at build time, it can now be changed at runtime
2979 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2980 specify that it is scanning data with non-default line endings.
2981
2982 24. Changed the definition of CXXLINK to make it agree with the definition of
2983 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2984
2985 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2986 recursions. This makes a big different to stack usage for some patterns.
2987
2988 26. If a subpattern containing a named recursion or subroutine reference such
2989 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2990 the space required for the compiled pattern went wrong and gave too small a
2991 value. Depending on the environment, this could lead to "Failed: internal
2992 error: code overflow at offset 49" or "glibc detected double free or
2993 corruption" errors.
2994
2995 27. Applied patches from Google (a) to support the new newline modes and (b) to
2996 advance over multibyte UTF-8 characters in GlobalReplace.
2997
2998 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2999 difference for some implementation of PCRE in some Windows version.
3000
3001 29. Added some extra testing facilities to pcretest:
3002
3003 \q<number> in a data line sets the "match limit" value
3004 \Q<number> in a data line sets the "match recursion limt" value
3005 -S <number> sets the stack size, where <number> is in megabytes
3006
3007 The -S option isn't available for Windows.
3008
3009
3010 Version 6.6 06-Feb-06
3011 ---------------------
3012
3013 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3014 in pcreposix.h. I have copied the definition from pcre.h.
3015
3016 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3017 because pcre.h is no longer a built file.
3018
3019 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3020 not normally included in the compiled code.
3021
3022
3023 Version 6.5 01-Feb-06
3024 ---------------------
3025
3026 1. When using the partial match feature with pcre_dfa_exec(), it was not
3027 anchoring the second and subsequent partial matches at the new starting
3028 point. This could lead to incorrect results. For example, with the pattern
3029 /1234/, partially matching against "123" and then "a4" gave a match.
3030
3031 2. Changes to pcregrep:
3032
3033 (a) All non-match returns from pcre_exec() were being treated as failures
3034 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3035 error message is output. Some extra information is given for the
3036 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3037 probably the only errors that are likely to be caused by users (by
3038 specifying a regex that has nested indefinite repeats, for instance).
3039 If there are more than 20 of these errors, pcregrep is abandoned.
3040
3041 (b) A binary zero was treated as data while matching, but terminated the
3042 output line if it was written out. This has been fixed: binary zeroes
3043 are now no different to any other data bytes.
3044
3045 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3046 used to set a locale for matching. The --locale=xxxx long option has
3047 been added (no short equivalent) to specify a locale explicitly on the
3048 pcregrep command, overriding the environment variables.
3049
3050 (d) When -B was used with -n, some line numbers in the output were one less
3051 than they should have been.
3052
3053 (e) Added the -o (--only-matching) option.
3054
3055 (f) If -A or -C was used with -c (count only), some lines of context were
3056 accidentally printed for the final match.
3057
3058 (g) Added the -H (--with-filename) option.
3059
3060 (h) The combination of options -rh failed to suppress file names for files
3061 that were found from directory arguments.
3062
3063 (i) Added the -D (--devices) and -d (--directories) options.
3064
3065 (j) Added the -F (--fixed-strings) option.
3066
3067 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3068
3069 (l) Added the --colo(u)r option.
3070
3071 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3072 is not present by default.
3073
3074 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3075 items such as (?R) or (?1), when the recursion could match a number of
3076 alternatives. If it matched one of the alternatives, but subsequently,
3077 outside the recursion, there was a failure, the code tried to back up into
3078 the recursion. However, because of the way PCRE is implemented, this is not
3079 possible, and the result was an incorrect result from the match.
3080
3081 In order to prevent this happening, the specification of recursion has
3082 been changed so that all such subpatterns are automatically treated as
3083 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3084
3085 4. I had overlooked the fact that, in some locales, there are characters for
3086 which isalpha() is true but neither isupper() nor islower() are true. In
3087 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3088 and ordfeminine) are like this. This affected the treatment of \w and \W
3089 when they appeared in character classes, but not when they appeared outside
3090 a character class. The bit map for "word" characters is now created
3091 separately from the results of isalnum() instead of just taking it from the
3092 upper, lower, and digit maps. (Plus the underscore character, of course.)
3093
3094 5. The above bug also affected the handling of POSIX character classes such as
3095 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3096 permanent tables. Instead, the bit maps for such a class were previously
3097 created as the appropriate unions of the upper, lower, and digit bitmaps.
3098 Now they are created by subtraction from the [[:word:]] class, which has
3099 its own bitmap.
3100
3101 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3102 It is created by subtracting the vertical space characters (\x09, \x0a,
3103 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3104 subtraction was done in the overall bitmap for a character class, meaning
3105 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3106 be recognized. This bug has been fixed.
3107
3108 7. Patches from the folks at Google:
3109
3110 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3111 real life, but is still worth protecting against".
3112
3113 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3114 regular expressions".
3115
3116 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3117 have it.
3118
3119 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3120 "configure" and the latter not, in order to fix a problem somebody had
3121 with compiling the Arg class on HP-UX.
3122
3123 (e) Improve the error-handling of the C++ wrapper a little bit.
3124
3125 (f) New tests for checking recursion limiting.
3126
3127 8. The pcre_memmove() function, which is used only if the environment does not
3128 have a standard memmove() function (and is therefore rarely compiled),
3129 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3130 returning a result (though PCRE never actually uses the result).
3131
3132 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3133 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3134 returned instead of calling malloc() with an overflowing number that would
3135 most likely cause subsequent chaos.
3136
3137 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3138
3139 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3140 with this option is matched, the nmatch and pmatch options of regexec() are
3141 ignored.
3142
3143 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3144 provided in case anyone wants to the the POSIX interface with UTF-8
3145 strings.
3146
3147 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3148 C++ linking (needed for some HP-UX environments).
3149
3150 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3151 (unused parameter) and in the pcre_printint() function (omitted "default"
3152 switch label when the default is to do nothing).
3153
3154 15. Added some code to make it possible, when PCRE is compiled as a C++
3155 library, to replace subject pointers for pcre_exec() with a smart pointer
3156 class, thus making it possible to process discontinuous strings.
3157
3158 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3159 much the same function. They were added by different people who were trying
3160 to make PCRE easy to compile on non-Unix systems. It has been suggested
3161 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3162 for compiling on Windows systems. I have therefore replaced it with
3163 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3164 defaults to "extern" for C or "extern C" for C++, which works fine on
3165 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3166 SCOPE with something explicit in config.h. In addition:
3167
3168 (a) pcreposix.h still had just "extern" instead of either of these macros;
3169 I have replaced it with PCRE_DATA_SCOPE.
3170
3171 (b) Functions such as _pcre_xclass(), which are internal to the library,
3172 but external in the C sense, all had PCRE_EXPORT in their definitions.
3173 This is apparently wrong for the Windows case, so I have removed it.
3174 (It makes no difference on Unix-like systems.)
3175
3176 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3177 of recursive calls to match(). This is different to MATCH_LIMIT because
3178 that limits the total number of calls to match(), not all of which increase
3179 the depth of recursion. Limiting the recursion depth limits the amount of
3180 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3181 when PCRE is compiled, and changed at run time. A patch from Google adds
3182 this functionality to the C++ interface.
3183
3184 18. Changes to the handling of Unicode character properties:
3185
3186 (a) Updated the table to Unicode 4.1.0.
3187
3188 (b) Recognize characters that are not in the table as "Cn" (undefined).
3189
3190 (c) I revised the way the table is implemented to a much improved format
3191 which includes recognition of ranges. It now supports the ranges that
3192 are defined in UnicodeData.txt, and it also amalgamates other
3193 characters into ranges. This has reduced the number of entries in the
3194 table from around 16,000 to around 3,000, thus reducing its size
3195 considerably. I realized I did not need to use a tree structure after
3196 all - a binary chop search is just as efficient. Having reduced the
3197 number of entries, I extended their size from 6 bytes to 8 bytes to
3198 allow for more data.
3199
3200 (d) Added support for Unicode script names via properties such as \p{Han}.
3201
3202 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3203 matching that character.
3204
3205 20. When matching a repeated Unicode property with a minimum greater than zero,
3206 (for example \pL{2,}), PCRE could look past the end of the subject if it
3207 reached it while seeking the minimum number of characters. This could
3208 happen only if some of the characters were more than one byte long, because
3209 there is a check for at least the minimum number of bytes.
3210
3211 21. Refactored the implementation of \p and \P so as to be more general, to
3212 allow for more different types of property in future. This has changed the
3213 compiled form incompatibly. Anybody with saved compiled patterns that use
3214 \p or \P will have to recompile them.
3215
3216 22. Added "Any" and "L&" to the supported property types.
3217
3218 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3219 but give a compile time error if the value is greater than 0xff.
3220
3221 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3222 accidentally not being installed or uninstalled.
3223
3224 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3225 made were to insert the current release number. This seemed silly, because
3226 it made things harder for people building PCRE on systems that don't run
3227 "configure". I have turned pcre.h into a distributed file, no longer built
3228 by "configure", with the version identification directly included. There is
3229 no longer a pcre.h.in file.
3230
3231 However, this change necessitated a change to the pcre-config script as
3232 well. It is built from pcre-config.in, and one of the substitutions was the
3233 release number. I have updated configure.ac so that ./configure now finds
3234 the release number by grepping pcre.h.
3235
3236 26. Added the ability to run the tests under valgrind.
3237
3238
3239 Version 6.4 05-Sep-05
3240 ---------------------
3241
3242 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3243 "--" to be printed when multiple files were scanned, even when none of the
3244 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3245 consider it to be a bug, and have restored the previous behaviour.
3246
3247 2. A couple of code tidies to get rid of compiler warnings.
3248
3249 3. The pcretest program used to cheat by referring to symbols in the library
3250 whose names begin with _pcre_. These are internal symbols that are not
3251 really supposed to be visible externally, and in some environments it is
3252 possible to suppress them. The cheating is now confined to including
3253 certain files from the library's source, which is a bit cleaner.
3254
3255 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3256 file's purpose clearer.
3257
3258 5. Reorganized pcre_ucp_findchar().
3259
3260
3261 Version 6.3 15-Aug-05
3262 ---------------------
3263
3264 1. The file libpcre.pc.in did not have general read permission in the tarball.
3265
3266 2. There were some problems when building without C++ support:
3267
3268 (a) If C++ support was not built, "make install" and "make test" still
3269 tried to test it.
3270
3271 (b) There were problems when the value of CXX was explicitly set. Some
3272 changes have been made to try to fix these, and ...
3273
3274 (c) --disable-cpp can now be used to explicitly disable C++ support.
3275
3276 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3277 backslash in a target when C++ was disabled. This confuses some
3278 versions of "make", apparently. Using an intermediate variable solves
3279 this. (Same for CPP_LOBJ.)
3280
3281 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3282 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3283 necessary on certain architectures.
3284
3285 4. Added a setting of -export-symbols-regex to the link command to remove
3286 those symbols that are exported in the C sense, but actually are local
3287 within the library, and not documented. Their names all begin with
3288 "_pcre_". This is not a perfect job, because (a) we have to except some
3289 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3290 available (and never for static libraries). I have made a note to try to
3291 find a way round (a) in the future.
3292
3293
3294 Version 6.2 01-Aug-05
3295 ---------------------
3296
3297 1. There was no test for integer overflow of quantifier values. A construction
3298 such as {1111111111111111} would give undefined results. What is worse, if
3299 a minimum quantifier for a parenthesized subpattern overflowed and became
3300 negative, the calculation of the memory size went wrong. This could have
3301 led to memory overwriting.
3302
3303 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3304
3305 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3306 operating environments where this matters.
3307
3308 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3309 PCRE options from within the C++ wrapper.
3310
3311 5. Named capturing subpatterns were not being correctly counted when a pattern
3312 was compiled. This caused two problems: (a) If there were more than 100
3313 such subpatterns, the calculation of the memory needed for the whole
3314 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3315 back references of the form \12, where the number was greater than 9, were
3316 not recognized as back references, even though there were sufficient
3317 previous subpatterns.
3318
3319 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3320 versions of gcc, e.g. 2.95.4.
3321
3322
3323 Version 6.1 21-Jun-05
3324 ---------------------
3325
3326 1. There was one reference to the variable "posix" in pcretest.c that was not
3327 surrounded by "#if !defined NOPOSIX".
3328
3329 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3330 the cross-check on the old pcre_info() function, for the benefit of the
3331 cut-down version of PCRE that is currently imported into Exim.
3332
3333 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3334 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3335 patterns that start (?i) but I don't think that matters. The use of (?i) is
3336 just an example; this all applies to the other options as well.
3337
3338 4. Since libtool seems to echo the compile commands it is issuing, the output
3339 from "make" can be reduced a bit by putting "@" in front of each libtool
3340 compile command.
3341
3342 5. Patch from the folks at Google for configure.in to be a bit more thorough
3343 in checking for a suitable C++ installation before trying to compile the
3344 C++ stuff. This should fix a reported problem when a compiler was present,
3345 but no suitable headers.
3346
3347 6. The man pages all had just "PCRE" as their title. I have changed them to
3348 be the relevant file name. I have also arranged that these names are
3349 retained in the file doc/pcre.txt, which is a concatenation in text format
3350 of all the man pages except the little individual ones for each function.
3351
3352 7. The NON-UNIX-USE file had not been updated for the different set of source
3353 files that come with release 6. I also added a few comments about the C++
3354 wrapper.
3355
3356
3357 Version 6.0 07-Jun-05
3358 ---------------------
3359
3360 1. Some minor internal re-organization to help with my DFA experiments.
3361
3362 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3363 didn't matter for the library itself when fully configured, but did matter
3364 when compiling without UCP support, or within Exim, where the ucp files are
3365 not imported.
3366
3367 3. Refactoring of the library code to split up the various functions into
3368 different source modules. The addition of the new DFA matching code (see
3369 below) to a single monolithic source would have made it really too
3370 unwieldy, quite apart from causing all the code to be include in a
3371 statically linked application, when only some functions are used. This is
3372 relevant even without the DFA addition now that patterns can be compiled in
3373 one application and matched in another.
3374
3375 The downside of splitting up is that there have to be some external
3376 functions and data tables that are used internally in different modules of
3377 the library but which are not part of the API. These have all had their
3378 names changed to start with "_pcre_" so that they are unlikely to clash
3379 with other external names.
3380
3381 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3382 a different (DFA) algorithm. Although it is slower than the original
3383 function, it does have some advantages for certain types of matching
3384 problem.
3385
3386 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3387 including restarting after a partial match.
3388
3389 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3390 defined when compiling for Windows was sent to me. I have put it into the
3391 code, though I have no means of testing or verifying it.
3392
3393 7. Added the pcre_refcount() auxiliary function.
3394
3395 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3396 match before or at the first newline in the subject string. In pcretest,
3397 the /f option on a pattern can be used to set this.
3398
3399 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3400 would behave wrongly. This has been present in PCRE since release 4.0.
3401
3402 10. A number of changes to the pcregrep command:
3403
3404 (a) Refactored how -x works; insert ^(...)$ instead of setting
3405 PCRE_ANCHORED and checking the length, in preparation for adding
3406 something similar for -w.
3407
3408 (b) Added the -w (match as a word) option.
3409
3410 (c) Refactored the way lines are read and buffered so as to have more
3411 than one at a time available.
3412
3413 (d) Implemented a pcregrep test script.
3414
3415 (e) Added the -M (multiline match) option. This allows patterns to match
3416 over several lines of the subject. The buffering ensures that at least
3417 8K, or the rest of the document (whichever is the shorter) is available
3418 for matching (and similarly the previous 8K for lookbehind assertions).
3419
3420 (f) Changed the --help output so that it now says
3421
3422 -w, --word-regex(p)
3423
3424 instead of two lines, one with "regex" and the other with "regexp"
3425 because that confused at least one person since the short forms are the
3426 same. (This required a bit of code, as the output is generated
3427 automatically from a table. It wasn't just a text change.)
3428
3429 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3430 option but starts with a hyphen. Could be a pattern or a path name
3431 starting with a hyphen, for instance.
3432
3433 (h) "-" can be given as a file name to represent stdin.
3434
3435 (i) When file names are being printed, "(standard input)" is used for
3436 the standard input, for compatibility with GNU grep. Previously
3437 "<stdin>" was used.
3438
3439 (j) The option --label=xxx can be used to supply a name to be used for
3440 stdin when file names are being printed. There is no short form.
3441
3442 (k) Re-factored the options decoding logic because we are going to add
3443 two more options that take data. Such options can now be given in four
3444 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3445
3446 (l) Added the -A, -B, and -C options for requesting that lines of context
3447 around matches be printed.
3448
3449 (m) Added the -L option to print the names of files that do not contain
3450 any matching lines, that is, the complement of -l.
3451
3452 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3453 continue to scan other files.
3454
3455 (o) The -s option was incorrectly implemented. For compatibility with other
3456 greps, it now suppresses the error message for a non-existent or non-
3457 accessible file (but not the return code). There is a new option called
3458 -q that suppresses the output of matching lines, which was what -s was
3459 previously doing.
3460
3461 (p) Added --include and --exclude options to specify files for inclusion
3462 and exclusion when recursing.
3463
3464 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3465 Hopefully, it now does.
3466
3467 12. Missing cast in pcre_study().
3468
3469 13. Added an "uninstall" target to the makefile.
3470
3471 14. Replaced "extern" in the function prototypes in Makefile.in with
3472 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3473 world, but is set differently for Windows.
3474
3475 15. Added a second compiling function called pcre_compile2(). The only
3476 difference is that it has an extra argument, which is a pointer to an
3477 integer error code. When there is a compile-time failure, this is set
3478 non-zero, in addition to the error test pointer being set to point to an
3479 error message. The new argument may be NULL if no error number is required
3480 (but then you may as well call pcre_compile(), which is now just a
3481 wrapper). This facility is provided because some applications need a
3482 numeric error indication, but it has also enabled me to tidy up the way
3483 compile-time errors are handled in the POSIX wrapper.
3484
3485 16. Added VPATH=.libs to the makefile; this should help when building with one
3486 prefix path and installing with another. (Or so I'm told by someone who
3487 knows more about this stuff than I do.)
3488
3489 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3490 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3491 match everything, including newlines. This is not POSIX-compatible, but
3492 somebody wanted the feature. From pcretest it can be activated by using
3493 both the P and the s flags.
3494
3495 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3496
3497 19. libpcre.pc was being incorrectly installed as executable.
3498
3499 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3500 it now also looks for '\r' so that it will work unmodified on Windows.
3501
3502 21. Added Google's contributed C++ wrapper to the distribution.
3503
3504 22. Added some untidy missing memory free() calls in pcretest, to keep
3505 Electric Fence happy when testing.
3506
3507
3508
3509 Version 5.0 13-Sep-04
3510 ---------------------
3511
3512 1. Internal change: literal characters are no longer packed up into items
3513 containing multiple characters in a single byte-string. Each character
3514 is now matched using a separate opcode. However, there may be more than one
3515 byte in the character in UTF-8 mode.
3516
3517 2. The pcre_callout_block structure has two new fields: pattern_position and
3518 next_item_length. These contain the offset in the pattern to the next match
3519 item, and its length, respectively.
3520
3521 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3522 insertion of callouts before each pattern item. Added the /C option to
3523 pcretest to make use of this.
3524
3525 4. On the advice of a Windows user, the lines
3526
3527 #if defined(_WIN32) || defined(WIN32)
3528 _setmode( _fileno( stdout ), 0x8000 );
3529 #endif /* defined(_WIN32) || defined(WIN32) */
3530
3531 have been added to the source of pcretest. This apparently does useful
3532 magic in relation to line terminators.
3533
3534 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3535 for the benefit of those environments where the "b" makes a difference.
3536
3537 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3538 to know about it. I have put a hack into configure.in that adds in code
3539 to set GCC=yes if CC=icc. This seems to end up at a point in the
3540 generated configure script that is early enough to affect the setting of
3541 compiler options, which is what is needed, but I have no means of testing
3542 whether it really works. (The user who reported this had patched the
3543 generated configure script, which of course I cannot do.)
3544
3545 LATER: After change 22 below (new libtool files), the configure script
3546 seems to know about icc (and also ecc). Therefore, I have commented out
3547 this hack in configure.in.
3548
3549 7. Added support for pkg-config (2 patches were sent in).
3550
3551 8. Negated POSIX character classes that used a combination of internal tables
3552 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3553 [[:^ascii]]. Typically, they would match almost any characters. The other
3554 POSIX classes were not broken in this way.
3555
3556 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3557 to find the match, as PCRE was deluded into thinking that the match had to
3558 start at the start point or following a newline. The same bug applied to
3559 patterns with negative forward assertions or any backward assertions
3560 preceding ".*" at the start, unless the pattern required a fixed first
3561 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3562
3563 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3564 starting at the last subject character, bytes beyond the end of the subject
3565 string were read.
3566
3567 11. Renamed the variable "class" as "classbits" to make life easier for C++
3568 users. (Previously there was a macro definition, but it apparently wasn't
3569 enough.)
3570
3571 12. Added the new field "tables" to the extra data so that tables can be passed
3572 in at exec time, or the internal tables can be re-selected. This allows
3573 a compiled regex to be saved and re-used at a later time by a different
3574 program that might have everything at different addresses.
3575
3576 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3577 -R library as well as a -L library.
3578
3579 14. The debugging options of pcretest (-d on the command line or D on a
3580 pattern) showed incorrect output for anything following an extended class
3581 that contained multibyte characters and which was followed by a quantifier.
3582
3583 15. Added optional support for general category Unicode character properties
3584 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3585 support. It adds about 90K to the size of the library. The meanings of the
3586 inbuilt class escapes such as \d and \s have NOT been changed.
3587
3588 16. Updated pcredemo.c to include calls to free() to release the memory for the
3589 compiled pattern.
3590
3591 17. The generated file chartables.c was being created in the source directory
3592 instead of in the building directory. This caused the build to fail if the
3593 source directory was different from the building directory, and was
3594 read-only.
3595
3596 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3597 file. No doubt somebody will tell me if they don't make sense... Also added
3598 Dan Mooney's comments about building on OpenVMS.
3599
3600 19. Added support for partial matching via the PCRE_PARTIAL option for
3601 pcre_exec() and the \P data escape in pcretest.
3602
3603 20. Extended pcretest with 3 new pattern features:
3604
3605 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3606 write the compiled pattern to the file whose name is "rest-of-line".
3607 This is a straight binary dump of the data, with the saved pointer to
3608 the character tables forced to be NULL. The study data, if any, is
3609 written too. After writing, pcretest reads a new pattern.
3610
3611 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3612 compiled pattern from the given file. There must not be any
3613 occurrences of "<" in the file name (pretty unlikely); if there are,
3614 pcretest will instead treat the initial "<" as a pattern delimiter.
3615 After reading in the pattern, pcretest goes on to read data lines as
3616 usual.
3617
3618 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3619 and 16-bit fields in a compiled pattern, to simulate a pattern that
3620 was compiled on a host of opposite endianness.
3621
3622 21. The pcre-exec() function can now cope with patterns that were compiled on
3623 hosts of opposite endianness, with this restriction:
3624
3625 As for any compiled expression that is saved and used later, the tables
3626 pointer field cannot be preserved; the extra_data field in the arguments
3627 to pcre_exec() should be used to pass in a tables address if a value
3628 other than the default internal tables were used at compile time.
3629
3630 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3631 now diagnosed as an error. Previously, most of the time, a negative number
3632 would have been treated as zero, but if in addition "ovector" was passed as
3633 NULL, a crash could occur.
3634
3635 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3636 new versions from the libtool 1.5 distribution (the last one is a copy of
3637 a file called libtool.m4). This seems to have fixed the need to patch
3638 "configure" to support Darwin 1.3 (which I used to do). However, I still
3639 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3640 workstation).
3641
3642 24. Changed the PCRE licence to be the more standard "BSD" licence.
3643
3644
3645 Version 4.5 01-Dec-03
3646 ---------------------
3647
3648 1. There has been some re-arrangement of the code for the match() function so
3649 that it can be compiled in a version that does not call itself recursively.
3650 Instead, it keeps those local variables that need separate instances for
3651 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3652 needs to "recurse". Keeping track of where control must go is done by means
3653 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3654 hide most of the details from the main code, and operates only if
3655 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3656 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3657 operating.
3658
3659 To make it easier for callers to provide specially tailored get/free
3660 functions for this usage, two new functions, pcre_stack_malloc, and
3661 pcre_stack_free, are used. They are always called in strict stacking order,
3662 and the size of block requested is always the same.
3663
3664 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3665 PCRE has been compiled to use the stack or the heap for recursion. The
3666 -C option of pcretest uses this to show which version is compiled.
3667
3668 A new data escape \S, is added to pcretest; it causes the amounts of store
3669 obtained and freed by both kinds of malloc/free at match time to be added
3670 to the output.
3671
3672 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3673 what's available on my current Linux desktop machine.
3674
3675 3. When matching a UTF-8 string, the test for a valid string at the start has
3676 been extended. If start_offset is not zero, PCRE now checks that it points
3677 to a byte that is the start of a UTF-8 character. If not, it returns
3678 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3679 this is necessary because there may be backward assertions in the pattern.
3680 When matching the same subject several times, it may save resources to use
3681 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3682
3683 4. The code for checking the validity of UTF-8 strings has been tightened so
3684 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3685 containing "overlong sequences".
3686
3687 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3688 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3689 should have been "&", but it just so happened that all the cases this let
3690 through by mistake were picked up later in the function.
3691
3692 6. I had used a variable called "isblank" - this is a C99 function, causing
3693 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3694
3695 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3696 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3697 so the version is shown; (c) stop "make test" echoing "./RunTest".
3698
3699 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3700
3701 9. The prototype for memmove() for systems that don't have it was using
3702 size_t, but the inclusion of the header that defines size_t was later. I've
3703 moved the #includes for the C headers earlier to avoid this.
3704
3705 10. Added some adjustments to the code to make it easier to compiler on certain
3706 special systems:
3707
3708 (a) Some "const" qualifiers were missing.
3709 (b) Added the macro EXPORT before all exported functions; by default this
3710 is defined to be empty.
3711 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3712 that it reads its output file name as an argument instead of writing
3713 to the standard output and assuming this can be redirected.
3714
3715 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3716 class containing characters with values greater than 255, PCRE compilation
3717 went into a loop.
3718
3719 12. A recursive reference to a subpattern that was within another subpattern
3720 that had a minimum quantifier of zero caused PCRE to crash. For example,
3721 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3722 recursion. If the recursively-called subpattern itself had a zero repeat,
3723 that was OK.
3724
3725 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3726 buffer into which it was copied (for escape processing) was still set at
3727 1024, so long lines caused crashes.
3728
3729 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3730 "internal error: code overflow...". This applied to any character class
3731 that was followed by a possessive quantifier.
3732
3733 15. Modified the Makefile to add libpcre.la as a prerequisite for
3734 libpcreposix.la because I was told this is needed for a parallel build to
3735 work.
3736
3737 16. If a pattern that contained .* following optional items at the start was
3738 studied, the wrong optimizing data was generated, leading to matching
3739 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3740 matching string must start with a or b or c. The correct conclusion for
3741 this pattern is that a match can start with any character.
3742
3743
3744 Version 4.4 13-Aug-03
3745 ---------------------
3746
3747 1. In UTF-8 mode, a character class containing characters with values between
3748 127 and 255 was not handled correctly if the compiled pattern was studied.
3749 In fixing this, I have also improved the studying algorithm for such
3750 classes (slightly).
3751
3752 2. Three internal functions had redundant arguments passed to them. Removal
3753 might give a very teeny performance improvement.
3754
3755 3. Documentation bug: the value of the capture_top field in a callout is *one
3756 more than* the number of the hightest numbered captured substring.
3757
3758 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3759 in incorrectly linking with a previously installed version. They now link
3760 explicitly with libpcre.la.
3761
3762 5. configure.in no longer needs to recognize Cygwin specially.
3763
3764 6. A problem in pcre.in for Windows platforms is fixed.
3765
3766 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3767 pcretest, it used to include the size of the study block as part of its
3768 output. Unfortunately, the structure contains a field that has a different
3769 size on different hardware architectures. This meant that the tests that
3770 showed this size failed. As the block is currently always of a fixed size,
3771 this information isn't actually particularly useful in pcretest output, so
3772 I have just removed it.
3773
3774 8. Three pre-processor statements accidentally did not start in column 1.
3775 Sadly, there are *still* compilers around that complain, even though
3776 standard C has not required this for well over a decade. Sigh.
3777
3778 9. In pcretest, the code for checking callouts passed small integers in the
3779 callout_data field, which is a void * field. However, some picky compilers
3780 complained about the casts involved for this on 64-bit systems. Now
3781 pcretest passes the address of the small integer instead, which should get
3782 rid of the warnings.
3783
3784 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3785 both compile and run time, and gives an error if an invalid UTF-8 sequence
3786 is found. There is a option for disabling this check in cases where the
3787 string is known to be correct and/or the maximum performance is wanted.
3788
3789 11. In response to a bug report, I changed one line in Makefile.in from
3790
3791 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3792 to
3793 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3794
3795 to look similar to other lines, but I have no way of telling whether this
3796 is the right thing to do, as I do not use Windows. No doubt I'll get told
3797 if it's wrong...
3798
3799
3800 Version 4.3 21-May-03
3801 ---------------------
3802
3803 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3804 Makefile.
3805
3806 2. Some refactoring to improve the quality of the code:
3807
3808 (i) The utf8_table... variables are now declared "const".
3809
3810 (ii) The code for \cx, which used the "case flipping" table to upper case
3811 lower case letters, now just substracts 32. This is ASCII-specific,
3812 but the whole concept of \cx is ASCII-specific, so it seems
3813 reasonable.
3814
3815 (iii) PCRE was using its character types table to recognize decimal and
3816 hexadecimal digits in the pattern. This is silly, because it handles
3817 only 0-9, a-f, and A-F, but the character types table is locale-
3818 specific, which means strange things might happen. A private
3819 table is now used for this - though it costs 256 bytes, a table is
3820 much faster than multiple explicit tests. Of course, the standard
3821 character types table is still used for matching digits in subject
3822 strings against \d.
3823
3824 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3825 ending in _t are). So I've renamed it as ESC_tee.
3826
3827 3. The first argument for regexec() in the POSIX wrapper should have been
3828 defined as "const".
3829
3830 4. Changed pcretest to use malloc() for its buffers so that they can be
3831 Electric Fenced for debugging.
3832
3833 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3834 to read one or more bytes before the start of the subject string. Often this
3835 had no effect on PCRE's behaviour, but in some circumstances it could
3836 provoke a segmentation fault.
3837
3838 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3839 to try to read one or more bytes before the start of the subject string.
3840
3841 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3842 UTF-8 support could misbehave in various ways if the subject string
3843 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3844 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3845 back over UTF-8 characters.)
3846
3847
3848 Version 4.2 14-Apr-03
3849 ---------------------
3850
3851 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3852
3853 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3854 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3855 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3856 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3857 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3858 and BUILD_EXEEXT
3859 Note: automatic setting of the BUILD variables is not yet working
3860 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3861 compile-time but not at link-time
3862 [LINK]: use for linking executables only
3863 make different versions for Windows and non-Windows
3864 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3865 libraries
3866 [LINK_FOR_BUILD]: new variable
3867 [OBJEXT]: use throughout
3868 [EXEEXT]: use throughout
3869 <winshared>: new target
3870 <wininstall>: new target
3871 <dftables.o>: use native compiler
3872 <dftables>: use native linker
3873 <install>: handle Windows platform correctly
3874 <clean>: ditto
3875 <check>: ditto
3876 copy DLL to top builddir before testing
3877
3878 As part of these changes, -no-undefined was removed again. This was reported
3879 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3880 in any case.
3881
3882 3. Some tidies to get rid of compiler warnings:
3883
3884 . In the match_data structure, match_limit was an unsigned long int, whereas
3885 match_call_count was an int. I've made them both unsigned long ints.
3886
3887 . In pcretest the fact that a const uschar * doesn't automatically cast to
3888 a void * provoked a warning.
3889
3890 . Turning on some more compiler warnings threw up some "shadow" variables
3891 and a few more missing casts.
3892
3893 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3894 option, a class that contained a single character with a value between 128
3895 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3896
3897 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3898 option, a class that contained several characters, but with at least one
3899 whose value was between 128 and 255 caused PCRE to crash.
3900
3901
3902 Version 4.1 12-Mar-03
3903 ---------------------
3904
3905 1. Compiling with gcc -pedantic found a couple of places where casts were
3906 needed, and a string in dftables.c that was longer than standard compilers are
3907 required to support.
3908
3909 2. Compiling with Sun's compiler found a few more places where the code could
3910 be tidied up in order to avoid warnings.
3911
3912 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3913 first of these names is deprecated in the latest Autoconf in favour of the name
3914 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3915 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3916 analogy I have changed it to CFLAGS_FOR_BUILD.
3917
3918 4. Added -no-undefined to the linking command in the Makefile, because this is
3919 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3920 linking step for the pcreposix library.
3921
3922 5. PCRE was failing to diagnose the case of two named groups with the same
3923 name.
3924
3925 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3926 literal character that is needed in the subject for a match, and scans along to
3927 ensure that it is present before embarking on the full matching process. This
3928 saves time in cases of nested unlimited repeats that are never going to match.
3929 Problem: the scan can take a lot of time if the subject is very long (e.g.
3930 megabytes), thus penalizing straightforward matches. It is now done only if the
3931 amount of subject to be scanned is less than 1000 bytes.
3932
3933 7. A lesser problem with the same optimization is that it was recording the
3934 first character of an anchored pattern as "needed", thus provoking a search
3935 right along the subject, even when the first match of the pattern was going to
3936 fail. The "needed" character is now not set for anchored patterns, unless it
3937 follows something in the pattern that is of non-fixed length. Thus, it still
3938 fulfils its original purpose of finding quick non-matches in cases of nested
3939 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3940
3941
3942 Version 4.0 17-Feb-03
3943 ---------------------
3944
3945 1. If a comment in an extended regex that started immediately after a meta-item
3946 extended to the end of string, PCRE compiled incorrect data. This could lead to
3947 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3948
3949 2. Moved to autoconf 2.53 and libtool 1.4.2.
3950
3951 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3952 the special perltest8 script is no longer needed - all the tests can be run
3953 from a single perltest script.
3954
3955 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3956 by \s. It has now been removed in PCRE. This means it isn't recognized as
3957 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3958 class [:space:] *does* include VT, thereby creating a mess.
3959
3960 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3961 space and tab.
3962
3963 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3964 its new features into the main test script, reducing the number of scripts.
3965
3966 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3967 were backward compatible, and made the (?i) apply to the whole pattern, as if
3968 /i were given. Now it behaves more logically, and applies the option setting
3969 only to what follows. PCRE has been changed to follow suit. However, if it
3970 finds options settings right at the start of the pattern, it extracts them into
3971 the global options, as before. Thus, they show up in the info data.
3972
3973 8. Added support for the \Q...\E escape sequence. Characters in between are
3974 treated as literals. This is slightly different from Perl in that $ and @ are
3975 also handled as literals inside the quotes. In Perl, they will cause variable
3976 interpolation. Note the following examples:
3977
3978 Pattern PCRE matches Perl matches
3979
3980 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3981 \Qabc\$xyz\E abc\$xyz abc\$xyz
3982 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3983
3984 For compatibility with Perl, \Q...\E sequences are recognized inside character
3985 classes as well as outside them.
3986
3987 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3988 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3989 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3990 signed/unsigned warnings.
3991
3992 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3993 option for pcretest, so I've replaced it by a simple function that does just
3994 that job.
3995
3996 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3997 "pcregrep -".
3998
3999 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4000 Java package. This provides some syntactic sugar for simple cases of what my
4001 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4002 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4003 item, you can use this simplified notation. Note that only makes sense with
4004 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4005 greediness, whatever the setting of the PCRE_UNGREEDY option.
4006
4007 13. A change of greediness default within a pattern was not taking effect at
4008 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4009 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4010 was abstracted outside.
4011
4012 14. PCRE now supports the \G assertion. It is true when the current matching
4013 position is at the start point of the match. This differs from \A when the
4014 starting offset is non-zero. Used with the /g option of pcretest (or similar
4015 code), it works in the same way as it does for Perl's /g option. If all
4016 alternatives of a regex begin with \G, the expression is anchored to the start
4017 match position, and the "anchored" flag is set in the compiled expression.
4018
4019 15. Some bugs concerning the handling of certain option changes within patterns
4020 have been fixed. These applied to options other than (?ims). For example,
4021 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4022 been the other way round. Some of this was related to change 7 above.
4023
4024 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4025 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4026 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4027 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4028
4029 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4030 mode. Unlike ".", it always matches newline, whatever the setting of
4031 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4032 assertions. Perl allows it, but it doesn't (in general) work because it can't
4033 calculate the length of the lookbehind. At least, that's the case for Perl
4034 5.8.0 - I've been told they are going to document that it doesn't work in
4035 future.
4036
4037 18. Added an error diagnosis for escapes that PCRE does not support: these are
4038 \L, \l, \N, \P, \p, \U, \u, and \X.
4039
4040 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4041 reading past the end of the pattern in cases such as /[abcd/.
4042
4043 20. PCRE was getting more memory than necessary for patterns with classes that
4044 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4045
4046 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4047 compiling PCRE for use with Virtual Pascal.
4048
4049 22. Small fix to the Makefile to make it work properly if the build is done
4050 outside the source tree.
4051
4052 23. Added a new extension: a condition to go with recursion. If a conditional
4053 subpattern starts with (?(R) the "true" branch is used if recursion has
4054 happened, whereas the "false" branch is used only at the top level.
4055
4056 24. When there was a very long string of literal characters (over 255 bytes
4057 without UTF support, over 250 bytes with UTF support), the computation of how
4058 much memory was required could be incorrect, leading to segfaults or other
4059 strange effects.
4060
4061 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4062 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4063 there was a subsequent back reference to those brackets. This meant that, for
4064 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4065 possible to check for precisely this case. All we can do is abandon the
4066 optimization if .* occurs inside capturing brackets when there are any back
4067 references whatsoever. (See below for a better fix that came later.)
4068
4069 26. The handling of the optimization for finding the first character of a
4070 non-anchored pattern, and for finding a character that is required later in the
4071 match were failing in some cases. This didn't break the matching; it just
4072 failed to optimize when it could. The way this is done has been re-implemented.
4073
4074 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4075
4076 28. Added a new feature that provides some of the functionality that Perl
4077 provides with (?{...}). The facility is termed a "callout". The way it is done
4078 in PCRE is for the caller to provide an optional function, by setting
4079 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4080 global variable. By default it is unset, which disables all calling out. To get
4081 the function called, the regex must include (?C) at appropriate points. This
4082 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4083 This provides a means of identifying different callout points. When PCRE
4084 reaches such a point in the regex, if pcre_callout has been set, the external
4085 function is called. It is provided with data in a structure called
4086 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4087 matching continues; if it returns a non-zero value, the match at the current
4088 point fails. However, backtracking will occur if possible. [This was changed
4089 later and other features added - see item 49 below.]
4090
4091 29. pcretest is upgraded to test the callout functionality. It provides a
4092 callout function that displays information. By default, it shows the start of
4093 the match and the current position in the text. There are some new data escapes
4094 to vary what happens:
4095
4096 \C+ in addition, show current contents of captured substrings
4097 \C- do not supply a callout function
4098 \C!n return 1 when callout number n is reached
4099 \C!n!m return 1 when callout number n is reached for the mth time
4100
4101 30. If pcregrep was called with the -l option and just a single file name, it
4102 output "<stdin>" if a match was found, instead of the file name.
4103
4104 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4105 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4106 pcre_exec(). This saves a malloc/free per call. The default value of
4107 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4108 when configuring.
4109
4110 32. The default maximum size of a compiled pattern is 64K. There have been a
4111 few cases of people hitting this limit. The code now uses macros to handle the
4112 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4113 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4114 configuring. Tests 2 and 5 work only with 2-byte links because they output
4115 debugging information about compiled patterns.
4116
4117 33. Internal code re-arrangements:
4118
4119 (a) Moved the debugging function for printing out a compiled regex into
4120 its own source file (printint.c) and used #include to pull it into
4121 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4122 separate copies.
4123
4124 (b) Defined the list of op-code names for debugging as a macro in
4125 internal.h so that it is next to the definition of the opcodes.
4126
4127 (c) Defined a table of op-code lengths for simpler skipping along compiled
4128 code. This is again a macro in internal.h so that it is next to the
4129 definition of the opcodes.
4130
4131 34. Added support for recursive calls to individual subpatterns, along the
4132 lines of Robin Houston's patch (but implemented somewhat differently).
4133
4134 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4135 allow it to read and process whole directories in Win32. This code was
4136 contributed by Lionel Fourquaux; it has not been tested by me.
4137
4138 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4139 used to name a group. Names consist of alphanumerics and underscores, and must
4140 be unique. Back references use the syntax (?P=name) and recursive calls use
4141 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4142 numbers. The function pcre_fullinfo() can be used after compilation to extract
4143 a name/number map. There are three relevant calls:
4144
4145 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4146 PCRE_INFO_NAMECOUNT yields the number of entries
4147 PCRE_INFO_NAMETABLE yields a pointer to the map.
4148
4149 The map is a vector of fixed-size entries. The size of each entry depends on
4150 the length of the longest name used. The first two bytes of each entry are the
4151 group number, most significant byte first. There follows the corresponding
4152 name, zero terminated. The names are in alphabetical order.
4153
4154 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4155 case instead of 255. Making it the same both with and without UTF-8 support
4156 means that the same test output works with both.
4157
4158 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4159 calling malloc() with a zero argument.
4160
4161 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4162 optimization. I've improved things by keeping a bitmap of backreferences with
4163 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4164 fact referenced, the optimization can be applied. It is unlikely that a
4165 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4166 the match to follow \n) will appear inside brackets with a number greater than
4167 31, but if it does, any back reference > 31 suppresses the optimization.
4168
4169 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4170 of disabling numbered capturing parentheses. Any opening parenthesis that is
4171 not followed by ? behaves as if it were followed by ?: but named parentheses
4172 can still be used for capturing (and they will acquire numbers in the usual
4173 way).
4174
4175 41. Redesigned the return codes from the match() function into yes/no/error so
4176 that errors can be passed back from deep inside the nested calls. A malloc
4177 failure while inside a recursive subpattern call now causes the
4178 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4179
4180 42. It is now possible to set a limit on the number of times the match()
4181 function is called in a call to pcre_exec(). This facility makes it possible to
4182 limit the amount of recursion and backtracking, though not in a directly
4183 obvious way, because the match() function is used in a number of different
4184 circumstances. The count starts from zero for each position in the subject
4185 string (for non-anchored patterns). The default limit is, for compatibility, a
4186 large number, namely 10 000 000. You can change this in two ways:
4187
4188 (a) When configuring PCRE before making, you can use --with-match-limit=n
4189 to set a default value for the compiled library.
4190
4191 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4192 a different value is set. See 45 below.
4193
4194 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4195
4196 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4197 of things that can be changed at compile time. The first argument specifies
4198 what is wanted and the second points to where the information is to be placed.
4199 The current list of available information is:
4200
4201 PCRE_CONFIG_UTF8
4202
4203 The output is an integer that is set to one if UTF-8 support is available;
4204 otherwise it is set to zero.
4205
4206 PCRE_CONFIG_NEWLINE
4207
4208 The output is an integer that it set to the value of the code that is used for
4209 newline. It is either LF (10) or CR (13).
4210
4211 PCRE_CONFIG_LINK_SIZE
4212
4213 The output is an integer that contains the number of bytes used for internal
4214 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4215
4216 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4217
4218 The output is an integer that contains the threshold above which the POSIX
4219 interface uses malloc() for output vectors. See item 31 above.
4220
4221 PCRE_CONFIG_MATCH_LIMIT
4222
4223 The output is an unsigned integer that contains the default limit of the number
4224 of match() calls in a pcre_exec() execution. See 42 above.
4225
4226 44. pcretest has been upgraded by the addition of the -C option. This causes it
4227 to extract all the available output from the new pcre_config() function, and to
4228 output it. The program then exits immediately.
4229
4230 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4231 order to support additional features. One way would have been to define
4232 pcre_exec2() (for example) with extra arguments, but this would not have been
4233 extensible, and would also have required all calls to the original function to
4234 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4235 is used for passing in "extra" data from pcre_study().
4236
4237 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4238 contains the following fields:
4239
4240 flags a bitmap indicating which of the following fields are set
4241 study_data opaque data from pcre_study()
4242 match_limit a way of specifying a limit on match() calls for a specific
4243 call to pcre_exec()
4244 callout_data data for callouts (see 49 below)
4245
4246 The flag bits are also defined in pcre.h, and are
4247
4248 PCRE_EXTRA_STUDY_DATA
4249 PCRE_EXTRA_MATCH_LIMIT
4250 PCRE_EXTRA_CALLOUT_DATA
4251
4252 The pcre_study() function now returns one of these new pcre_extra blocks, with
4253 the actual study data pointed to by the study_data field, and the
4254 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4255 before. That is, this change is entirely upwards-compatible and requires no
4256 change to existing code.
4257
4258 If you want to pass in additional data to pcre_exec(), you can either place it
4259 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4260 block.
4261
4262 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4263 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4264 times with different match limits, until it finds the minimum value needed for
4265 pcre_exec() to complete. The value is then output. This can be instructive; for
4266 most simple matches the number is quite small, but for pathological cases it
4267 gets very large very quickly.
4268
4269 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4270 returns the size of the data block pointed to by the study_data field in a
4271 pcre_extra block, that is, the value that was passed as the argument to
4272 pcre_malloc() when PCRE was getting memory in which to place the information
4273 created by pcre_study(). The fourth argument should point to a size_t variable.
4274 pcretest has been extended so that this information is shown after a successful
4275 pcre_study() call when information about the compiled regex is being displayed.
4276
4277 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4278 because what follows is always an absolute path. (Later: it turns out that this
4279 is more than cosmetic for MinGW, because it doesn't like empty path
4280 components.)
4281
4282 49. Some changes have been made to the callout feature (see 28 above):
4283
4284 (i) A callout function now has three choices for what it returns:
4285
4286 0 => success, carry on matching
4287 > 0 => failure at this point, but backtrack if possible
4288 < 0 => serious error, return this value from pcre_exec()
4289
4290 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4291 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4292 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4293 use by callout functions. It will never be used by PCRE itself.
4294
4295 (ii) The pcre_extra structure (see 45 above) has a void * field called
4296 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4297 pcre_callout_block structure has a field of the same name. The contents of
4298 the field passed in the pcre_extra structure are passed to the callout
4299 function in the corresponding field in the callout block. This makes it
4300 easier to use the same callout-containing regex from multiple threads. For
4301 testing, the pcretest program has a new data escape
4302
4303 \C*n pass the number n (may be negative) as callout_data
4304
4305 If the callout function in pcretest receives a non-zero value as
4306 callout_data, it returns that value.
4307
4308 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4309 there were some redundant $(CFLAGS) in commands that are now specified as
4310 $(LINK), which already includes $(CFLAGS).
4311
4312 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4313 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4314 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4315 one-byte characters throughout. Note that case-insensitive matching applies
4316 only to characters whose values are less than 256. PCRE doesn't support the
4317 notion of cases for higher-valued characters.
4318
4319 (i) A character class whose characters are all within 0-255 is handled as
4320 a bit map, and the map is inverted for negative classes. Previously, a
4321 character > 255 always failed to match such a class; however it should
4322 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4323
4324 (ii) A negated character class with a single character < 255 is coded as
4325 "not this character" (OP_NOT). This wasn't working properly when the test
4326 character was multibyte, either singly or repeated.
4327
4328 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4329 mode, for example: \x{100}{2,3}.
4330
4331 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4332 singly or repeated) now correctly test multibyte characters. However,
4333 PCRE doesn't recognize any characters with values greater than 255 as
4334 digits, spaces, or word characters. Such characters always match \D, \S,
4335 and \W, and never match \d, \s, or \w.
4336
4337 (v) Classes may now contain characters and character ranges with values
4338 greater than 255. For example: [ab\x{100}-\x{400}].
4339
4340 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4341 PCRE in UTF-8 mode.
4342
4343 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4344 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4345 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4346 value.)
4347
4348 53. The single man page has become too large. I have therefore split it up into
4349 a number of separate man pages. These also give rise to individual HTML pages;
4350 these are now put in a separate directory, and there is an index.html page that
4351 lists them all. Some hyperlinking between the pages has been installed.
4352
4353 54. Added convenience functions for handling named capturing parentheses.
4354
4355 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4356 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4357 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4358 are faulted.
4359
4360 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4361 calling configure. These values are used when compiling the dftables.c program
4362 which is run to generate the source of the default character tables. They
4363 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4364 you will need to set these values.
4365
4366 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4367
4368
4369 Version 3.9 02-Jan-02
4370 ---------------------
4371
4372 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4373
4374 2. If --disable-static was given, the building process failed when trying to
4375 build pcretest and pcregrep. (For some reason it was using libtool to compile
4376 them, which is not right, as they aren't part of the library.)
4377
4378
4379 Version 3.8 18-Dec-01
4380 ---------------------
4381
4382 1. The experimental UTF-8 code was completely screwed up. It was packing the
4383 bytes in the wrong order. How dumb can you get?
4384
4385
4386 Version 3.7 29-Oct-01
4387 ---------------------
4388
4389 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4390 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4391 this didn't happen under Solaris 8, where I normally test things.
4392
4393 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4394 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4395 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4396 this fix an hour or so after the initial 3.7 release.)
4397
4398
4399 Version 3.6 23-Oct-01
4400 ---------------------
4401
4402 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4403 offsets passed as NULL with zero offset count.
4404
4405 2. The config.guess and config.sub files had not been updated when I moved to
4406 the latest autoconf.
4407
4408
4409 Version 3.5 15-Aug-01
4410 ---------------------
4411
4412 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4413 had been forgotten.
4414
4415 2. By using declared but undefined structures, we can avoid using "void"
4416 definitions in pcre.h while keeping the internal definitions of the structures
4417 private.
4418
4419 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4420 user point of view, this means that both static and shared libraries are built
4421 by default, but this can be individually controlled. More of the work of
4422 handling this static/shared cases is now inside libtool instead of PCRE's make
4423 file.
4424
4425 4. The pcretest utility is now installed along with pcregrep because it is
4426 useful for users (to test regexs) and by doing this, it automatically gets
4427 relinked by libtool. The documentation has been turned into a man page, so
4428 there are now .1, .txt, and .html versions in /doc.
4429
4430 5. Upgrades to pcregrep:
4431 (i) Added long-form option names like gnu grep.
4432 (ii) Added --help to list all options with an explanatory phrase.
4433 (iii) Added -r, --recursive to recurse into sub-directories.
4434 (iv) Added -f, --file to read patterns from a file.
4435
4436 6. pcre_exec() was referring to its "code" argument before testing that
4437 argument for NULL (and giving an error if it was NULL).
4438
4439 7. Upgraded Makefile.in to allow for compiling in a different directory from
4440 the source directory.
4441
4442 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4443 options bits, the pointer it was passed was to an int instead of to an unsigned
4444 long int. This mattered only on 64-bit systems.
4445
4446 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4447 generated) instead of pcre.in, which it its source. Also made the same change
4448 in several of the .c files.
4449
4450 10. A new release of gcc defines printf() as a macro, which broke pcretest
4451 because it had an ifdef in the middle of a string argument for printf(). Fixed
4452 by using separate calls to printf().
4453
4454 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4455 script, to force use of CR or LF instead of \n in the source. On non-Unix
4456 systems, the value can be set in config.h.
4457
4458 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4459 absolute limit. Changed the text of the error message to make this clear, and
4460 likewise updated the man page.
4461
4462 13. The limit of 99 on the number of capturing subpatterns has been removed.
4463 The new limit is 65535, which I hope will not be a "real" limit.
4464
4465
4466 Version 3.4 22-Aug-00
4467 ---------------------
4468
4469 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4470
4471 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4472
4473
4474 Version 3.3 01-Aug-00
4475 ---------------------
4476
4477 1. If an octal character was given, but the value was greater than \377, it
4478 was not getting masked to the least significant bits, as documented. This could
4479 lead to crashes in some systems.
4480
4481 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4482 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4483
4484 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4485 These just pass their arguments on to (pcre_free)(), but they are provided
4486 because some uses of PCRE bind it to non-C systems that can call its functions,
4487 but cannot call free() or pcre_free() directly.
4488
4489 4. Add "make test" as a synonym for "make check". Corrected some comments in
4490 the Makefile.
4491
4492 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4493 Makefile.
4494
4495 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4496 command called pgrep for grepping around the active processes.
4497
4498 7. Added the beginnings of support for UTF-8 character strings.
4499
4500 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4501 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4502 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4503 out for the ar command.)
4504
4505
4506 Version 3.2 12-May-00
4507 ---------------------
4508
4509 This is purely a bug fixing release.
4510
4511 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4512 of ZA. This was just one example of several cases that could provoke this bug,
4513 which was introduced by change 9 of version 2.00. The code for breaking
4514 infinite loops after an iteration that matches an empty string was't working
4515 correctly.
4516
4517 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4518 when matched against abbab (for example). After matching an empty string, it
4519 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4520 caused it to match further down the string than it should.
4521
4522 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4523 was there because it doesn't seem to be needed, and it causes trouble on some
4524 systems, as it is not a Standard C header. It has been removed.
4525
4526 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4527 were reported on the Macintosh. The changes were from
4528
4529 while ((c = *(++ptr)) != 0 && c != '\n');
4530 to
4531 while ((c = *(++ptr)) != 0 && c != '\n') ;
4532
4533 Totally extraordinary, but if that's what it takes...
4534
4535 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4536 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4537 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4538 assumes the way PCRE uses memmove() (always moving upwards).
4539
4540 6. PCRE is being used in one environment where strchr() is not available. There
4541 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4542 faster code anyway.
4543
4544
4545 Version 3.1 09-Feb-00
4546 ---------------------
4547
4548 The only change in this release is the fixing of some bugs in Makefile.in for
4549 the "install" target:
4550
4551 (1) It was failing to install pcreposix.h.
4552
4553 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4554
4555
4556 Version 3.0 01-Feb-00
4557 ---------------------
4558
4559 1. Add support for the /+ modifier to perltest (to output $` like it does in
4560 pcretest).
4561
4562 2. Add support for the /g modifier to perltest.
4563
4564 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4565 matches null strings.
4566
4567 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4568 pattern. Perl treats empty patterns specially - it reuses the most recent
4569 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4570 effect.
4571
4572 5. The POSIX interface was broken in that it was just handing over the POSIX
4573 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4574 required a bigger vector, with some working space on the end. This means that
4575 the POSIX wrapper now has to get and free some memory, and copy the results.
4576
4577 6. Added some simple autoconf support, placing the test data and the
4578 documentation in separate directories, re-organizing some of the
4579 information files, and making it build pcre-config (a GNU standard). Also added
4580 libtool support for building PCRE as a shared library, which is now the
4581 default.
4582
4583 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4584 09 are not valid octal constants. Single digits will be used for minor values
4585 less than 10.
4586
4587 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4588 existing programs that set these in the POSIX interface can use PCRE without
4589 modification.
4590
4591 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4592 return all that pcre_info() returns, plus additional data. The pcre_info()
4593 function is retained for compatibility, but is considered to be obsolete.
4594
4595 10. Added experimental recursion feature (?R) to handle one common case that
4596 Perl 5.6 will be able to do with (?p{...}).
4597
4598 11. Added support for POSIX character classes like [:alpha:], which Perl is
4599 adopting.
4600
4601
4602 Version 2.08 31-Aug-99
4603 ----------------------
4604
4605 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4606 trying to match at the startoffset position, but instead was moving forward to
4607 the next newline as if a previous match had failed.
4608
4609 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4610 and could get into a loop if a null string was matched other than at the start
4611 of the subject.
4612
4613 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4614 be distinguished at compile time, and for completeness also added PCRE_DATE.
4615
4616 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4617 in GnuWin32 environments.
4618
4619
4620 Version 2.07 29-Jul-99
4621 ----------------------
4622
4623 1. The documentation is now supplied in plain text form and HTML as well as in
4624 the form of man page sources.
4625
4626 2. C++ compilers don't like assigning (void *) values to other pointer types.
4627 In particular this affects malloc(). Although there is no problem in Standard
4628 C, I've put in casts to keep C++ compilers happy.
4629
4630 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4631 should be (const char *).
4632
4633 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4634 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4635 However, I haven't made this a standard facility. The documentation doesn't
4636 mention it, and the Makefile doesn't support it.
4637
4638 5. The Makefile now contains an "install" target, with editable destinations at
4639 the top of the file. The pcretest program is not installed.
4640
4641 6. pgrep -V now gives the PCRE version number and date.
4642
4643 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4644 causing the entire string to be ignored, instead of just the last character.
4645
4646 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4647 non-matching string, it can take a very, very long time, even for strings of
4648 quite modest length, because of the nested recursion. PCRE now does better in
4649 some of these cases. It does this by remembering the last required literal
4650 character in the pattern, and pre-searching the subject to ensure it is present
4651 before running the real match. In other words, it applies a heuristic to detect
4652 some types of certain failure quickly, and in the above example, if presented
4653 with a string that has no trailing " it gives "no match" very quickly.
4654
4655 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4656 other alternatives are tried instead.
4657
4658
4659 Version 2.06 09-Jun-99
4660 ----------------------
4661
4662 1. Change pcretest's output for amount of store used to show just the code
4663 space, because the remainder (the data block) varies in size between 32-bit and
4664 64-bit systems.
4665
4666 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4667 start matching at. This allows lookbehinds to work when searching for multiple
4668 occurrences in a string.
4669
4670 3. Added additional options to pcretest for testing multiple occurrences:
4671
4672 /+ outputs the rest of the string that follows a match
4673 /g loops for multiple occurrences, using the new startoffset argument
4674 /G loops for multiple occurrences by passing an incremented pointer
4675
4676 4. PCRE wasn't doing the "first character" optimization for patterns starting
4677 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4678 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4679 the letter 'x'. On long subject strings, this gives a significant speed-up.
4680
4681
4682 Version 2.05 21-Apr-99
4683 ----------------------
4684
4685 1. Changed the type of magic_number from int to long int so that it works
4686 properly on 16-bit systems.
4687
4688 2. Fixed a bug which caused patterns starting with .* not to work correctly
4689 when the subject string contained newline characters. PCRE was assuming
4690 anchoring for such patterns in all cases, which is not correct because .* will
4691 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4692 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4693 must be retried after every newline in the subject.
4694
4695
4696 Version 2.04 18-Feb-99
4697 ----------------------
4698
4699 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4700 computation of the store needed to hold the pattern was incorrect (too large).
4701 If such patterns were nested a few deep, this could multiply and become a real
4702 problem.
4703
4704 2. Added /M option to pcretest to show the memory requirement of a specific
4705 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4706
4707 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4708 compiled in such a way that the backtracking after subsequent failure was
4709 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4710 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4711
4712
4713 Version 2.03 02-Feb-99
4714 ----------------------
4715
4716 1. Fixed typo and small mistake in man page.
4717
4718 2. Added 4th condition (GPL supersedes if conflict) and created separate
4719 LICENCE file containing the conditions.
4720
4721 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4722 Perl, that is the internal \ allows the delimiter to be included in the
4723 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4724 the final delimiter, add \ to the end of the pattern (to test the error).
4725
4726 4. Added the convenience functions for extracting substrings after a successful
4727 match. Updated pcretest to make it able to test these functions.
4728
4729
4730 Version 2.02 14-Jan-99
4731 ----------------------
4732
4733 1. Initialized the working variables associated with each extraction so that
4734 their saving and restoring doesn't refer to uninitialized store.
4735
4736 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4737 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4738 fix the problem.
4739
4740 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4741 calls, and wasn't printing the correct value for compiling calls. Increased the
4742 default value of LOOPREPEAT, and the number of significant figures in the
4743 times.
4744
4745 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4746
4747 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4748 a building problem on Windows NT with a FAT file system.
4749
4750
4751 Version 2.01 21-Oct-98
4752 ----------------------
4753
4754 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4755 to character tables built by pcre_maketables() in the current locale. If NULL
4756 is passed, the default tables are used.
4757
4758
4759 Version 2.00 24-Sep-98
4760 ----------------------
4761
4762 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4763 it any more.
4764
4765 2. Allow quantification of (?>) groups, and make it work correctly.
4766
4767 3. The first character computation wasn't working for (?>) groups.
4768
4769 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4770 end of the subject) and add 5.005's \z, which really does match only at the
4771 very end of the subject.
4772
4773 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4774
4775 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4776 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4777 localized options. All options to pcre_study() were also removed.
4778
4779 7. Add other new features from 5.005:
4780
4781 $(?<= positive lookbehind
4782 $(?<! negative lookbehind
4783 (?imsx-imsx) added the unsetting capability
4784 such a setting is global if at outer level; local otherwise
4785 (?imsx-imsx:) non-capturing groups with option setting
4786 (?(cond)re|re) conditional pattern matching
4787
4788 A backreference to itself in a repeated group matches the previous
4789 captured string.
4790
4791 8. General tidying up of studying (both automatic and via "study")
4792 consequential on the addition of new assertions.
4793
4794 9. As in 5.005, unlimited repeated groups that could match an empty substring
4795 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4796 runtime if any iteration does actually match an empty substring.
4797
4798 10. Include the RunTest script in the distribution.
4799
4800 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4801 discrepancies, some of which were old and were also with respect to 5.004. They
4802 have now been fixed.
4803
4804
4805 Version 1.09 28-Apr-98
4806 ----------------------
4807
4808 1. A negated single character class followed by a quantifier with a minimum
4809 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4810 program crashes, or just wrong answers. This did not apply to negated classes
4811 containing more than one character, or to minima other than one.
4812
4813
4814 Version 1.08 27-Mar-98
4815 ----------------------
4816
4817 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4818
4819 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4820 latter must appear before anything that relies on it in the pattern.
4821
4822
4823 Version 1.07 16-Feb-98
4824 ----------------------
4825
4826 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4827 repeat of a potentially empty string).
4828
4829
4830 Version 1.06 23-Jan-98
4831 ----------------------
4832
4833 1. Added Markus Oberhumer's little patches for C++.
4834
4835 2. Literal strings longer than 255 characters were broken.
4836
4837
4838 Version 1.05 23-Dec-97
4839 ----------------------
4840
4841 1. Negated character classes containing more than one character were failing if
4842 PCRE_CASELESS was set at run time.
4843
4844
4845 Version 1.04 19-Dec-97
4846 ----------------------
4847
4848 1. Corrected the man page, where some "const" qualifiers had been omitted.
4849
4850 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4851 input syntax.
4852
4853 3. Fixed memory leak which occurred when a regex with back references was
4854 matched with an offsets vector that wasn't big enough. The temporary memory
4855 that is used in this case wasn't being freed if the match failed.
4856
4857 4. Tidied pcretest to ensure it frees memory that it gets.
4858
4859 5. Temporary memory was being obtained in the case where the passed offsets
4860 vector was exactly big enough.
4861
4862 6. Corrected definition of offsetof() from change 5 below.
4863
4864 7. I had screwed up change 6 below and broken the rules for the use of
4865 setjmp(). Now fixed.
4866
4867
4868 Version 1.03 18-Dec-97
4869 ----------------------
4870
4871 1. A erroneous regex with a missing opening parenthesis was correctly
4872 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4873 on some systems.
4874
4875 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4876 it was reported that one broken compiler failed on the former because "code" is
4877 also an independent variable.
4878
4879 3. The erroneous regex a[]b caused an array overrun reference.
4880
4881 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4882 fail on data ending with that character. (It was going on too far, and checking
4883 the next character, typically a binary zero.) This was specific to the
4884 optimized code for single-character negative classes.
4885
4886 5. Added a contributed patch from the TIN world which does the following:
4887
4888 + Add an undef for memmove, in case the the system defines a macro for it.
4889
4890 + Add a definition of offsetof(), in case there isn't one. (I don't know
4891 the reason behind this - offsetof() is part of the ANSI standard - but
4892 it does no harm).
4893
4894 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4895 most of the places where whitespace preceded '#'. I have given up and
4896 allowed the remaining 2 cases to be at the margin.
4897
4898 + Rename some variables in pcre to eliminate shadowing. This seems very
4899 pedantic, but does no harm, of course.
4900
4901 6. Moved the call to setjmp() into its own function, to get rid of warnings
4902 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4903
4904 7. Constructs such as \d{8,} were compiling into the equivalent of
4905 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4906 outcome, but in this particular case used more store than had been allocated,
4907 which caused the bug to be discovered because it threw up an internal error.
4908
4909 8. The debugging code in both pcre and pcretest for outputting the compiled
4910 form of a regex was going wrong in the case of back references followed by
4911 curly-bracketed repeats.
4912
4913
4914 Version 1.02 12-Dec-97
4915 ----------------------
4916
4917 1. Typos in pcre.3 and comments in the source fixed.
4918
4919 2. Applied a contributed patch to get rid of places where it used to remove
4920 'const' from variables, and fixed some signed/unsigned and uninitialized
4921 variable warnings.
4922
4923 3. Added the "runtest" target to Makefile.
4924
4925 4. Set default compiler flag to -O2 rather than just -O.
4926
4927
4928 Version 1.01 19-Nov-97
4929 ----------------------
4930
4931 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4932 like /([ab]*)*/, that is, for classes with more than one character in them.
4933
4934 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4935 as /((?>a*))*/ (a PCRE_EXTRA facility).
4936
4937
4938 Version 1.00 18-Nov-97
4939 ----------------------
4940
4941 1. Added compile-time macros to support systems such as SunOS4 which don't have
4942 memmove() or strerror() but have other things that can be used instead.
4943
4944 2. Arranged that "make clean" removes the executables.
4945
4946
4947 Version 0.99 27-Oct-97
4948 ----------------------
4949
4950 1. Fixed bug in code for optimizing classes with only one character. It was
4951 initializing a 32-byte map regardless, which could cause it to run off the end
4952 of the memory it had got.
4953
4954 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4955
4956
4957 Version 0.98 22-Oct-97
4958 ----------------------
4959
4960 1. Fixed bug in code for handling temporary memory usage when there are more
4961 back references than supplied space in the ovector. This could cause segfaults.
4962
4963
4964 Version 0.97 21-Oct-97
4965 ----------------------
4966
4967 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4968
4969 2. Optimized negated single characters not to use a bit map.
4970
4971 3. Brought error texts together as macro definitions; clarified some of them;
4972 fixed one that was wrong - it said "range out of order" when it meant "invalid
4973 escape sequence".
4974
4975 4. Changed some char * arguments to const char *.
4976
4977 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4978
4979 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4980 pcretest.
4981
4982
4983 Version 0.96 16-Oct-97
4984 ----------------------
4985
4986 1. Added a simple "pgrep" utility to the distribution.
4987
4988 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4989 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4990 where "ddd" means "one or more decimal digits".
4991
4992 3. Fixed serious bug. If a pattern had a back reference, but the call to
4993 pcre_exec() didn't supply a large enough ovector to record the related
4994 identifying subpattern, the match always failed. PCRE now remembers the number
4995 of the largest back reference, and gets some temporary memory in which to save
4996 the offsets during matching if necessary, in order to ensure that
4997 backreferences always work.
4998
4999 4. Increased the compatibility with Perl in a number of ways:
5000
5001 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5002 to request this handling. The option can be set at compile or exec time.
5003
5004 (b) $ matches before a terminating newline by default; an option
5005 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5006 mode). The option can be set at compile or exec time.
5007
5008 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5009 the same as Perl's. If the decimal number it represents is less than 10
5010 or there aren't that many previous left capturing parentheses, an octal
5011 escape is read. Inside a character class, it's always an octal escape,
5012 even if it is a single digit.
5013
5014 (d) An escaped but undefined alphabetic character is taken as a literal,
5015 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5016 escapes.
5017
5018 (e) {0} is now permitted. (The previous item is removed from the compiled
5019 pattern).
5020
5021 5. Changed all the names of code files so that the basic parts are no longer
5022 than 10 characters, and abolished the teeny "globals.c" file.
5023
5024 6. Changed the handling of character classes; they are now done with a 32-byte
5025 bit map always.
5026
5027 7. Added the -d and /D options to pcretest to make it possible to look at the
5028 internals of compilation without having to recompile pcre.
5029
5030
5031 Version 0.95 23-Sep-97
5032 ----------------------
5033
5034 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5035 \x20 at the start of a run of normal characters. These were being treated as
5036 real characters, instead of the source characters being re-checked.
5037
5038
5039 Version 0.94 18-Sep-97
5040 ----------------------
5041
5042 1. The functions are now thread-safe, with the caveat that the global variables
5043 containing pointers to malloc() and free() or alternative functions are the
5044 same for all threads.
5045
5046 2. Get pcre_study() to generate a bitmap of initial characters for non-
5047 anchored patterns when this is possible, and use it if passed to pcre_exec().
5048
5049
5050 Version 0.93 15-Sep-97
5051 ----------------------
5052
5053 1. /(b)|(:+)/ was computing an incorrect first character.
5054
5055 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5056 but not actually doing anything yet.
5057
5058 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5059 as Perl does (e.g. [-az] or [az-]).
5060
5061 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5062 all possible positions.
5063
5064 5. Split up into different modules to avoid including unneeded functions in a
5065 compiled binary. However, compile and exec are still in one module. The "study"
5066 function is split off.
5067
5068 6. The character tables are now in a separate module whose source is generated
5069 by an auxiliary program - but can then be edited by hand if required. There are
5070 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5071 toupper() in the code.
5072
5073 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5074 make them global. Abolish the function for setting them, as the caller can now
5075 set them directly.
5076
5077
5078 Version 0.92 11-Sep-97
5079 ----------------------
5080
5081 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5082 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5083
5084 2. Caseless matching was not working in character classes if the characters in
5085 the pattern were in upper case.
5086
5087 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5088
5089 4. Make PCRE_ANCHORED public and accept as a compile option.
5090
5091 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5092 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5093 pass them.
5094
5095 6. Give an error if bad option bits passed at compile or run time.
5096
5097 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5098 pcretest to cause it to pass that flag.
5099
5100 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5101 options, and the first character, if set.
5102
5103 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5104
5105
5106 Version 0.91 10-Sep-97
5107 ----------------------
5108
5109 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5110 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5111
5112 2. PCRE was looping on encountering an indefinitely repeated back reference to
5113 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5114 Perl does - treats the match as successful.
5115
5116 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5