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

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1313 - (show annotations)
Wed Apr 24 12:07:09 2013 UTC (6 years, 7 months ago) by ph10
File size: 246287 byte(s)
Code (but not yet documentation) for *LIMIT_MATCH and *LIMIT_RECURSION.
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 36. In the interpreter, maximizing pattern repetitions for characters and
139 character types now use tail recursion, which reduces stack usage.
140
141 37. The value of the max lookbehind was not correctly preserved if a compiled
142 and saved regex was reloaded on a host of different endianness.
143
144 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION).
145
146
147 Version 8.32 30-November-2012
148 -----------------------------
149
150 1. Improved JIT compiler optimizations for first character search and single
151 character iterators.
152
153 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
154 Patch by Daniel Richard G.
155
156 3. Single character iterator optimizations in the JIT compiler.
157
158 4. Improved JIT compiler optimizations for character ranges.
159
160 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
161 Reported by Giuseppe D'Angelo.
162
163 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
164 of a line, was being set incorrectly in cases where .* appeared inside
165 atomic brackets at the start of a pattern, or where there was a subsequent
166 *PRUNE or *SKIP.
167
168 7. Improved instruction cache flush for POWER/PowerPC.
169 Patch by Daniel Richard G.
170
171 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
172 grep:
173
174 (a) There is now no limit to the number of patterns to be matched.
175
176 (b) An error is given if a pattern is too long.
177
178 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
179 are now supported.
180
181 (d) --exclude-from and --include-from (multiple use) have been added.
182
183 (e) Exclusions and inclusions now apply to all files and directories, not
184 just to those obtained from scanning a directory recursively.
185
186 (f) Multiple uses of -f and --file-list are now supported.
187
188 (g) In a Windows environment, the default for -d has been changed from
189 "read" (the GNU grep default) to "skip", because otherwise the presence
190 of a directory in the file list provokes an error.
191
192 (h) The documentation has been revised and clarified in places.
193
194 9. Improve the matching speed of capturing brackets.
195
196 10. Changed the meaning of \X so that it now matches a Unicode extended
197 grapheme cluster.
198
199 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
200 out POSIX threads when JIT support is configured.
201
202 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
203
204 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
205 structure could go wrong in environments where size_t is not the same size
206 as int.
207
208 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
209 set.
210
211 15. The EBCDIC support had decayed; later updates to the code had included
212 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
213 general tidy up of EBCDIC-related issues, and the documentation was also
214 not quite right. There is now a test that can be run on ASCII systems to
215 check some of the EBCDIC-related things (but is it not a full test).
216
217 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
218 in a small tidy to the code.
219
220 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
221
222 18. If the --only-matching (-o) option in pcregrep is specified multiple
223 times, each one causes appropriate output. For example, -o1 -o2 outputs the
224 substrings matched by the 1st and 2nd capturing parentheses. A separating
225 string can be specified by --om-separator (default empty).
226
227 19. Improving the first n character searches.
228
229 20. Turn case lists for horizontal and vertical white space into macros so that
230 they are defined only once.
231
232 21. This set of changes together give more compatible Unicode case-folding
233 behaviour for characters that have more than one other case when UCP
234 support is available.
235
236 (a) The Unicode property table now has offsets into a new table of sets of
237 three or more characters that are case-equivalent. The MultiStage2.py
238 script that generates these tables (the pcre_ucd.c file) now scans
239 CaseFolding.txt instead of UnicodeData.txt for character case
240 information.
241
242 (b) The code for adding characters or ranges of characters to a character
243 class has been abstracted into a generalized function that also handles
244 case-independence. In UTF-mode with UCP support, this uses the new data
245 to handle characters with more than one other case.
246
247 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
248 whose other case is greater than 256 are now correctly matched
249 caselessly. Previously, the high codepoint matched the low one, but not
250 vice versa.
251
252 (d) The processing of \h, \H, \v, and \ in character classes now makes use
253 of the new class addition function, using character lists defined as
254 macros alongside the case definitions of 20 above.
255
256 (e) Caseless back references now work with characters that have more than
257 one other case.
258
259 (f) General caseless matching of characters with more than one other case
260 is supported.
261
262 22. Unicode character properties were updated from Unicode 6.2.0
263
264 23. Improved CMake support under Windows. Patch by Daniel Richard G.
265
266 24. Add support for 32-bit character strings, and UTF-32
267
268 25. Major JIT compiler update (code refactoring and bugfixing).
269 Experimental Sparc 32 support is added.
270
271 26. Applied a modified version of Daniel Richard G's patch to create
272 pcre.h.generic and config.h.generic by "make" instead of in the
273 PrepareRelease script.
274
275 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
276 pcre_compile.c when checking for a zero character.
277
278 28. Introducing a native interface for JIT. Through this interface, the compiled
279 machine code can be directly executed. The purpose of this interface is to
280 provide fast pattern matching, so several sanity checks are not performed.
281 However, feature tests are still performed. The new interface provides
282 1.4x speedup compared to the old one.
283
284 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
285 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
286 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
287
288 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
289 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
290 the "old" RFC 2279). Instead, it ended up passing a negative length to
291 pcre_exec().
292
293 31. Add support for GCC's visibility feature to hide internal functions.
294
295 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
296 "unknown -C option" after outputting 0 or 1.
297
298 33. There is now support for generating a code coverage report for the test
299 suite in environments where gcc is the compiler and lcov is installed. This
300 is mainly for the benefit of the developers.
301
302 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
303 unaddressable using valgrind annotations, allowing valgrind to detect
304 invalid memory accesses. This is mainly for the benefit of the developers.
305
306 25. (*UTF) can now be used to start a pattern in any of the three libraries.
307
308 26. Give configure error if --enable-cpp but no C++ compiler found.
309
310
311 Version 8.31 06-July-2012
312 -------------------------
313
314 1. Fixing a wrong JIT test case and some compiler warnings.
315
316 2. Removed a bashism from the RunTest script.
317
318 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
319 to unsigned type, result still unsigned" that was given by an MS compiler
320 on encountering the code "-sizeof(xxx)".
321
322 4. Partial matching support is added to the JIT compiler.
323
324 5. Fixed several bugs concerned with partial matching of items that consist
325 of more than one character:
326
327 (a) /^(..)\1/ did not partially match "aba" because checking references was
328 done on an "all or nothing" basis. This also applied to repeated
329 references.
330
331 (b) \R did not give a hard partial match if \r was found at the end of the
332 subject.
333
334 (c) \X did not give a hard partial match after matching one or more
335 characters at the end of the subject.
336
337 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
338 a partial match for the string "\r".
339
340 (e) When newline was set to CRLF, the metacharacter "." did not recognize
341 a partial match for a CR character at the end of the subject string.
342
343 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
344 running pcretest, the text "(JIT)" added to the output whenever JIT is
345 actually used to run the match.
346
347 7. Individual JIT compile options can be set in pcretest by following -s+[+]
348 or /S+[+] with a digit between 1 and 7.
349
350 8. OP_NOT now supports any UTF character not just single-byte ones.
351
352 9. (*MARK) control verb is now supported by the JIT compiler.
353
354 10. The command "./RunTest list" lists the available tests without actually
355 running any of them. (Because I keep forgetting what they all are.)
356
357 11. Add PCRE_INFO_MAXLOOKBEHIND.
358
359 12. Applied a (slightly modified) user-supplied patch that improves performance
360 when the heap is used for recursion (compiled with --disable-stack-for-
361 recursion). Instead of malloc and free for each heap frame each time a
362 logical recursion happens, frames are retained on a chain and re-used where
363 possible. This sometimes gives as much as 30% improvement.
364
365 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
366 call.
367
368 14. As documented, (*COMMIT) is now confined to within a positive assertion.
369
370 15. It is now possible to link pcretest with libedit as an alternative to
371 libreadline.
372
373 16. (*COMMIT) control verb is now supported by the JIT compiler.
374
375 17. The Unicode data tables have been updated to Unicode 6.1.0.
376
377 18. Added --file-list option to pcregrep.
378
379 19. Added binary file support to pcregrep, including the -a, --binary-files,
380 -I, and --text options.
381
382 20. The madvise function is renamed for posix_madvise for QNX compatibility
383 reasons. Fixed by Giuseppe D'Angelo.
384
385 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
386
387 22. Changed the option for creating symbolic links for 16-bit man pages from
388 -s to -sf so that re-installing does not cause issues.
389
390 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
391
392 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
393 matches in certain environments (the workspace was not being correctly
394 retained). Also added to pcre_dfa_exec() a simple plausibility check on
395 some of the workspace data at the beginning of a restart.
396
397 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
398 was not doing so when it should - probably a typo introduced by SVN 528
399 (change 8.10/14).
400
401 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
402 \w+ when the character tables indicated that \x{c4} was a word character.
403 There were several related cases, all because the tests for doing a table
404 lookup were testing for characters less than 127 instead of 255.
405
406 27. If a pattern contains capturing parentheses that are not used in a match,
407 their slots in the ovector are set to -1. For those that are higher than
408 any matched groups, this happens at the end of processing. In the case when
409 there were back references that the ovector was too small to contain
410 (causing temporary malloc'd memory to be used during matching), and the
411 highest capturing number was not used, memory off the end of the ovector
412 was incorrectly being set to -1. (It was using the size of the temporary
413 memory instead of the true size.)
414
415 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
416 ovector size, it uses memory at the end of the block that it has got.
417
418 29. Check for an overlong MARK name and give an error at compile time. The
419 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
420
421 30. JIT compiler update.
422
423 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
424 Rill for the patch.
425
426 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
427
428 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
429
430 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
431 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
432 was enabled.
433
434 35. Improve JIT code generation for greedy plus quantifier.
435
436 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
437 1 to "aa" instead of to an empty string. The bug affected repeated groups
438 that could potentially match an empty string.
439
440 37. Optimizing single character iterators in JIT.
441
442 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
443 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
444 codepoints that are too big for the mode are faulted, and in a UTF mode,
445 disallowed codepoints are also faulted.
446
447 39. If PCRE was compiled with UTF support, in three places in the DFA
448 matcher there was code that should only have been obeyed in UTF mode, but
449 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
450 processing when bytes with values greater than 127 were present. In 16-bit
451 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
452 both cases the values are those that cannot be the first data item in a UTF
453 character. The three items that might have provoked this were recursions,
454 possessively repeated groups, and atomic groups.
455
456 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
457 and pcregrep, because some OS require shared objects to be explicitly
458 passed to ld, causing the link step to fail if they are not.
459
460 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
461 mode, patterns that started with \h* or \R* might be incorrectly matched.
462
463
464 Version 8.30 04-February-2012
465 -----------------------------
466
467 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
468 name is defined in ctype.h.
469
470 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
471 only in quite long subpatterns.
472
473 3. Removed the function pcre_info(), which has been obsolete and deprecated
474 since it was replaced by pcre_fullinfo() in February 2000.
475
476 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
477 match a (*MARK), and the match failed at the start of the subject, a
478 reference to memory before the start of the subject could occur. This bug
479 was introduced by fix 17 of release 8.21.
480
481 5. A reference to an unset group with zero minimum repetition was giving
482 totally wrong answers (in non-JavaScript-compatibility mode). For example,
483 /(another)?(\1?)test/ matched against "hello world test". This bug was
484 introduced in release 8.13.
485
486 6. Add support for 16-bit character strings (a large amount of work involving
487 many changes and refactorings).
488
489 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
490 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
491 from a file.
492
493 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
494 rounding is not applied in this particular case).
495
496 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
497 if they appear, or are escaped, in patterns.
498
499 10. Get rid of a number of -Wunused-but-set-variable warnings.
500
501 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
502 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
503 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
504 also returns the mark "x". This bug applied to capturing parentheses,
505 non-capturing parentheses, and atomic parentheses. It also applied to some
506 assertions.
507
508 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
509 information out of configure.ac instead of relying on pcre.h.generic, which
510 is not stored in the repository.
511
512 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
513 -lreadline.
514
515 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
516
517 15. Applied Graycode's patch to put the top-level frame on the stack rather
518 than the heap when not using the stack for recursion. This gives a
519 performance improvement in many cases when recursion is not deep.
520
521 16. Experimental code added to "pcretest -C" to output the stack frame size.
522
523
524 Version 8.21 12-Dec-2011
525 ------------------------
526
527 1. Updating the JIT compiler.
528
529 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
530 are added as well.
531
532 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
533 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
534 calling _pcre_jit_exec. Some extra comments are added.
535
536 4. (*MARK) settings inside atomic groups that do not contain any capturing
537 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
538 was introduced by change 18 for 8.20.
539
540 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
541 ECMA-262 standard.
542
543 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
544 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
545 This bug was probably introduced by change 9 of 8.13.
546
547 7. While fixing 6 above, I noticed that a number of other items were being
548 incorrectly rejected as "not fixed length". This arose partly because newer
549 opcodes had not been added to the fixed-length checking code. I have (a)
550 corrected the bug and added tests for these items, and (b) arranged for an
551 error to occur if an unknown opcode is encountered while checking for fixed
552 length instead of just assuming "not fixed length". The items that were
553 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
554 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
555 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
556
557 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
558 being incorrectly compiled and would have given unpredicatble results.
559
560 9. A possessively repeated subpattern with minimum repeat count greater than
561 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
562 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
563 the first (A) could occur when it should not.
564
565 10. Add a cast and remove a redundant test from the code.
566
567 11. JIT should use pcre_malloc/pcre_free for allocation.
568
569 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
570 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
571 is anything other than /usr, -L is still shown).
572
573 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
574
575 14. Perl does not support \N without a following name in a [] class; PCRE now
576 also gives an error.
577
578 15. If a forward reference was repeated with an upper limit of around 2000,
579 it caused the error "internal error: overran compiling workspace". The
580 maximum number of forward references (including repeats) was limited by the
581 internal workspace, and dependent on the LINK_SIZE. The code has been
582 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
583 the default depends on LINK_SIZE. There is a new upper limit (for safety)
584 of around 200,000 forward references. While doing this, I also speeded up
585 the filling in of repeated forward references.
586
587 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
588 incorrectly expecting the subject to contain another "a" after the start.
589
590 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
591 in the match, the SKIP should be ignored. This was not happening; instead
592 the SKIP was being treated as NOMATCH. For patterns such as
593 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
594 tested.
595
596 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
597 now much more compatible with Perl, in particular in cases where the result
598 is a non-match for a non-anchored pattern. For example, if
599 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
600 "m", where previously it did not return a name. A side effect of this
601 change is that for partial matches, the last encountered mark name is
602 returned, as for non matches. A number of tests that were previously not
603 Perl-compatible have been moved into the Perl-compatible test files. The
604 refactoring has had the pleasing side effect of removing one argument from
605 the match() function, thus reducing its stack requirements.
606
607 19. If the /S+ option was used in pcretest to study a pattern using JIT,
608 subsequent uses of /S (without +) incorrectly behaved like /S+.
609
610 21. Retrieve executable code size support for the JIT compiler and fixing
611 some warnings.
612
613 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
614 not work when the shorter character appeared right at the end of the
615 subject string.
616
617 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
618 systems.
619
620 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
621 output it when the /M option is used in pcretest.
622
623 25. The CheckMan script was not being included in the distribution. Also, added
624 an explicit "perl" to run Perl scripts from the PrepareRelease script
625 because this is reportedly needed in Windows.
626
627 26. If study data was being save in a file and studying had not found a set of
628 "starts with" bytes for the pattern, the data written to the file (though
629 never used) was taken from uninitialized memory and so caused valgrind to
630 complain.
631
632 27. Updated RunTest.bat as provided by Sheri Pierce.
633
634 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
635
636 29. Computation of memory usage for the table of capturing group names was
637 giving an unnecessarily large value.
638
639
640 Version 8.20 21-Oct-2011
641 ------------------------
642
643 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
644 a POSIX class. After further experiments with Perl, which convinced me that
645 Perl has bugs and confusions, a closing square bracket is no longer allowed
646 in a POSIX name. This bug also affected patterns with classes that started
647 with full stops.
648
649 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
650 captured substring, but while checking the failing first alternative,
651 substring 1 is temporarily captured. If the output vector supplied to
652 pcre_exec() was not big enough for this capture, the yield of the function
653 was still zero ("insufficient space for captured substrings"). This cannot
654 be totally fixed without adding another stack variable, which seems a lot
655 of expense for a edge case. However, I have improved the situation in cases
656 such as /(a)(b)x|abc/ matched against "abc", where the return code
657 indicates that fewer than the maximum number of slots in the ovector have
658 been set.
659
660 3. Related to (2) above: when there are more back references in a pattern than
661 slots in the output vector, pcre_exec() uses temporary memory during
662 matching, and copies in the captures as far as possible afterwards. It was
663 using the entire output vector, but this conflicts with the specification
664 that only 2/3 is used for passing back captured substrings. Now it uses
665 only the first 2/3, for compatibility. This is, of course, another edge
666 case.
667
668 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
669 main code base, and can be used by building with --enable-jit. When this is
670 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
671 runtime --no-jit option is given.
672
673 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
674 ovector, the return from the function was zero, implying that there were
675 other matches that did not fit. The correct "exactly full" value is now
676 returned.
677
678 6. If a subpattern that was called recursively or as a subroutine contained
679 (*PRUNE) or any other control that caused it to give a non-standard return,
680 invalid errors such as "Error -26 (nested recursion at the same subject
681 position)" or even infinite loops could occur.
682
683 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
684 computing the minimum length on reaching *ACCEPT, and so ended up with the
685 wrong value of 1 rather than 0. Further investigation indicates that
686 computing a minimum subject length in the presence of *ACCEPT is difficult
687 (think back references, subroutine calls), and so I have changed the code
688 so that no minimum is registered for a pattern that contains *ACCEPT.
689
690 8. If (*THEN) was present in the first (true) branch of a conditional group,
691 it was not handled as intended. [But see 16 below.]
692
693 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
694 Sheri Pierce.
695
696 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
697 the first byte in a match must be "a".
698
699 11. Change 17 for 8.13 increased the recursion depth for patterns like
700 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
701 pattern contains any instances of (*THEN). If it does not, the old
702 optimizations are restored. It would be nice to do this on a per-group
703 basis, but at the moment that is not feasible.
704
705 12. In some environments, the output of pcretest -C is CRLF terminated. This
706 broke RunTest's code that checks for the link size. A single white space
707 character after the value is now allowed for.
708
709 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
710 For "fr", it uses the Windows-specific input and output files.
711
712 14. If (*THEN) appeared in a group that was called recursively or as a
713 subroutine, it did not work as intended. [But see next item.]
714
715 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
716 pattern fragments (but not containing any | characters). If A and B are
717 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
718 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
719 D. In other words, Perl considers parentheses that do not contain any |
720 characters to be part of a surrounding alternative, whereas PCRE was
721 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
722 differently. PCRE now behaves in the same way as Perl, except in the case
723 of subroutine/recursion calls such as (?1) which have in any case always
724 been different (but PCRE had them first :-).
725
726 16. Related to 15 above: Perl does not treat the | in a conditional group as
727 creating alternatives. Such a group is treated in the same way as an
728 ordinary group without any | characters when processing (*THEN). PCRE has
729 been changed to match Perl's behaviour.
730
731 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
732 RunGrepTest script failed.
733
734 18. Change 22 for version 13 caused atomic groups to use more stack. This is
735 inevitable for groups that contain captures, but it can lead to a lot of
736 stack use in large patterns. The old behaviour has been restored for atomic
737 groups that do not contain any capturing parentheses.
738
739 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
740 suppress the check for a minimum subject length at run time. (If it was
741 given to pcre_exec() or pcre_dfa_exec() it did work.)
742
743 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
744 fail to work when decoding hex characters in data strings in EBCDIC
745 environments.
746
747 21. It appears that in at least one Mac OS environment, the isxdigit() function
748 is implemented as a macro that evaluates to its argument more than once,
749 contravening the C 90 Standard (I haven't checked a later standard). There
750 was an instance in pcretest which caused it to go wrong when processing
751 \x{...} escapes in subject strings. The has been rewritten to avoid using
752 things like p++ in the argument of isxdigit().
753
754
755 Version 8.13 16-Aug-2011
756 ------------------------
757
758 1. The Unicode data tables have been updated to Unicode 6.0.0.
759
760 2. Two minor typos in pcre_internal.h have been fixed.
761
762 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
763 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
764 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
765
766 4. There were a number of related bugs in the code for matching backrefences
767 caselessly in UTF-8 mode when codes for the characters concerned were
768 different numbers of bytes. For example, U+023A and U+2C65 are an upper
769 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
770 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
771 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
772 2-byte code at the end of the subject (it thought there wasn't enough data
773 left).
774
775 5. Comprehensive information about what went wrong is now returned by
776 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
777 as the output vector has at least 2 elements. The offset of the start of
778 the failing character and a reason code are placed in the vector.
779
780 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
781 now returned is for the first byte of the failing character, instead of the
782 last byte inspected. This is an incompatible change, but I hope it is small
783 enough not to be a problem. It makes the returned offset consistent with
784 pcre_exec() and pcre_dfa_exec().
785
786 7. pcretest now gives a text phrase as well as the error number when
787 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
788 failure, the offset and reason code are output.
789
790 8. When \R was used with a maximizing quantifier it failed to skip backwards
791 over a \r\n pair if the subsequent match failed. Instead, it just skipped
792 back over a single character (\n). This seems wrong (because it treated the
793 two characters as a single entity when going forwards), conflicts with the
794 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
795 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
796 has been changed.
797
798 9. Some internal refactoring has changed the processing so that the handling
799 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
800 time (the PCRE_DOTALL option was changed this way some time ago: version
801 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
802 which was always a bit of a fudge. It also means that there is one less
803 argument for the match() function, which reduces its stack requirements
804 slightly. This change also fixes an incompatibility with Perl: the pattern
805 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
806
807 10. More internal refactoring has drastically reduced the number of recursive
808 calls to match() for possessively repeated groups such as (abc)++ when
809 using pcre_exec().
810
811 11. While implementing 10, a number of bugs in the handling of groups were
812 discovered and fixed:
813
814 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
815 (a|)*(?1) gave a compile-time internal error.
816 ((a|)+)+ did not notice that the outer group could match an empty string.
817 (^a|^)+ was not marked as anchored.
818 (.*a|.*)+ was not marked as matching at start or after a newline.
819
820 12. Yet more internal refactoring has removed another argument from the match()
821 function. Special calls to this function are now indicated by setting a
822 value in a variable in the "match data" data block.
823
824 13. Be more explicit in pcre_study() instead of relying on "default" for
825 opcodes that mean there is no starting character; this means that when new
826 ones are added and accidentally left out of pcre_study(), testing should
827 pick them up.
828
829 14. The -s option of pcretest has been documented for ages as being an old
830 synonym of -m (show memory usage). I have changed it to mean "force study
831 for every regex", that is, assume /S for every regex. This is similar to -i
832 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
833 using it. It makes it easier to run collections of tests with and without
834 study enabled, and thereby test pcre_study() more easily. All the standard
835 tests are now run with and without -s (but some patterns can be marked as
836 "never study" - see 20 below).
837
838 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
839 restoration of the capturing data to the outer values was not happening
840 correctly.
841
842 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
843 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
844 pattern had matched an empty string, and so incorrectly returned a no
845 match.
846
847 17. There was optimizing code for the last branch of non-capturing parentheses,
848 and also for the obeyed branch of a conditional subexpression, which used
849 tail recursion to cut down on stack usage. Unfortunately, now that there is
850 the possibility of (*THEN) occurring in these branches, tail recursion is
851 no longer possible because the return has to be checked for (*THEN). These
852 two optimizations have therefore been removed. [But see 8.20/11 above.]
853
854 18. If a pattern containing \R was studied, it was assumed that \R always
855 matched two bytes, thus causing the minimum subject length to be
856 incorrectly computed because \R can also match just one byte.
857
858 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
859 was incorrectly computed.
860
861 20. If /S is present twice on a test pattern in pcretest input, it now
862 *disables* studying, thereby overriding the use of -s on the command line
863 (see 14 above). This is necessary for one or two tests to keep the output
864 identical in both cases.
865
866 21. When (*ACCEPT) was used in an assertion that matched an empty string and
867 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
868
869 22. When an atomic group that contained a capturing parenthesis was
870 successfully matched, but the branch in which it appeared failed, the
871 capturing was not being forgotten if a higher numbered group was later
872 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
873 group 1 to "a", when in fact it should be unset. This applied to multi-
874 branched capturing and non-capturing groups, repeated or not, and also to
875 positive assertions (capturing in negative assertions does not happen
876 in PCRE) and also to nested atomic groups.
877
878 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
879 subject after a captured substring, to make it easier to tell which of a
880 number of identical substrings has been captured.
881
882 24. The way atomic groups are processed by pcre_exec() has been changed so that
883 if they are repeated, backtracking one repetition now resets captured
884 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
885 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
886 "aaa". Previously, it would have been "a". As part of this code
887 refactoring, the way recursive calls are handled has also been changed.
888
889 25. If an assertion condition captured any substrings, they were not passed
890 back unless some other capturing happened later. For example, if
891 (?(?=(a))a) was matched against "a", no capturing was returned.
892
893 26. When studying a pattern that contained subroutine calls or assertions,
894 the code for finding the minimum length of a possible match was handling
895 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
896 group 1 called group 2 while simultaneously a separate group 2 called group
897 1). A stack overflow occurred in this case. I have fixed this by limiting
898 the recursion depth to 10.
899
900 27. Updated RunTest.bat in the distribution to the version supplied by Tom
901 Fortmann. This supports explicit test numbers on the command line, and has
902 argument validation and error reporting.
903
904 28. An instance of \X with an unlimited repeat could fail if at any point the
905 first character it looked at was a mark character.
906
907 29. Some minor code refactoring concerning Unicode properties and scripts
908 should reduce the stack requirement of match() slightly.
909
910 30. Added the '=' option to pcretest to check the setting of unused capturing
911 slots at the end of the pattern, which are documented as being -1, but are
912 not included in the return count.
913
914 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
915 compiled something random. Now it gives a compile-time error (as does
916 Perl).
917
918 32. A *MARK encountered during the processing of a positive assertion is now
919 recorded and passed back (compatible with Perl).
920
921 33. If --only-matching or --colour was set on a pcregrep call whose pattern
922 had alternative anchored branches, the search for a second match in a line
923 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
924 matched the line "0102" twice. The same bug affected patterns that started
925 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
926 twice.
927
928 34. Previously, PCRE did not allow quantification of assertions. However, Perl
929 does, and because of capturing effects, quantifying parenthesized
930 assertions may at times be useful. Quantifiers are now allowed for
931 parenthesized assertions.
932
933 35. A minor code tidy in pcre_compile() when checking options for \R usage.
934
935 36. \g was being checked for fancy things in a character class, when it should
936 just be a literal "g".
937
938 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
939 appearance of a nested POSIX class supersedes an apparent external class.
940 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
941 unescaped square brackets may also appear as part of class names. For
942 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
943 more like Perl. (But see 8.20/1 above.)
944
945 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
946 was because it thought it was \N{name}, which is not supported).
947
948 39. Add minix to OS list not supporting the -S option in pcretest.
949
950 40. PCRE tries to detect cases of infinite recursion at compile time, but it
951 cannot analyze patterns in sufficient detail to catch mutual recursions
952 such as ((?1))((?2)). There is now a runtime test that gives an error if a
953 subgroup is called recursively as a subpattern for a second time at the
954 same position in the subject string. In previous releases this might have
955 been caught by the recursion limit, or it might have run out of stack.
956
957 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
958 happen only once. PCRE was, however incorrectly giving a compile time error
959 "recursive call could loop indefinitely" because it cannot analyze the
960 pattern in sufficient detail. The compile time test no longer happens when
961 PCRE is compiling a conditional subpattern, but actual runaway loops are
962 now caught at runtime (see 40 above).
963
964 42. It seems that Perl allows any characters other than a closing parenthesis
965 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
966 has been changed to be the same.
967
968 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
969 as not to get warnings when autogen.sh is called. Also changed
970 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
971
972 44. To help people who use pcregrep to scan files containing exceedingly long
973 lines, the following changes have been made:
974
975 (a) The default value of the buffer size parameter has been increased from
976 8K to 20K. (The actual buffer used is three times this size.)
977
978 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
979 PCRE is built.
980
981 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
982 to be set at run time.
983
984 (d) Numerical values in pcregrep options can be followed by K or M, for
985 example --buffer-size=50K.
986
987 (e) If a line being scanned overflows pcregrep's buffer, an error is now
988 given and the return code is set to 2.
989
990 45. Add a pointer to the latest mark to the callout data block.
991
992 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
993 partial match of an empty string instead of no match. This was specific to
994 the use of ".".
995
996 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
997 complete match instead of a partial match. This bug was dependent on both
998 the PCRE_UTF8 and PCRE_DOTALL options being set.
999
1000 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1001 starting byte set, because \b was not being ignored.
1002
1003
1004 Version 8.12 15-Jan-2011
1005 ------------------------
1006
1007 1. Fixed some typos in the markup of the man pages, and wrote a script that
1008 checks for such things as part of the documentation building process.
1009
1010 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1011 --match-limit and --recursion-limit options (added for 8.11). In
1012 particular, this made one of the standard tests fail. (The integer value
1013 went into the wrong half of a long int.)
1014
1015 3. If the --colour option was given to pcregrep with -v (invert match), it
1016 did strange things, either producing crazy output, or crashing. It should,
1017 of course, ignore a request for colour when reporting lines that do not
1018 match.
1019
1020 4. Another pcregrep bug caused similar problems if --colour was specified with
1021 -M (multiline) and the pattern match finished with a line ending.
1022
1023 5. In pcregrep, when a pattern that ended with a literal newline sequence was
1024 matched in multiline mode, the following line was shown as part of the
1025 match. This seems wrong, so I have changed it.
1026
1027 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
1028 the check for further matches in the same line (so they could be coloured)
1029 to overrun the end of the current line. If another match was found, it was
1030 incorrectly shown (and then shown again when found in the next line).
1031
1032 7. If pcregrep was compiled under Windows, there was a reference to the
1033 function pcregrep_exit() before it was defined. I am assuming this was
1034 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
1035 reported by a user. I've moved the definition above the reference.
1036
1037
1038 Version 8.11 10-Dec-2010
1039 ------------------------
1040
1041 1. (*THEN) was not working properly if there were untried alternatives prior
1042 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
1043 backtracked to try for "b" instead of moving to the next alternative branch
1044 at the same level (in this case, to look for "c"). The Perl documentation
1045 is clear that when (*THEN) is backtracked onto, it goes to the "next
1046 alternative in the innermost enclosing group".
1047
1048 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
1049 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
1050 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
1051 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
1052 (*THEN).
1053
1054 3. If \s appeared in a character class, it removed the VT character from
1055 the class, even if it had been included by some previous item, for example
1056 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
1057 of \s, but is part of the POSIX "space" class.)
1058
1059 4. A partial match never returns an empty string (because you can always
1060 match an empty string at the end of the subject); however the checking for
1061 an empty string was starting at the "start of match" point. This has been
1062 changed to the "earliest inspected character" point, because the returned
1063 data for a partial match starts at this character. This means that, for
1064 example, /(?<=abc)def/ gives a partial match for the subject "abc"
1065 (previously it gave "no match").
1066
1067 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
1068 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
1069 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
1070 has an implication that the given string is incomplete (because a partial
1071 match is preferred over a full match). For this reason, these items now
1072 give a partial match in this situation. [Aside: previously, the one case
1073 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
1074 match rather than a full match, which was wrong by the old rules, but is
1075 now correct.]
1076
1077 6. There was a bug in the handling of #-introduced comments, recognized when
1078 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
1079 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
1080 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
1081 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
1082 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
1083 places in pcre_compile().
1084
1085 7. Related to (6) above, when pcre_compile() was skipping #-introduced
1086 comments when looking ahead for named forward references to subpatterns,
1087 the only newline sequence it recognized was NL. It now handles newlines
1088 according to the set newline convention.
1089
1090 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
1091 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
1092 cater for a lack of strerror(). These oversights have been fixed.
1093
1094 9. Added --match-limit and --recursion-limit to pcregrep.
1095
1096 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
1097
1098 11. When the -o option was used, pcregrep was setting a return code of 1, even
1099 when matches were found, and --line-buffered was not being honoured.
1100
1101 12. Added an optional parentheses number to the -o and --only-matching options
1102 of pcregrep.
1103
1104 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
1105 can match an empty string. The code to do it in pcretest and pcredemo
1106 needed fixing:
1107
1108 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
1109 only one byte after an empty string match just before CRLF (this case
1110 just got forgotten; "any" and "anycrlf" were OK).
1111
1112 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
1113 mode when an empty string match preceded an ASCII character followed by
1114 a non-ASCII character. (The code for advancing by one character rather
1115 than one byte was nonsense.)
1116
1117 (c) The pcredemo.c sample program did not have any code at all to handle
1118 the cases when CRLF is a valid newline sequence.
1119
1120 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
1121 as a starting offset was within the subject string. There is now a new
1122 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
1123 negative or greater than the length of the string. In order to test this,
1124 pcretest is extended to allow the setting of negative starting offsets.
1125
1126 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
1127 starting offset points to the beginning of a UTF-8 character was
1128 unnecessarily clumsy. I tidied it up.
1129
1130 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
1131 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
1132
1133 17. Nobody had reported that the --include_dir option, which was added in
1134 release 7.7 should have been called --include-dir (hyphen, not underscore)
1135 for compatibility with GNU grep. I have changed it to --include-dir, but
1136 left --include_dir as an undocumented synonym, and the same for
1137 --exclude-dir, though that is not available in GNU grep, at least as of
1138 release 2.5.4.
1139
1140 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
1141 characters from a string of bytes) have been redefined so as not to use
1142 loops, in order to improve performance in some environments. At the same
1143 time, I abstracted some of the common code into auxiliary macros to save
1144 repetition (this should not affect the compiled code).
1145
1146 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
1147 compile-time error is now given if \c is not followed by an ASCII
1148 character, that is, a byte less than 128. (In EBCDIC mode, the code is
1149 different, and any byte value is allowed.)
1150
1151 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
1152 START_OPTIMIZE option, which is now allowed at compile time - but just
1153 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
1154 to pcregrep and other applications that have no direct access to PCRE
1155 options. The new /Y option in pcretest sets this option when calling
1156 pcre_compile().
1157
1158 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
1159 back references. Groups containing recursive back references were forced to
1160 be atomic by that change, but in the case of named groups, the amount of
1161 memory required was incorrectly computed, leading to "Failed: internal
1162 error: code overflow". This has been fixed.
1163
1164 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
1165 pcretest.c, to avoid build problems in some Borland environments.
1166
1167
1168 Version 8.10 25-Jun-2010
1169 ------------------------
1170
1171 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
1172 THEN.
1173
1174 2. (*ACCEPT) was not working when inside an atomic group.
1175
1176 3. Inside a character class, \B is treated as a literal by default, but
1177 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
1178 causes the error). The code is unchanged, but I tidied the documentation.
1179
1180 4. Inside a character class, PCRE always treated \R and \X as literals,
1181 whereas Perl faults them if its -w option is set. I have changed PCRE so
1182 that it faults them when PCRE_EXTRA is set.
1183
1184 5. Added support for \N, which always matches any character other than
1185 newline. (It is the same as "." when PCRE_DOTALL is not set.)
1186
1187 6. When compiling pcregrep with newer versions of gcc which may have
1188 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
1189 declared with attribute warn_unused_result" were given. Just casting the
1190 result to (void) does not stop the warnings; a more elaborate fudge is
1191 needed. I've used a macro to implement this.
1192
1193 7. Minor change to pcretest.c to avoid a compiler warning.
1194
1195 8. Added four artifical Unicode properties to help with an option to make
1196 \s etc use properties (see next item). The new properties are: Xan
1197 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
1198
1199 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
1200 use Unicode properties. (*UCP) at the start of a pattern can be used to set
1201 this option. Modified pcretest to add /W to test this facility. Added
1202 REG_UCP to make it available via the POSIX interface.
1203
1204 10. Added --line-buffered to pcregrep.
1205
1206 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
1207 studied, and the match started with a letter with a code point greater than
1208 127 whose first byte was different to the first byte of the other case of
1209 the letter, the other case of this starting letter was not recognized
1210 (#976).
1211
1212 12. If a pattern that was studied started with a repeated Unicode property
1213 test, for example, \p{Nd}+, there was the theoretical possibility of
1214 setting up an incorrect bitmap of starting bytes, but fortunately it could
1215 not have actually happened in practice until change 8 above was made (it
1216 added property types that matched character-matching opcodes).
1217
1218 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
1219 possible starting bytes for non-anchored patterns.
1220
1221 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
1222 \R, and also a number of cases that involve Unicode properties, both
1223 explicit and implicit when PCRE_UCP is set.
1224
1225 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
1226 input, it could crash or give wrong results if characters with values
1227 greater than 0xc0 were present in the subject string. (Detail: it assumed
1228 UTF-8 input when processing these items.)
1229
1230 16. Added a lot of (int) casts to avoid compiler warnings in systems where
1231 size_t is 64-bit (#991).
1232
1233 17. Added a check for running out of memory when PCRE is compiled with
1234 --disable-stack-for-recursion (#990).
1235
1236 18. If the last data line in a file for pcretest does not have a newline on
1237 the end, a newline was missing in the output.
1238
1239 19. The default pcre_chartables.c file recognizes only ASCII characters (values
1240 less than 128) in its various bitmaps. However, there is a facility for
1241 generating tables according to the current locale when PCRE is compiled. It
1242 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
1243 characters, are recognized by isspace() and therefore were getting set in
1244 these tables, and indeed these tables seem to approximate to ISO 8859. This
1245 caused a problem in UTF-8 mode when pcre_study() was used to create a list
1246 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
1247 which of course cannot start UTF-8 characters. I have changed the code so
1248 that only real ASCII characters (less than 128) and the correct starting
1249 bytes for UTF-8 encodings are set for characters greater than 127 when in
1250 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
1251 altogether.)
1252
1253 20. Added the /T option to pcretest so as to be able to run tests with non-
1254 standard character tables, thus making it possible to include the tests
1255 used for 19 above in the standard set of tests.
1256
1257 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
1258 reference to a subpattern the other side of a comment that contains an
1259 opening parenthesis caused either an internal compiling error, or a
1260 reference to the wrong subpattern.
1261
1262
1263 Version 8.02 19-Mar-2010
1264 ------------------------
1265
1266 1. The Unicode data tables have been updated to Unicode 5.2.0.
1267
1268 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
1269 configured.
1270
1271 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
1272 original author of that file, following a query about its status.
1273
1274 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
1275 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
1276
1277 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
1278 quantifier applied to a forward-referencing subroutine call, could compile
1279 incorrect code or give the error "internal error: previously-checked
1280 referenced subpattern not found".
1281
1282 6. Both MS Visual Studio and Symbian OS have problems with initializing
1283 variables to point to external functions. For these systems, therefore,
1284 pcre_malloc etc. are now initialized to local functions that call the
1285 relevant global functions.
1286
1287 7. There were two entries missing in the vectors called coptable and poptable
1288 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
1289 I've fixed the data, and added a kludgy way of testing at compile time that
1290 the lengths are correct (equal to the number of opcodes).
1291
1292 8. Following on from 7, I added a similar kludge to check the length of the
1293 eint vector in pcreposix.c.
1294
1295 9. Error texts for pcre_compile() are held as one long string to avoid too
1296 much relocation at load time. To find a text, the string is searched,
1297 counting zeros. There was no check for running off the end of the string,
1298 which could happen if a new error number was added without updating the
1299 string.
1300
1301 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
1302
1303 11. \K was not working if it appeared in an atomic group or in a group that
1304 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
1305 \K is "not well defined" if used in an assertion. PCRE now accepts it if
1306 the assertion is positive, but not if it is negative.
1307
1308 12. Change 11 fortuitously reduced the size of the stack frame used in the
1309 "match()" function of pcre_exec.c by one pointer. Forthcoming
1310 implementation of support for (*MARK) will need an extra pointer on the
1311 stack; I have reserved it now, so that the stack frame size does not
1312 decrease.
1313
1314 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
1315 item in branch that calls a recursion is a subroutine call - as in the
1316 second branch in the above example - was incorrectly given the compile-
1317 time error "recursive call could loop indefinitely" because pcre_compile()
1318 was not correctly checking the subroutine for matching a non-empty string.
1319
1320 14. The checks for overrunning compiling workspace could trigger after an
1321 overrun had occurred. This is a "should never occur" error, but it can be
1322 triggered by pathological patterns such as hundreds of nested parentheses.
1323 The checks now trigger 100 bytes before the end of the workspace.
1324
1325 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
1326
1327
1328 Version 8.01 19-Jan-2010
1329 ------------------------
1330
1331 1. If a pattern contained a conditional subpattern with only one branch (in
1332 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
1333 computed the wrong minimum data length (which is of course zero for such
1334 subpatterns). This could cause incorrect "no match" results.
1335
1336 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
1337 the pattern is reset in the first branch, pcre_compile() failed with
1338 "internal error: code overflow at offset...". This happened only when
1339 the reset was to the original external option setting. (An optimization
1340 abstracts leading options settings into an external setting, which was the
1341 cause of this.)
1342
1343 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
1344 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
1345 assertion pattern did not match (meaning that the assertion was true), it
1346 was incorrectly treated as false if the SKIP had been reached during the
1347 matching. This also applied to assertions used as conditions.
1348
1349 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
1350 assertion subpattern, including such a pattern used as a condition,
1351 unpredictable results occurred, instead of the error return
1352 PCRE_ERROR_DFA_UITEM.
1353
1354 5. The C++ GlobalReplace function was not working like Perl for the special
1355 situation when an empty string is matched. It now does the fancy magic
1356 stuff that is necessary.
1357
1358 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
1359 removed. (These were left over from very, very early versions of PCRE.)
1360
1361 7. Some cosmetic changes to the code to make life easier when compiling it
1362 as part of something else:
1363
1364 (a) Change DEBUG to PCRE_DEBUG.
1365
1366 (b) In pcre_compile(), rename the member of the "branch_chain" structure
1367 called "current" as "current_branch", to prevent a collision with the
1368 Linux macro when compiled as a kernel module.
1369
1370 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
1371 prevent a collision with the Linux macro when compiled as a kernel
1372 module.
1373
1374 8. In pcre_compile() there are some checks for integer overflows that used to
1375 cast potentially large values to (double). This has been changed to that
1376 when building, a check for int64_t is made, and if it is found, it is used
1377 instead, thus avoiding the use of floating point arithmetic. (There is no
1378 other use of FP in PCRE.) If int64_t is not found, the fallback is to
1379 double.
1380
1381 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
1382 2005 (difference between two addresses compared to an unsigned value).
1383
1384 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
1385 custom one, because of the following reported problem in Windows:
1386
1387 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
1388 under Win32.
1389 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
1390 therefore missing the function definition.
1391 - The compiler thus generates a "C" signature for the test function.
1392 - The linker fails to find the "C" function.
1393 - PCRE fails to configure if asked to do so against libbz2.
1394
1395 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
1396 messages were output:
1397
1398 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
1399 rerunning libtoolize, to keep the correct libtool macros in-tree.
1400 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
1401
1402 I have done both of these things.
1403
1404 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
1405 most of the time, it *can* run out if it is given a pattern that contains a
1406 runaway infinite recursion. I updated the discussion in the pcrestack man
1407 page.
1408
1409 13. Now that we have gone to the x.xx style of version numbers, the minor
1410 version may start with zero. Using 08 or 09 is a bad idea because users
1411 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
1412 interpreted as invalid octal numbers. I've updated the previous comment in
1413 configure.ac, and also added a check that gives an error if 08 or 09 are
1414 used.
1415
1416 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
1417 causing partial matching to fail when the end of the subject matched \W
1418 in a UTF-8 pattern where \W was quantified with a minimum of 3.
1419
1420 15. There were some discrepancies between the declarations in pcre_internal.h
1421 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
1422 their definitions. The declarations used "const uschar *" and the
1423 definitions used USPTR. Even though USPTR is normally defined as "const
1424 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
1425 reported that: "This difference in casting confuses some C++ compilers, for
1426 example, SunCC recognizes above declarations as different functions and
1427 generates broken code for hbpcre." I have changed the declarations to use
1428 USPTR.
1429
1430 16. GNU libtool is named differently on some systems. The autogen.sh script now
1431 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
1432 (FreeBSD).
1433
1434 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
1435 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
1436 comment: "Figure out how to create a longlong from a string: strtoll and
1437 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
1438 instance, but it only takes 2 args instead of 3!"
1439
1440 18. A subtle bug concerned with back references has been fixed by a change of
1441 specification, with a corresponding code fix. A pattern such as
1442 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
1443 refers, was giving matches when it shouldn't. For example, xa=xaaa would
1444 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
1445 same bug. Such groups have to be quantified to be useful, or contained
1446 inside another quantified group. (If there's no repetition, the reference
1447 can never match.) The problem arises because, having left the group and
1448 moved on to the rest of the pattern, a later failure that backtracks into
1449 the group uses the captured value from the final iteration of the group
1450 rather than the correct earlier one. I have fixed this in PCRE by forcing
1451 any group that contains a reference to itself to be an atomic group; that
1452 is, there cannot be any backtracking into it once it has completed. This is
1453 similar to recursive and subroutine calls.
1454
1455
1456 Version 8.00 19-Oct-09
1457 ----------------------
1458
1459 1. The table for translating pcre_compile() error codes into POSIX error codes
1460 was out-of-date, and there was no check on the pcre_compile() error code
1461 being within the table. This could lead to an OK return being given in
1462 error.
1463
1464 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
1465 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
1466 in a Windows environment.
1467
1468 3. The pcregrep --count option prints the count for each file even when it is
1469 zero, as does GNU grep. However, pcregrep was also printing all files when
1470 --files-with-matches was added. Now, when both options are given, it prints
1471 counts only for those files that have at least one match. (GNU grep just
1472 prints the file name in this circumstance, but including the count seems
1473 more useful - otherwise, why use --count?) Also ensured that the
1474 combination -clh just lists non-zero counts, with no names.
1475
1476 4. The long form of the pcregrep -F option was incorrectly implemented as
1477 --fixed_strings instead of --fixed-strings. This is an incompatible change,
1478 but it seems right to fix it, and I didn't think it was worth preserving
1479 the old behaviour.
1480
1481 5. The command line items --regex=pattern and --regexp=pattern were not
1482 recognized by pcregrep, which required --regex pattern or --regexp pattern
1483 (with a space rather than an '='). The man page documented the '=' forms,
1484 which are compatible with GNU grep; these now work.
1485
1486 6. No libpcreposix.pc file was created for pkg-config; there was just
1487 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
1488
1489 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
1490 when UCP support is not needed, by modifying the Python script that
1491 generates it from Unicode data files. This should not matter if the module
1492 is correctly used as a library, but I received one complaint about 50K of
1493 unwanted data. My guess is that the person linked everything into his
1494 program rather than using a library. Anyway, it does no harm.
1495
1496 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
1497 was a minimum greater than 1 for a wide character in a possessive
1498 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
1499 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
1500 character. Chaos in the form of incorrect output or a compiling loop could
1501 result.
1502
1503 9. The restrictions on what a pattern can contain when partial matching is
1504 requested for pcre_exec() have been removed. All patterns can now be
1505 partially matched by this function. In addition, if there are at least two
1506 slots in the offset vector, the offset of the earliest inspected character
1507 for the match and the offset of the end of the subject are set in them when
1508 PCRE_ERROR_PARTIAL is returned.
1509
1510 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
1511 synonymous with PCRE_PARTIAL, for backwards compatibility, and
1512 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
1513 and may be more useful for multi-segment matching.
1514
1515 11. Partial matching with pcre_exec() is now more intuitive. A partial match
1516 used to be given if ever the end of the subject was reached; now it is
1517 given only if matching could not proceed because another character was
1518 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
1519 string "Z", which now yields "no match" instead of "partial match". In the
1520 case of pcre_dfa_exec(), "no match" is given if every matching path for the
1521 final character ended with (*FAIL).
1522
1523 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
1524 if the pattern had a "must contain" character that was already found in the
1525 earlier partial match, unless partial matching was again requested. For
1526 example, with the pattern /dog.(body)?/, the "must contain" character is
1527 "g". If the first part-match was for the string "dog", restarting with
1528 "sbody" failed. This bug has been fixed.
1529
1530 13. The string returned by pcre_dfa_exec() after a partial match has been
1531 changed so that it starts at the first inspected character rather than the
1532 first character of the match. This makes a difference only if the pattern
1533 starts with a lookbehind assertion or \b or \B (\K is not supported by
1534 pcre_dfa_exec()). It's an incompatible change, but it makes the two
1535 matching functions compatible, and I think it's the right thing to do.
1536
1537 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
1538 so that the demonstration program is easily available in environments where
1539 PCRE has not been installed from source.
1540
1541 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
1542 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
1543 library.
1544
1545 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
1546 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
1547 is not the first non-POSIX option to be added. Clearly some people find
1548 these options useful.
1549
1550 17. If a caller to the POSIX matching function regexec() passes a non-zero
1551 value for nmatch with a NULL value for pmatch, the value of
1552 nmatch is forced to zero.
1553
1554 18. RunGrepTest did not have a test for the availability of the -u option of
1555 the diff command, as RunTest does. It now checks in the same way as
1556 RunTest, and also checks for the -b option.
1557
1558 19. If an odd number of negated classes containing just a single character
1559 interposed, within parentheses, between a forward reference to a named
1560 subpattern and the definition of the subpattern, compilation crashed with
1561 an internal error, complaining that it could not find the referenced
1562 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
1563 [The bug was that it was starting one character too far in when skipping
1564 over the character class, thus treating the ] as data rather than
1565 terminating the class. This meant it could skip too much.]
1566
1567 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
1568 /g option in pcretest when the pattern contains \K, which makes it possible
1569 to have an empty string match not at the start, even when the pattern is
1570 anchored. Updated pcretest and pcredemo to use this option.
1571
1572 21. If the maximum number of capturing subpatterns in a recursion was greater
1573 than the maximum at the outer level, the higher number was returned, but
1574 with unset values at the outer level. The correct (outer level) value is
1575 now given.
1576
1577 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
1578 PCRE did not set those parentheses (unlike Perl). I have now found a way to
1579 make it do so. The string so far is captured, making this feature
1580 compatible with Perl.
1581
1582 23. The tests have been re-organized, adding tests 11 and 12, to make it
1583 possible to check the Perl 5.10 features against Perl 5.10.
1584
1585 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
1586 pattern matches a fixed length string. PCRE did not allow this; now it
1587 does. Neither allows recursion.
1588
1589 25. I finally figured out how to implement a request to provide the minimum
1590 length of subject string that was needed in order to match a given pattern.
1591 (It was back references and recursion that I had previously got hung up
1592 on.) This code has now been added to pcre_study(); it finds a lower bound
1593 to the length of subject needed. It is not necessarily the greatest lower
1594 bound, but using it to avoid searching strings that are too short does give
1595 some useful speed-ups. The value is available to calling programs via
1596 pcre_fullinfo().
1597
1598 26. While implementing 25, I discovered to my embarrassment that pcretest had
1599 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
1600 study optimizations had never been tested with that matching function.
1601 Oops. What is worse, even when it was passed study data, there was a bug in
1602 pcre_dfa_exec() that meant it never actually used it. Double oops. There
1603 were also very few tests of studied patterns with pcre_dfa_exec().
1604
1605 27. If (?| is used to create subpatterns with duplicate numbers, they are now
1606 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
1607 on the other side of the coin, they are no longer allowed to have different
1608 names, because these cannot be distinguished in PCRE, and this has caused
1609 confusion. (This is a difference from Perl.)
1610
1611 28. When duplicate subpattern names are present (necessarily with different
1612 numbers, as required by 27 above), and a test is made by name in a
1613 conditional pattern, either for a subpattern having been matched, or for
1614 recursion in such a pattern, all the associated numbered subpatterns are
1615 tested, and the overall condition is true if the condition is true for any
1616 one of them. This is the way Perl works, and is also more like the way
1617 testing by number works.
1618
1619
1620 Version 7.9 11-Apr-09
1621 ---------------------
1622
1623 1. When building with support for bzlib/zlib (pcregrep) and/or readline
1624 (pcretest), all targets were linked against these libraries. This included
1625 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
1626 libraries. This caused unwanted dependencies to be created. This problem
1627 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
1628 pcretest is linked with readline.
1629
1630 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
1631 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
1632 moved outside it again, because FALSE and TRUE are already defined in AIX,
1633 but BOOL is not.
1634
1635 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
1636 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
1637
1638 4. The pcregrep documentation said spaces were inserted as well as colons (or
1639 hyphens) following file names and line numbers when outputting matching
1640 lines. This is not true; no spaces are inserted. I have also clarified the
1641 wording for the --colour (or --color) option.
1642
1643 5. In pcregrep, when --colour was used with -o, the list of matching strings
1644 was not coloured; this is different to GNU grep, so I have changed it to be
1645 the same.
1646
1647 6. When --colo(u)r was used in pcregrep, only the first matching substring in
1648 each matching line was coloured. Now it goes on to look for further matches
1649 of any of the test patterns, which is the same behaviour as GNU grep.
1650
1651 7. A pattern that could match an empty string could cause pcregrep to loop; it
1652 doesn't make sense to accept an empty string match in pcregrep, so I have
1653 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
1654 seems to be how GNU grep behaves.
1655
1656 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
1657 start or after a newline", because the conditional assertion was not being
1658 correctly handled. The rule now is that both the assertion and what follows
1659 in the first alternative must satisfy the test.
1660
1661 9. If auto-callout was enabled in a pattern with a conditional group whose
1662 condition was an assertion, PCRE could crash during matching, both with
1663 pcre_exec() and pcre_dfa_exec().
1664
1665 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
1666 used for matching.
1667
1668 11. Unicode property support in character classes was not working for
1669 characters (bytes) greater than 127 when not in UTF-8 mode.
1670
1671 12. Added the -M command line option to pcretest.
1672
1673 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
1674
1675 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
1676
1677 16. Added comments and documentation about mis-use of no_arg in the C++
1678 wrapper.
1679
1680 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
1681 from Martin Jerabek that uses macro names for all relevant character and
1682 string constants.
1683
1684 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
1685 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
1686 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
1687 these, but not everybody uses configure.
1688
1689 19. A conditional group that had only one branch was not being correctly
1690 recognized as an item that could match an empty string. This meant that an
1691 enclosing group might also not be so recognized, causing infinite looping
1692 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
1693 with the subject "ab", where knowledge that the repeated group can match
1694 nothing is needed in order to break the loop.
1695
1696 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
1697 exec(), but without supplying a callout function, matching went wrong.
1698
1699 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
1700 leak if the size of the offset vector was greater than 30. When the vector
1701 is smaller, the saved offsets during recursion go onto a local stack
1702 vector, but for larger vectors malloc() is used. It was failing to free
1703 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1704 error, in fact).
1705
1706 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1707 heapframe that is used only when UTF-8 support is enabled. This caused no
1708 problem, but was untidy.
1709
1710 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1711 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1712 included within another project.
1713
1714 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1715 slightly modified by me:
1716
1717 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1718 not building pcregrep.
1719
1720 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1721 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1722
1723 25. Forward references, both numeric and by name, in patterns that made use of
1724 duplicate group numbers, could behave incorrectly or give incorrect errors,
1725 because when scanning forward to find the reference group, PCRE was not
1726 taking into account the duplicate group numbers. A pattern such as
1727 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1728
1729 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1730 the feature of a custom pointer more persuasive (as requested by a user).
1731
1732 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1733 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1734 pre-defined.
1735
1736 28. Added support for (*UTF8) at the start of a pattern.
1737
1738 29. Arrange for flags added by the "release type" setting in CMake to be shown
1739 in the configuration summary.
1740
1741
1742 Version 7.8 05-Sep-08
1743 ---------------------
1744
1745 1. Replaced UCP searching code with optimized version as implemented for Ad
1746 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1747 stage table and inline lookup instead of a function, giving speed ups of 2
1748 to 5 times on some simple patterns that I tested. Permission was given to
1749 distribute the MultiStage2.py script that generates the tables (it's not in
1750 the tarball, but is in the Subversion repository).
1751
1752 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1753 scripts.
1754
1755 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1756 a group with a zero qualifier. The result of the study could be incorrect,
1757 or the function might crash, depending on the pattern.
1758
1759 4. Caseless matching was not working for non-ASCII characters in back
1760 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1761 It now works when Unicode Property Support is available.
1762
1763 5. In pcretest, an escape such as \x{de} in the data was always generating
1764 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1765 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1766 truncation.
1767
1768 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1769
1770 7. Added two (int) casts to pcregrep when printing the difference of two
1771 pointers, in case they are 64-bit values.
1772
1773 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1774 test 2 if it fails.
1775
1776 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1777 and a #define of that name to empty if it is not externally set. This is to
1778 allow users of MSVC to set it if necessary.
1779
1780 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1781 the convenience functions in the pcre_get.c source file.
1782
1783 11. An option change at the start of a pattern that had top-level alternatives
1784 could cause overwriting and/or a crash. This command provoked a crash in
1785 some environments:
1786
1787 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1788
1789 This potential security problem was recorded as CVE-2008-2371.
1790
1791 12. For a pattern where the match had to start at the beginning or immediately
1792 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1793 pcre_dfa_exec() could read past the end of the passed subject if there was
1794 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1795 pcretest so that it places the subject at the end of its malloc-ed buffer.
1796
1797 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1798 exec() might read past the end of the data buffer in UTF-8 mode.
1799
1800 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1801 the data contained the byte 0x85 as part of a UTF-8 character within its
1802 first line. This applied both to normal and DFA matching.
1803
1804 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1805 /^[^d]*?$/8 failed to match "abc".
1806
1807 16. Added a missing copyright notice to pcrecpp_internal.h.
1808
1809 17. Make it more clear in the documentation that values returned from
1810 pcre_exec() in ovector are byte offsets, not character counts.
1811
1812 18. Tidied a few places to stop certain compilers from issuing warnings.
1813
1814 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1815 supplied by Stefan Weber. I made a further small update for 7.8 because
1816 there is a change of source arrangements: the pcre_searchfuncs.c module is
1817 replaced by pcre_ucd.c.
1818
1819
1820 Version 7.7 07-May-08
1821 ---------------------
1822
1823 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1824 a string to a long long, pretend we don't even have a long long." This is
1825 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1826
1827 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1828 pre-7.6 versions, which defined a global no_arg variable instead of putting
1829 it in the RE class. (See also #8 below.)
1830
1831 3. Remove a line of dead code, identified by coverity and reported by Nuno
1832 Lopes.
1833
1834 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1835
1836 (1) The include/exclude patterns were being applied to the whole pathnames
1837 of files, instead of just to the final components.
1838
1839 (2) If there was more than one level of directory, the subdirectories were
1840 skipped unless they satisfied the include/exclude conditions. This is
1841 inconsistent with GNU grep (and could even be seen as contrary to the
1842 pcregrep specification - which I improved to make it absolutely clear).
1843 The action now is always to scan all levels of directory, and just
1844 apply the include/exclude patterns to regular files.
1845
1846 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1847 --exclude_dir in the tests to avoid scanning .svn directories.
1848
1849 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1850 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1851 doesn't support NULs in patterns.
1852
1853 7. Added some missing "const"s to declarations of static tables in
1854 pcre_compile.c and pcre_dfa_exec.c.
1855
1856 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1857 caused by fix #2 above. (Subsequently also a second patch to fix the
1858 first patch. And a third patch - this was a messy problem.)
1859
1860 9. Applied Craig's patch to remove the use of push_back().
1861
1862 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1863 matching function regexec().
1864
1865 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1866 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1867 references. PCRE supports relative numbers with this syntax (I don't think
1868 Oniguruma does).
1869
1870 12. Previously, a group with a zero repeat such as (...){0} was completely
1871 omitted from the compiled regex. However, this means that if the group
1872 was called as a subroutine from elsewhere in the pattern, things went wrong
1873 (an internal error was given). Such groups are now left in the compiled
1874 pattern, with a new opcode that causes them to be skipped at execution
1875 time.
1876
1877 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1878 to the way PCRE behaves:
1879
1880 (a) A lone ] character is dis-allowed (Perl treats it as data).
1881
1882 (b) A back reference to an unmatched subpattern matches an empty string
1883 (Perl fails the current match path).
1884
1885 (c) A data ] in a character class must be notated as \] because if the
1886 first data character in a class is ], it defines an empty class. (In
1887 Perl it is not possible to have an empty class.) The empty class []
1888 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1889 The negative empty class [^] matches any one character, independently
1890 of the DOTALL setting.
1891
1892 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1893 non-existent subpattern following a character class starting with ']' and
1894 containing () gave an internal compiling error instead of "reference to
1895 non-existent subpattern". Fortunately, when the pattern did exist, the
1896 compiled code was correct. (When scanning forwards to check for the
1897 existencd of the subpattern, it was treating the data ']' as terminating
1898 the class, so got the count wrong. When actually compiling, the reference
1899 was subsequently set up correctly.)
1900
1901 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1902 it was being rejected as not supported by pcre_dfa_exec(), even though
1903 other assertions are supported. I have made pcre_dfa_exec() support
1904 (*FAIL).
1905
1906 16. The implementation of 13c above involved the invention of a new opcode,
1907 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1908 cannot be changed at match time, I realized I could make a small
1909 improvement to matching performance by compiling OP_ALLANY instead of
1910 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1911 on the OP_ANY path.
1912
1913 17. Compiling pcretest on Windows with readline support failed without the
1914 following two fixes: (1) Make the unistd.h include conditional on
1915 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1916
1917 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1918 ncurses library to be included for pcretest when ReadLine support is
1919 requested, but also to allow for it to be overridden. This patch came from
1920 Daniel Bergström.
1921
1922 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1923 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1924 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1925 spotting this.
1926
1927
1928 Version 7.6 28-Jan-08
1929 ---------------------
1930
1931 1. A character class containing a very large number of characters with
1932 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1933 overflow.
1934
1935 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1936 HAVE_LONG_LONG is not defined.
1937
1938 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1939 bring them up to date and include new features. This patch includes:
1940
1941 - Fixed PH's badly added libz and libbz2 support.
1942 - Fixed a problem with static linking.
1943 - Added pcredemo. [But later removed - see 7 below.]
1944 - Fixed dftables problem and added an option.
1945 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1946 HAVE_LONG_LONG.
1947 - Added readline support for pcretest.
1948 - Added an listing of the option settings after cmake has run.
1949
1950 4. A user submitted a patch to Makefile that makes it easy to create
1951 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1952 Makefile.am that cause it to include this special target, without
1953 affecting anything else. Note that the same mingw target plus all
1954 the other distribution libraries and programs are now supported
1955 when configuring with CMake (see 6 below) instead of with
1956 Configure/Make.
1957
1958 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1959 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1960 exported in the Windows port". It has not yet been confirmed that the patch
1961 solves the problem, but it does no harm.
1962
1963 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1964 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1965 with CMake, and also correct the comment about stack recursion.
1966
1967 7. Remove the automatic building of pcredemo from the ./configure system and
1968 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1969 of a program that users should build themselves after PCRE is installed, so
1970 building it automatically is not really right. What is more, it gave
1971 trouble in some build environments.
1972
1973 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1974
1975
1976 Version 7.5 10-Jan-08
1977 ---------------------
1978
1979 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1980 values in parens when parsing an RE using the C++ wrapper."
1981
1982 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1983 Characters greater than 255 were excluded from the class instead of being
1984 included.
1985
1986 3. The same bug as (2) above applied to negated POSIX classes such as
1987 [:^space:].
1988
1989 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1990 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1991 I have changed it.
1992
1993 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1994 first named subpattern) and a construct such as (?&a) would reference the
1995 first named subpattern whose name started with "a" (in other words, the
1996 length check was missing). Both these problems are fixed. "Subpattern name
1997 expected" is now given for (?&) (a zero-length name), and this patch also
1998 makes it give the same error for \k'' (previously it complained that that
1999 was a reference to a non-existent subpattern).
2000
2001 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2002 this is right because (?- can be followed by option settings as well as by
2003 digits. I have, however, made the messages clearer.
2004
2005 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2006 than the number used in the conditional) now cause a compile-time error.
2007 This is actually not compatible with Perl, which accepts such patterns, but
2008 treats the conditional as always being FALSE (as PCRE used to), but it
2009 seems to me that giving a diagnostic is better.
2010
2011 8. Change "alphameric" to the more common word "alphanumeric" in comments
2012 and messages.
2013
2014 9. Fix two occurrences of "backslash" in comments that should have been
2015 "backspace".
2016
2017 10. Remove two redundant lines of code that can never be obeyed (their function
2018 was moved elsewhere).
2019
2020 11. The program that makes PCRE's Unicode character property table had a bug
2021 which caused it to generate incorrect table entries for sequences of
2022 characters that have the same character type, but are in different scripts.
2023 It amalgamated them into a single range, with the script of the first of
2024 them. In other words, some characters were in the wrong script. There were
2025 thirteen such cases, affecting characters in the following ranges:
2026
2027 U+002b0 - U+002c1
2028 U+0060c - U+0060d
2029 U+0061e - U+00612
2030 U+0064b - U+0065e
2031 U+0074d - U+0076d
2032 U+01800 - U+01805
2033 U+01d00 - U+01d77
2034 U+01d9b - U+01dbf
2035 U+0200b - U+0200f
2036 U+030fc - U+030fe
2037 U+03260 - U+0327f
2038 U+0fb46 - U+0fbb1
2039 U+10450 - U+1049d
2040
2041 12. The -o option (show only the matching part of a line) for pcregrep was not
2042 compatible with GNU grep in that, if there was more than one match in a
2043 line, it showed only the first of them. It now behaves in the same way as
2044 GNU grep.
2045
2046 13. If the -o and -v options were combined for pcregrep, it printed a blank
2047 line for every non-matching line. GNU grep prints nothing, and pcregrep now
2048 does the same. The return code can be used to tell if there were any
2049 non-matching lines.
2050
2051 14. Added --file-offsets and --line-offsets to pcregrep.
2052
2053 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
2054 infinitely looping recursion. The bug was that positive lookaheads were not
2055 being skipped when checking for a possible empty match (negative lookaheads
2056 and both kinds of lookbehind were skipped).
2057
2058 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
2059 inclusion of <windows.h> to before rather than after the definition of
2060 INVALID_FILE_ATTRIBUTES (patch from David Byron).
2061
2062 17. Specifying a possessive quantifier with a specific limit for a Unicode
2063 character property caused pcre_compile() to compile bad code, which led at
2064 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
2065 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
2066 caused the error; without that there was no problem.
2067
2068 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
2069
2070 19. Added --enable-pcretest-libreadline.
2071
2072 20. In pcrecpp.cc, the variable 'count' was incremented twice in
2073 RE::GlobalReplace(). As a result, the number of replacements returned was
2074 double what it should be. I removed one of the increments, but Craig sent a
2075 later patch that removed the other one (the right fix) and added unit tests
2076 that check the return values (which was not done before).
2077
2078 21. Several CMake things:
2079
2080 (1) Arranged that, when cmake is used on Unix, the libraries end up with
2081 the names libpcre and libpcreposix, not just pcre and pcreposix.
2082
2083 (2) The above change means that pcretest and pcregrep are now correctly
2084 linked with the newly-built libraries, not previously installed ones.
2085
2086 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
2087
2088 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
2089 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
2090 UTF-8 newline character). The key issue is that the pattern starts .*;
2091 this means that the match must be either at the beginning, or after a
2092 newline. The bug was in the code for advancing after a failed match and
2093 checking that the new position followed a newline. It was not taking
2094 account of UTF-8 characters correctly.
2095
2096 23. PCRE was behaving differently from Perl in the way it recognized POSIX
2097 character classes. PCRE was not treating the sequence [:...:] as a
2098 character class unless the ... were all letters. Perl, however, seems to
2099 allow any characters between [: and :], though of course it rejects as
2100 unknown any "names" that contain non-letters, because all the known class
2101 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
2102 for example, whereas PCRE did not - it did not recognize a POSIX character
2103 class. This seemed a bit dangerous, so the code has been changed to be
2104 closer to Perl. The behaviour is not identical to Perl, because PCRE will
2105 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
2106 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
2107 Perl does, and where it didn't before.
2108
2109 24. Rewrite so as to remove the single use of %n from pcregrep because in some
2110 Windows environments %n is disabled by default.
2111
2112
2113 Version 7.4 21-Sep-07
2114 ---------------------
2115
2116 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
2117 means that a class such as [\s] counted as "explicit reference to CR or
2118 LF". That isn't really right - the whole point of the change was to try to
2119 help when there was an actual mention of one of the two characters. So now
2120 the change happens only if \r or \n (or a literal CR or LF) character is
2121 encountered.
2122
2123 2. The 32-bit options word was also used for 6 internal flags, but the numbers
2124 of both had grown to the point where there were only 3 bits left.
2125 Fortunately, there was spare space in the data structure, and so I have
2126 moved the internal flags into a new 16-bit field to free up more option
2127 bits.
2128
2129 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
2130 but did not set the internal JCHANGED flag - either of these is enough to
2131 control the way the "get" function works - but the PCRE_INFO_JCHANGED
2132 facility is supposed to tell if (?J) was ever used, so now (?J) at the
2133 start sets both bits.
2134
2135 4. Added options (at build time, compile time, exec time) to change \R from
2136 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
2137
2138 5. doc/pcresyntax.html was missing from the distribution.
2139
2140 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
2141 compatibility, even though it is no longer used.
2142
2143 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
2144 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
2145 windows.h file is present (where different names are used). [This was
2146 reversed later after testing - see 16 below.]
2147
2148 8. Changed all #include <config.h> to #include "config.h". There were also
2149 some further <pcre.h> cases that I changed to "pcre.h".
2150
2151 9. When pcregrep was used with the --colour option, it missed the line ending
2152 sequence off the lines that it output.
2153
2154 10. It was pointed out to me that arrays of string pointers cause lots of
2155 relocations when a shared library is dynamically loaded. A technique of
2156 using a single long string with a table of offsets can drastically reduce
2157 these. I have refactored PCRE in four places to do this. The result is
2158 dramatic:
2159
2160 Originally: 290
2161 After changing UCP table: 187
2162 After changing error message table: 43
2163 After changing table of "verbs" 36
2164 After changing table of Posix names 22
2165
2166 Thanks to the folks working on Gregex for glib for this insight.
2167
2168 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
2169 unicode-properties was also set.
2170
2171 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
2172
2173 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
2174 checked only for CRLF.
2175
2176 14. Added casts to pcretest.c to avoid compiler warnings.
2177
2178 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
2179
2180 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
2181 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
2182 entirely. This removes changes made in 7 above.
2183
2184 17. The CMake files have been updated, and there is now more information about
2185 building with CMake in the NON-UNIX-USE document.
2186
2187
2188 Version 7.3 28-Aug-07
2189 ---------------------
2190
2191 1. In the rejigging of the build system that eventually resulted in 7.1, the
2192 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
2193 brackets there is not right, since it causes compilers to look for an
2194 installed pcre.h, not the version that is in the source that is being
2195 compiled (which of course may be different). I have changed it back to:
2196
2197 #include "pcre.h"
2198
2199 I have a vague recollection that the change was concerned with compiling in
2200 different directories, but in the new build system, that is taken care of
2201 by the VPATH setting the Makefile.
2202
2203 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
2204 when the subject happened to end in the byte 0x85 (e.g. if the last
2205 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
2206 characters but of course it shouldn't be taken as a newline when it is part
2207 of another character. The bug was that, for an unlimited repeat of . in
2208 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
2209 characters when looking for a newline.
2210
2211 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
2212
2213 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
2214 in debug output.
2215
2216 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
2217 long printing in the pcrecpp unittest when running under MinGW.
2218
2219 6. ESC_K was left out of the EBCDIC table.
2220
2221 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
2222 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
2223 limit also applies to "virtual nesting" when a pattern is recursive, and in
2224 this case 1000 isn't so big. I have been able to remove this limit at the
2225 expense of backing off one optimization in certain circumstances. Normally,
2226 when pcre_exec() would call its internal match() function recursively and
2227 immediately return the result unconditionally, it uses a "tail recursion"
2228 feature to save stack. However, when a subpattern that can match an empty
2229 string has an unlimited repetition quantifier, it no longer makes this
2230 optimization. That gives it a stack frame in which to save the data for
2231 checking that an empty string has been matched. Previously this was taken
2232 from the 1000-entry workspace that had been reserved. So now there is no
2233 explicit limit, but more stack is used.
2234
2235 8. Applied Daniel's patches to solve problems with the import/export magic
2236 syntax that is required for Windows, and which was going wrong for the
2237 pcreposix and pcrecpp parts of the library. These were overlooked when this
2238 problem was solved for the main library.
2239
2240 9. There were some crude static tests to avoid integer overflow when computing
2241 the size of patterns that contain repeated groups with explicit upper
2242 limits. As the maximum quantifier is 65535, the maximum group length was
2243 set at 30,000 so that the product of these two numbers did not overflow a
2244 32-bit integer. However, it turns out that people want to use groups that
2245 are longer than 30,000 bytes (though not repeat them that many times).
2246 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
2247 made it possible to implement the integer overflow checks in a much more
2248 dynamic way, which I have now done. The artificial limitation on group
2249 length has been removed - we now have only the limit on the total length of
2250 the compiled pattern, which depends on the LINK_SIZE setting.
2251
2252 10. Fixed a bug in the documentation for get/copy named substring when
2253 duplicate names are permitted. If none of the named substrings are set, the
2254 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
2255 empty string.
2256
2257 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
2258 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
2259 because the ] is interpreted as the first data character and the
2260 terminating ] is not found. PCRE has been made compatible with Perl in this
2261 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
2262 cause memory overwriting.
2263
2264 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
2265 string has been matched (to stop an infinite loop). It was not recognizing
2266 a conditional subpattern that could match an empty string if that
2267 subpattern was within another subpattern. For example, it looped when
2268 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
2269 condition was not nested. This bug has been fixed.
2270
2271 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
2272 past the start of the subject in the presence of bytes with the top bit
2273 set, for example "\x8aBCD".
2274
2275 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
2276 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
2277
2278 14. Optimized (?!) to (*FAIL).
2279
2280 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
2281 This restricts code points to be within the range 0 to 0x10FFFF, excluding
2282 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
2283 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
2284 does: it's just the validity check that is more restrictive.
2285
2286 16. Inserted checks for integer overflows during escape sequence (backslash)
2287 processing, and also fixed erroneous offset values for syntax errors during
2288 backslash processing.
2289
2290 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
2291 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
2292
2293 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
2294 caused an overrun.
2295
2296 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
2297 something other than just ASCII characters) inside a group that had an
2298 unlimited repeat caused a loop at compile time (while checking to see
2299 whether the group could match an empty string).
2300
2301 20. Debugging a pattern containing \p or \P could cause a crash. For example,
2302 [\P{Any}] did so. (Error in the code for printing property names.)
2303
2304 21. An orphan \E inside a character class could cause a crash.
2305
2306 22. A repeated capturing bracket such as (A)? could cause a wild memory
2307 reference during compilation.
2308
2309 23. There are several functions in pcre_compile() that scan along a compiled
2310 expression for various reasons (e.g. to see if it's fixed length for look
2311 behind). There were bugs in these functions when a repeated \p or \P was
2312 present in the pattern. These operators have additional parameters compared
2313 with \d, etc, and these were not being taken into account when moving along
2314 the compiled data. Specifically:
2315
2316 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
2317 length.
2318
2319 (b) An item such as \pL+ within a repeated group could cause crashes or
2320 loops.
2321
2322 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
2323 "reference to non-existent subpattern" error.
2324
2325 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
2326
2327 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
2328 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
2329
2330 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
2331
2332 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
2333 character were causing crashes (broken optimization).
2334
2335 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
2336 \p or \P) caused a compile-time loop.
2337
2338 28. More problems have arisen in unanchored patterns when CRLF is a valid line
2339 break. For example, the unstudied pattern [\r\n]A does not match the string
2340 "\r\nA" because change 7.0/46 below moves the current point on by two
2341 characters after failing to match at the start. However, the pattern \nA
2342 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
2343 the same is true. There doesn't seem any very clean way out of this, but
2344 what I have chosen to do makes the common cases work: PCRE now takes note
2345 of whether there can be an explicit match for \r or \n anywhere in the
2346 pattern, and if so, 7.0/46 no longer applies. As part of this change,
2347 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
2348 pattern has explicit CR or LF references.
2349
2350 29. Added (*CR) etc for changing newline setting at start of pattern.
2351
2352
2353 Version 7.2 19-Jun-07
2354 ---------------------
2355
2356 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
2357 which is apparently normally available under Windows.
2358
2359 2. Re-jig the pcregrep tests with different newline settings in an attempt
2360 to make them independent of the local environment's newline setting.
2361
2362 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
2363
2364 4. Some of the "internals" tests were previously cut out when the link size
2365 was not 2, because the output contained actual offsets. The recent new
2366 "Z" feature of pcretest means that these can be cut out, making the tests
2367 usable with all link sizes.
2368
2369 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
2370 stack recursion. This gives a massive performance boost under BSD, but just
2371 a small improvement under Linux. However, it saves one field in the frame
2372 in all cases.
2373
2374 6. Added more features from the forthcoming Perl 5.10:
2375
2376 (a) (?-n) (where n is a string of digits) is a relative subroutine or
2377 recursion call. It refers to the nth most recently opened parentheses.
2378
2379 (b) (?+n) is also a relative subroutine call; it refers to the nth next
2380 to be opened parentheses.
2381
2382 (c) Conditions that refer to capturing parentheses can be specified
2383 relatively, for example, (?(-2)... or (?(+3)...
2384
2385 (d) \K resets the start of the current match so that everything before
2386 is not part of it.
2387
2388 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
2389
2390 (f) \g{name} is another synonym - part of Perl 5.10's unification of
2391 reference syntax.
2392
2393 (g) (?| introduces a group in which the numbering of parentheses in each
2394 alternative starts with the same number.
2395
2396 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
2397
2398 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
2399 PCRE_INFO_JCHANGED.
2400
2401 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
2402 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
2403 for detecting groups that can match an empty string.
2404
2405 9. A pattern with a very large number of alternatives (more than several
2406 hundred) was running out of internal workspace during the pre-compile
2407 phase, where pcre_compile() figures out how much memory will be needed. A
2408 bit of new cunning has reduced the workspace needed for groups with
2409 alternatives. The 1000-alternative test pattern now uses 12 bytes of
2410 workspace instead of running out of the 4096 that are available.
2411
2412 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
2413
2414 11. Applied patch from Google to remove an optimization that didn't quite work.
2415 The report of the bug said:
2416
2417 pcrecpp::RE("a*").FullMatch("aaa") matches, while
2418 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
2419 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
2420
2421 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
2422 it matched the wrong number of bytes.
2423
2424
2425 Version 7.1 24-Apr-07
2426 ---------------------
2427
2428 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
2429 that is more "standard", making use of automake and other Autotools. There
2430 is some re-arrangement of the files and adjustment of comments consequent
2431 on this.
2432
2433 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
2434 for recursive directory scanning broke on some systems because the files
2435 are not scanned in any specific order and on different systems the order
2436 was different. A call to "sort" has been inserted into RunGrepTest for the
2437 approprate test as a short-term fix. In the longer term there may be an
2438 alternative.
2439
2440 3. I had an email from Eric Raymond about problems translating some of PCRE's
2441 man pages to HTML (despite the fact that I distribute HTML pages, some
2442 people do their own conversions for various reasons). The problems
2443 concerned the use of low-level troff macros .br and .in. I have therefore
2444 removed all such uses from the man pages (some were redundant, some could
2445 be replaced by .nf/.fi pairs). The 132html script that I use to generate
2446 HTML has been updated to handle .nf/.fi and to complain if it encounters
2447 .br or .in.
2448
2449 4. Updated comments in configure.ac that get placed in config.h.in and also
2450 arranged for config.h to be included in the distribution, with the name
2451 config.h.generic, for the benefit of those who have to compile without
2452 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
2453
2454 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
2455 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
2456 makevp.bat for the current PCRE, using the additional files
2457 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
2458
2459 6. A Windows user reported a minor discrepancy with test 2, which turned out
2460 to be caused by a trailing space on an input line that had got lost in his
2461 copy. The trailing space was an accident, so I've just removed it.
2462
2463 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
2464 that is needed.
2465
2466 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
2467 as "const" (a) because they are and (b) because it helps the PHP
2468 maintainers who have recently made a script to detect big data structures
2469 in the php code that should be moved to the .rodata section. I remembered
2470 to update Builducptable as well, so it won't revert if ucptable.h is ever
2471 re-created.
2472
2473 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
2474 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
2475 order to be able to cut out the UTF-8 tables in the latter when UTF-8
2476 support is not required. This saves 1.5-2K of code, which is important in
2477 some applications.
2478
2479 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
2480 so as not to refer to the tables, even though these functions will never be
2481 called when UTF-8 support is disabled. Otherwise there are problems with a
2482 shared library.
2483
2484 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
2485
2486 (a) It was defining its arguments as char * instead of void *.
2487
2488 (b) It was assuming that all moves were upwards in memory; this was true
2489 a long time ago when I wrote it, but is no longer the case.
2490
2491 The emulated memove() is provided for those environments that have neither
2492 memmove() nor bcopy(). I didn't think anyone used it these days, but that
2493 is clearly not the case, as these two bugs were recently reported.
2494
2495 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
2496 and Detrail to create the HTML documentation, the .txt form of the man
2497 pages, and it removes trailing spaces from listed files. It also creates
2498 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
2499 case, it wraps all the #defines with #ifndefs. This script should be run
2500 before "make dist".
2501
2502 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
2503 with Unicode property support.
2504
2505 (a) For a maximizing quantifier, if the two different cases of the
2506 character were of different lengths in their UTF-8 codings (there are
2507 some cases like this - I found 11), and the matching function had to
2508 back up over a mixture of the two cases, it incorrectly assumed they
2509 were both the same length.
2510
2511 (b) When PCRE was configured to use the heap rather than the stack for
2512 recursion during matching, it was not correctly preserving the data for
2513 the other case of a UTF-8 character when checking ahead for a match
2514 while processing a minimizing repeat. If the check also involved
2515 matching a wide character, but failed, corruption could cause an
2516 erroneous result when trying to check for a repeat of the original
2517 character.
2518
2519 13. Some tidying changes to the testing mechanism:
2520
2521 (a) The RunTest script now detects the internal link size and whether there
2522 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
2523 values substituted by "configure". (The RunGrepTest script already did
2524 this for UTF-8.) The configure.ac script no longer substitutes the
2525 relevant variables.
2526
2527 (b) The debugging options /B and /D in pcretest show the compiled bytecode
2528 with length and offset values. This means that the output is different
2529 for different internal link sizes. Test 2 is skipped for link sizes
2530 other than 2 because of this, bypassing the problem. Unfortunately,
2531 there was also a test in test 3 (the locale tests) that used /B and
2532 failed for link sizes other than 2. Rather than cut the whole test out,
2533 I have added a new /Z option to pcretest that replaces the length and
2534 offset values with spaces. This is now used to make test 3 independent
2535 of link size. (Test 2 will be tidied up later.)
2536
2537 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
2538 segmentation fault instead of returning the appropriate error message.
2539
2540 15. In multiline mode when the newline sequence was set to "any", the pattern
2541 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
2542 This doesn't seem right; it now treats the CRLF combination as the line
2543 ending, and so does not match in that case. It's only a pattern such as ^$
2544 that would hit this one: something like ^ABC$ would have failed after \r
2545 and then tried again after \r\n.
2546
2547 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
2548 in an attempt to make files that differ only in their line terminators
2549 compare equal. This works on Linux.
2550
2551 17. Under certain error circumstances pcregrep might try to free random memory
2552 as it exited. This is now fixed, thanks to valgrind.
2553
2554 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
2555 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
2556 was because its rules for how to advance for /g after matching an empty
2557 string at the end of a line did not allow for this case. They now check for
2558 it specially.
2559
2560 20. pcretest is supposed to handle patterns and data of any length, by
2561 extending its buffers when necessary. It was getting this wrong when the
2562 buffer for a data line had to be extended.
2563
2564 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
2565 CRLF as a newline sequence.
2566
2567 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
2568 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
2569 I have nevertheless tidied it up.
2570
2571 23. Added some casts to kill warnings from HP-UX ia64 compiler.
2572
2573 24. Added a man page for pcre-config.
2574
2575
2576 Version 7.0 19-Dec-06
2577 ---------------------
2578
2579 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
2580 moving to gcc 4.1.1.
2581
2582 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
2583 sys/time.h, which is documented as needed for this function. It doesn't
2584 seem to matter on Linux, but it showed up on some releases of OS X.
2585
2586 3. It seems that there are systems where bytes whose values are greater than
2587 127 match isprint() in the "C" locale. The "C" locale should be the
2588 default when a C program starts up. In most systems, only ASCII printing
2589 characters match isprint(). This difference caused the output from pcretest
2590 to vary, making some of the tests fail. I have changed pcretest so that:
2591
2592 (a) When it is outputting text in the compiled version of a pattern, bytes
2593 other than 32-126 are always shown as hex escapes.
2594
2595 (b) When it is outputting text that is a matched part of a subject string,
2596 it does the same, unless a different locale has been set for the match
2597 (using the /L modifier). In this case, it uses isprint() to decide.
2598
2599 4. Fixed a major bug that caused incorrect computation of the amount of memory
2600 required for a compiled pattern when options that changed within the
2601 pattern affected the logic of the preliminary scan that determines the
2602 length. The relevant options are -x, and -i in UTF-8 mode. The result was
2603 that the computed length was too small. The symptoms of this bug were
2604 either the PCRE error "internal error: code overflow" from pcre_compile(),
2605 or a glibc crash with a message such as "pcretest: free(): invalid next
2606 size (fast)". Examples of patterns that provoked this bug (shown in
2607 pcretest format) are:
2608
2609 /(?-x: )/x
2610 /(?x)(?-x: \s*#\s*)/
2611 /((?i)[\x{c0}])/8
2612 /(?i:[\x{c0}])/8
2613
2614 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
2615 is now done differently.
2616
2617 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
2618 wrapper classes; (b) implement a new function in the C++ scanner that is
2619 more efficient than the old way of doing things because it avoids levels of
2620 recursion in the regex matching; (c) add a paragraph to the documentation
2621 for the FullMatch() function.
2622
2623 6. The escape sequence \n was being treated as whatever was defined as
2624 "newline". Not only was this contrary to the documentation, which states
2625 that \n is character 10 (hex 0A), but it also went horribly wrong when
2626 "newline" was defined as CRLF. This has been fixed.
2627
2628 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
2629 was being set to -1 for the "end of line" case (supposedly a value that no
2630 character can have). Though this value is never used (the check for end of
2631 line is "zero bytes in current character"), it caused compiler complaints.
2632 I've changed it to 0xffffffff.
2633
2634 8. In pcre_version.c, the version string was being built by a sequence of
2635 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
2636 string (as it is for production releases) called a macro with an empty
2637 argument. The C standard says the result of this is undefined. The gcc
2638 compiler treats it as an empty string (which was what was wanted) but it is
2639 reported that Visual C gives an error. The source has been hacked around to
2640 avoid this problem.
2641
2642 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
2643 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
2644 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
2645 of them did).
2646
2647 10. Originally, pcretest opened its input and output without "b"; then I was
2648 told that "b" was needed in some environments, so it was added for release
2649 5.0 to both the input and output. (It makes no difference on Unix-like
2650 systems.) Later I was told that it is wrong for the input on Windows. I've
2651 now abstracted the modes into two macros, to make it easier to fiddle with
2652 them, and removed "b" from the input mode under Windows.
2653
2654 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
2655
2656 12. Added -help and --help to pcretest as an official way of being reminded
2657 of the options.
2658
2659 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
2660 and pcrecpp.cc because they annoy compilers at high warning levels.
2661
2662 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
2663
2664 15. Fixed an occurrence of == in configure.ac that should have been = (shell
2665 scripts are not C programs :-) and which was not noticed because it works
2666 on Linux.
2667
2668 16. pcretest is supposed to handle any length of pattern and data line (as one
2669 line or as a continued sequence of lines) by extending its input buffer if
2670 necessary. This feature was broken for very long pattern lines, leading to
2671 a string of junk being passed to pcre_compile() if the pattern was longer
2672 than about 50K.
2673
2674 17. I have done a major re-factoring of the way pcre_compile() computes the
2675 amount of memory needed for a compiled pattern. Previously, there was code
2676 that made a preliminary scan of the pattern in order to do this. That was
2677 OK when PCRE was new, but as the facilities have expanded, it has become
2678 harder and harder to keep it in step with the real compile phase, and there
2679 have been a number of bugs (see for example, 4 above). I have now found a
2680 cunning way of running the real compile function in a "fake" mode that
2681 enables it to compute how much memory it would need, while actually only
2682 ever using a few hundred bytes of working memory and without too many
2683 tests of the mode. This should make future maintenance and development
2684 easier. A side effect of this work is that the limit of 200 on the nesting
2685 depth of parentheses has been removed (though this was never a serious
2686 limitation, I suspect). However, there is a downside: pcre_compile() now
2687 runs more slowly than before (30% or more, depending on the pattern). I
2688 hope this isn't a big issue. There is no effect on runtime performance.
2689
2690 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
2691 newline (only possible for the last line of a file) and it was a
2692 pattern that set a locale (followed by /Lsomething), pcretest crashed.
2693
2694 19. Added additional timing features to pcretest. (1) The -tm option now times
2695 matching only, not compiling. (2) Both -t and -tm can be followed, as a
2696 separate command line item, by a number that specifies the number of
2697 repeats to use when timing. The default is 50000; this gives better
2698 precision, but takes uncomfortably long for very large patterns.
2699
2700 20. Extended pcre_study() to be more clever in cases where a branch of a
2701 subpattern has no definite first character. For example, (a*|b*)[cd] would
2702 previously give no result from pcre_study(). Now it recognizes that the
2703 first character must be a, b, c, or d.
2704
2705 21. There was an incorrect error "recursive call could loop indefinitely" if
2706 a subpattern (or the entire pattern) that was being tested for matching an
2707 empty string contained only one non-empty item after a nested subpattern.
2708 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2709 incorrectly, because the \d was being skipped in the check.
2710
2711 22. The pcretest program now has a new pattern option /B and a command line
2712 option -b, which is equivalent to adding /B to every pattern. This causes
2713 it to show the compiled bytecode, without the additional information that
2714 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2715 is the same as /B/I).
2716
2717 23. A new optimization is now able automatically to treat some sequences such
2718 as a*b as a*+b. More specifically, if something simple (such as a character
2719 or a simple class like \d) has an unlimited quantifier, and is followed by
2720 something that cannot possibly match the quantified thing, the quantifier
2721 is automatically "possessified".
2722
2723 24. A recursive reference to a subpattern whose number was greater than 39
2724 went wrong under certain circumstances in UTF-8 mode. This bug could also
2725 have affected the operation of pcre_study().
2726
2727 25. Realized that a little bit of performance could be had by replacing
2728 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2729
2730 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2731
2732 27. Possessive quantifiers such as a++ were previously implemented by turning
2733 them into atomic groups such as ($>a+). Now they have their own opcodes,
2734 which improves performance. This includes the automatically created ones
2735 from 23 above.
2736
2737 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2738 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2739 the first matched character to be a colon. This applied both to named and
2740 numbered groups.
2741
2742 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2743
2744 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2745 building PCRE on Windows easier, so I have included it in the distribution.
2746
2747 31. There is now a check in pcretest against a ridiculously large number being
2748 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2749 loop, the loop is abandoned.
2750
2751 32. Forward references to subpatterns in conditions such as (?(2)...) where
2752 subpattern 2 is defined later cause pcre_compile() to search forwards in
2753 the pattern for the relevant set of parentheses. This search went wrong
2754 when there were unescaped parentheses in a character class, parentheses
2755 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2756
2757 33. "Subroutine" calls and backreferences were previously restricted to
2758 referencing subpatterns earlier in the regex. This restriction has now
2759 been removed.
2760
2761 34. Added a number of extra features that are going to be in Perl 5.10. On the
2762 whole, these are just syntactic alternatives for features that PCRE had
2763 previously implemented using the Python syntax or my own invention. The
2764 other formats are all retained for compatibility.
2765
2766 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2767 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2768 also .NET compatible.
2769
2770 (b) A recursion or subroutine call to a named group can now be defined as
2771 (?&name) as well as (?P>name).
2772
2773 (c) A backreference to a named group can now be defined as \k<name> or
2774 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2775 5.10, are also .NET compatible.
2776
2777 (d) A conditional reference to a named group can now use the syntax
2778 (?(<name>) or (?('name') as well as (?(name).
2779
2780 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2781 groups (named and numbered) that are never evaluated inline, but can be
2782 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2783 is always false. There may be only one alternative in such a group.
2784
2785 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2786 as the simple (?(R). The condition is true only if the most recent
2787 recursion is that of the given number or name. It does not search out
2788 through the entire recursion stack.
2789
2790 (g) The escape \gN or \g{N} has been added, where N is a positive or
2791 negative number, specifying an absolute or relative reference.
2792
2793 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2794 some "unreachable code" warnings.
2795
2796 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2797 things, this adds five new scripts.
2798
2799 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2800 There were also incompatibilities regarding the handling of \Q..\E inside
2801 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2802 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2803
2804 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2805 matches an empty string, and forcibly breaks the loop. There were bugs in
2806 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2807 against aaaa the result was just "a" rather than "aaaa", for example. Two
2808 separate and independent bugs (that affected different cases) have been
2809 fixed.
2810
2811 39. Refactored the code to abolish the use of different opcodes for small
2812 capturing bracket numbers. This is a tidy that I avoided doing when I
2813 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2814 The new approach is not only tidier, it makes it possible to reduce the
2815 memory needed to fix the previous bug (38).
2816
2817 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2818 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2819 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2820 mode.
2821
2822 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2823 report.
2824
2825 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2826 copy construction and assignment in the C++ wrapper.
2827
2828 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2829 couple of bugs that could have given wrong results in the "--newline=crlf"
2830 case.
2831
2832 44. Added a number of casts and did some reorganization of signed/unsigned int
2833 variables following suggestions from Dair Grant. Also renamed the variable
2834 "this" as "item" because it is a C++ keyword.
2835
2836 45. Arranged for dftables to add
2837
2838 #include "pcre_internal.h"
2839
2840 to pcre_chartables.c because without it, gcc 4.x may remove the array
2841 definition from the final binary if PCRE is built into a static library and
2842 dead code stripping is activated.
2843
2844 46. For an unanchored pattern, if a match attempt fails at the start of a
2845 newline sequence, and the newline setting is CRLF or ANY, and the next two
2846 characters are CRLF, advance by two characters instead of one.
2847
2848
2849 Version 6.7 04-Jul-06
2850 ---------------------
2851
2852 1. In order to handle tests when input lines are enormously long, pcretest has
2853 been re-factored so that it automatically extends its buffers when
2854 necessary. The code is crude, but this _is_ just a test program. The
2855 default size has been increased from 32K to 50K.
2856
2857 2. The code in pcre_study() was using the value of the re argument before
2858 testing it for NULL. (Of course, in any sensible call of the function, it
2859 won't be NULL.)
2860
2861 3. The memmove() emulation function in pcre_internal.h, which is used on
2862 systems that lack both memmove() and bcopy() - that is, hardly ever -
2863 was missing a "static" storage class specifier.
2864
2865 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2866 containing an extended class (one that cannot be represented by a bitmap
2867 because it contains high-valued characters or Unicode property items, e.g.
2868 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2869 pattern, but PCRE should not loop if you do not (it no longer does).
2870 [Detail: two cases were found: (a) a repeated subpattern containing an
2871 extended class; (b) a recursive reference to a subpattern that followed a
2872 previous extended class. It wasn't skipping over the extended class
2873 correctly when UTF-8 mode was not set.]
2874
2875 5. A negated single-character class was not being recognized as fixed-length
2876 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2877 compile error "lookbehind assertion is not fixed length".
2878
2879 6. The RunPerlTest auxiliary script was showing an unexpected difference
2880 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2881 write a Perl script that can interpret lines of an input file either as
2882 byte characters or as UTF-8, which is what "perltest" was being required to
2883 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2884 can't do is switch easily at run time between having the "use utf8;" pragma
2885 or not. In the end, I fudged it by using the RunPerlTest script to insert
2886 "use utf8;" explicitly for the UTF-8 tests.
2887
2888 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2889 the end of the subject string, contrary to the documentation and to what
2890 Perl does. This was true of both matching functions. Now it matches only at
2891 the start of the subject and immediately after *internal* newlines.
2892
2893 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2894 a pointer to an int instead of a pointer to an unsigned long int. This
2895 caused problems on 64-bit systems.
2896
2897 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2898 instance of the 'standard' template library not being so standard".
2899
2900 10. There was no check on the number of named subpatterns nor the maximum
2901 length of a subpattern name. The product of these values is used to compute
2902 the size of the memory block for a compiled pattern. By supplying a very
2903 long subpattern name and a large number of named subpatterns, the size
2904 computation could be caused to overflow. This is now prevented by limiting
2905 the length of names to 32 characters, and the number of named subpatterns
2906 to 10,000.
2907
2908 11. Subpatterns that are repeated with specific counts have to be replicated in
2909 the compiled pattern. The size of memory for this was computed from the
2910 length of the subpattern and the repeat count. The latter is limited to
2911 65535, but there was no limit on the former, meaning that integer overflow
2912 could in principle occur. The compiled length of a repeated subpattern is
2913 now limited to 30,000 bytes in order to prevent this.
2914
2915 12. Added the optional facility to have named substrings with the same name.
2916
2917 13. Added the ability to use a named substring as a condition, using the
2918 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2919 are numbers (not recommended). Forward references are permitted.
2920
2921 14. Added forward references in named backreferences (if you see what I mean).
2922
2923 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2924 pattern could run off the end of the subject. For example, the pattern
2925 "(?s)(.{1,5})"8 did this with the subject "ab".
2926
2927 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2928 PCRE_CASELESS was set when matching characters that were quantified with ?
2929 or *.
2930
2931 17. A character class other than a single negated character that had a minimum
2932 but no maximum quantifier - for example [ab]{6,} - was not handled
2933 correctly by pce_dfa_exec(). It would match only one character.
2934
2935 18. A valid (though odd) pattern that looked like a POSIX character
2936 class but used an invalid character after [ (for example [[,abc,]]) caused
2937 pcre_compile() to give the error "Failed: internal error: code overflow" or
2938 in some cases to crash with a glibc free() error. This could even happen if
2939 the pattern terminated after [[ but there just happened to be a sequence of
2940 letters, a binary zero, and a closing ] in the memory that followed.
2941
2942 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2943 over the years. Originally (before any Unicode support), just the bottom 8
2944 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2945 output from "man perlunicode" includes this:
2946
2947 The regular expression compiler produces polymorphic opcodes. That
2948 is, the pattern adapts to the data and automatically switches to
2949 the Unicode character scheme when presented with Unicode data--or
2950 instead uses a traditional byte scheme when presented with byte
2951 data.
2952
2953 Sadly, a wide octal escape does not cause a switch, and in a string with
2954 no other multibyte characters, these octal escapes are treated as before.
2955 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2956 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2957 Unicode string.
2958
2959 I have not perpetrated such confusion in PCRE. Up till now, it took just
2960 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2961 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2962 translate to the appropriate multibyte character.
2963
2964 29. Applied some refactoring to reduce the number of warnings from Microsoft
2965 and Borland compilers. This has included removing the fudge introduced
2966 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2967 a warning about an unused variable.
2968
2969 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2970 characters since release 4.0, because Perl (from release 5.004) does not.
2971 [Or at least, is documented not to: some releases seem to be in conflict
2972 with the documentation.] However, when a pattern was studied with
2973 pcre_study() and all its branches started with \s, PCRE still included VT
2974 as a possible starting character. Of course, this did no harm; it just
2975 caused an unnecessary match attempt.
2976
2977 22. Removed a now-redundant internal flag bit that recorded the fact that case
2978 dependency changed within the pattern. This was once needed for "required
2979 byte" processing, but is no longer used. This recovers a now-scarce options
2980 bit. Also moved the least significant internal flag bit to the most-
2981 significant bit of the word, which was not previously used (hangover from
2982 the days when it was an int rather than a uint) to free up another bit for
2983 the future.
2984
2985 23. Added support for CRLF line endings as well as CR and LF. As well as the
2986 default being selectable at build time, it can now be changed at runtime
2987 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2988 specify that it is scanning data with non-default line endings.
2989
2990 24. Changed the definition of CXXLINK to make it agree with the definition of
2991 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2992
2993 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2994 recursions. This makes a big different to stack usage for some patterns.
2995
2996 26. If a subpattern containing a named recursion or subroutine reference such
2997 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2998 the space required for the compiled pattern went wrong and gave too small a
2999 value. Depending on the environment, this could lead to "Failed: internal
3000 error: code overflow at offset 49" or "glibc detected double free or
3001 corruption" errors.
3002
3003 27. Applied patches from Google (a) to support the new newline modes and (b) to
3004 advance over multibyte UTF-8 characters in GlobalReplace.
3005
3006 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3007 difference for some implementation of PCRE in some Windows version.
3008
3009 29. Added some extra testing facilities to pcretest:
3010
3011 \q<number> in a data line sets the "match limit" value
3012 \Q<number> in a data line sets the "match recursion limt" value
3013 -S <number> sets the stack size, where <number> is in megabytes
3014
3015 The -S option isn't available for Windows.
3016
3017
3018 Version 6.6 06-Feb-06
3019 ---------------------
3020
3021 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
3022 in pcreposix.h. I have copied the definition from pcre.h.
3023
3024 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
3025 because pcre.h is no longer a built file.
3026
3027 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
3028 not normally included in the compiled code.
3029
3030
3031 Version 6.5 01-Feb-06
3032 ---------------------
3033
3034 1. When using the partial match feature with pcre_dfa_exec(), it was not
3035 anchoring the second and subsequent partial matches at the new starting
3036 point. This could lead to incorrect results. For example, with the pattern
3037 /1234/, partially matching against "123" and then "a4" gave a match.
3038
3039 2. Changes to pcregrep:
3040
3041 (a) All non-match returns from pcre_exec() were being treated as failures
3042 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
3043 error message is output. Some extra information is given for the
3044 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
3045 probably the only errors that are likely to be caused by users (by
3046 specifying a regex that has nested indefinite repeats, for instance).
3047 If there are more than 20 of these errors, pcregrep is abandoned.
3048
3049 (b) A binary zero was treated as data while matching, but terminated the
3050 output line if it was written out. This has been fixed: binary zeroes
3051 are now no different to any other data bytes.
3052
3053 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
3054 used to set a locale for matching. The --locale=xxxx long option has
3055 been added (no short equivalent) to specify a locale explicitly on the
3056 pcregrep command, overriding the environment variables.
3057
3058 (d) When -B was used with -n, some line numbers in the output were one less
3059 than they should have been.
3060
3061 (e) Added the -o (--only-matching) option.
3062
3063 (f) If -A or -C was used with -c (count only), some lines of context were
3064 accidentally printed for the final match.
3065
3066 (g) Added the -H (--with-filename) option.
3067
3068 (h) The combination of options -rh failed to suppress file names for files
3069 that were found from directory arguments.
3070
3071 (i) Added the -D (--devices) and -d (--directories) options.
3072
3073 (j) Added the -F (--fixed-strings) option.
3074
3075 (k) Allow "-" to be used as a file name for -f as well as for a data file.
3076
3077 (l) Added the --colo(u)r option.
3078
3079 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
3080 is not present by default.
3081
3082 3. A nasty bug was discovered in the handling of recursive patterns, that is,
3083 items such as (?R) or (?1), when the recursion could match a number of
3084 alternatives. If it matched one of the alternatives, but subsequently,
3085 outside the recursion, there was a failure, the code tried to back up into
3086 the recursion. However, because of the way PCRE is implemented, this is not
3087 possible, and the result was an incorrect result from the match.
3088
3089 In order to prevent this happening, the specification of recursion has
3090 been changed so that all such subpatterns are automatically treated as
3091 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
3092
3093 4. I had overlooked the fact that, in some locales, there are characters for
3094 which isalpha() is true but neither isupper() nor islower() are true. In
3095 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
3096 and ordfeminine) are like this. This affected the treatment of \w and \W
3097 when they appeared in character classes, but not when they appeared outside
3098 a character class. The bit map for "word" characters is now created
3099 separately from the results of isalnum() instead of just taking it from the
3100 upper, lower, and digit maps. (Plus the underscore character, of course.)
3101
3102 5. The above bug also affected the handling of POSIX character classes such as
3103 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
3104 permanent tables. Instead, the bit maps for such a class were previously
3105 created as the appropriate unions of the upper, lower, and digit bitmaps.
3106 Now they are created by subtraction from the [[:word:]] class, which has
3107 its own bitmap.
3108
3109 6. The [[:blank:]] character class matches horizontal, but not vertical space.
3110 It is created by subtracting the vertical space characters (\x09, \x0a,
3111 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
3112 subtraction was done in the overall bitmap for a character class, meaning
3113 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
3114 be recognized. This bug has been fixed.
3115
3116 7. Patches from the folks at Google:
3117
3118 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
3119 real life, but is still worth protecting against".
3120
3121 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
3122 regular expressions".
3123
3124 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
3125 have it.
3126
3127 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
3128 "configure" and the latter not, in order to fix a problem somebody had
3129 with compiling the Arg class on HP-UX.
3130
3131 (e) Improve the error-handling of the C++ wrapper a little bit.
3132
3133 (f) New tests for checking recursion limiting.
3134
3135 8. The pcre_memmove() function, which is used only if the environment does not
3136 have a standard memmove() function (and is therefore rarely compiled),
3137 contained two bugs: (a) use of int instead of size_t, and (b) it was not
3138 returning a result (though PCRE never actually uses the result).
3139
3140 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
3141 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
3142 returned instead of calling malloc() with an overflowing number that would
3143 most likely cause subsequent chaos.
3144
3145 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
3146
3147 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
3148 with this option is matched, the nmatch and pmatch options of regexec() are
3149 ignored.
3150
3151 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
3152 provided in case anyone wants to the the POSIX interface with UTF-8
3153 strings.
3154
3155 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
3156 C++ linking (needed for some HP-UX environments).
3157
3158 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
3159 (unused parameter) and in the pcre_printint() function (omitted "default"
3160 switch label when the default is to do nothing).
3161
3162 15. Added some code to make it possible, when PCRE is compiled as a C++
3163 library, to replace subject pointers for pcre_exec() with a smart pointer
3164 class, thus making it possible to process discontinuous strings.
3165
3166 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
3167 much the same function. They were added by different people who were trying
3168 to make PCRE easy to compile on non-Unix systems. It has been suggested
3169 that PCRE_EXPORT be abolished now that there is more automatic apparatus
3170 for compiling on Windows systems. I have therefore replaced it with
3171 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
3172 defaults to "extern" for C or "extern C" for C++, which works fine on
3173 Unix-like systems. It is now possible to override the value of PCRE_DATA_
3174 SCOPE with something explicit in config.h. In addition:
3175
3176 (a) pcreposix.h still had just "extern" instead of either of these macros;
3177 I have replaced it with PCRE_DATA_SCOPE.
3178
3179 (b) Functions such as _pcre_xclass(), which are internal to the library,
3180 but external in the C sense, all had PCRE_EXPORT in their definitions.
3181 This is apparently wrong for the Windows case, so I have removed it.
3182 (It makes no difference on Unix-like systems.)
3183
3184 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
3185 of recursive calls to match(). This is different to MATCH_LIMIT because
3186 that limits the total number of calls to match(), not all of which increase
3187 the depth of recursion. Limiting the recursion depth limits the amount of
3188 stack (or heap if NO_RECURSE is set) that is used. The default can be set
3189 when PCRE is compiled, and changed at run time. A patch from Google adds
3190 this functionality to the C++ interface.
3191
3192 18. Changes to the handling of Unicode character properties:
3193
3194 (a) Updated the table to Unicode 4.1.0.
3195
3196 (b) Recognize characters that are not in the table as "Cn" (undefined).
3197
3198 (c) I revised the way the table is implemented to a much improved format
3199 which includes recognition of ranges. It now supports the ranges that
3200 are defined in UnicodeData.txt, and it also amalgamates other
3201 characters into ranges. This has reduced the number of entries in the
3202 table from around 16,000 to around 3,000, thus reducing its size
3203 considerably. I realized I did not need to use a tree structure after
3204 all - a binary chop search is just as efficient. Having reduced the
3205 number of entries, I extended their size from 6 bytes to 8 bytes to
3206 allow for more data.
3207
3208 (d) Added support for Unicode script names via properties such as \p{Han}.
3209
3210 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
3211 matching that character.
3212
3213 20. When matching a repeated Unicode property with a minimum greater than zero,
3214 (for example \pL{2,}), PCRE could look past the end of the subject if it
3215 reached it while seeking the minimum number of characters. This could
3216 happen only if some of the characters were more than one byte long, because
3217 there is a check for at least the minimum number of bytes.
3218
3219 21. Refactored the implementation of \p and \P so as to be more general, to
3220 allow for more different types of property in future. This has changed the
3221 compiled form incompatibly. Anybody with saved compiled patterns that use
3222 \p or \P will have to recompile them.
3223
3224 22. Added "Any" and "L&" to the supported property types.
3225
3226 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
3227 but give a compile time error if the value is greater than 0xff.
3228
3229 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
3230 accidentally not being installed or uninstalled.
3231
3232 25. The pcre.h file was built from pcre.h.in, but the only changes that were
3233 made were to insert the current release number. This seemed silly, because
3234 it made things harder for people building PCRE on systems that don't run
3235 "configure". I have turned pcre.h into a distributed file, no longer built
3236 by "configure", with the version identification directly included. There is
3237 no longer a pcre.h.in file.
3238
3239 However, this change necessitated a change to the pcre-config script as
3240 well. It is built from pcre-config.in, and one of the substitutions was the
3241 release number. I have updated configure.ac so that ./configure now finds
3242 the release number by grepping pcre.h.
3243
3244 26. Added the ability to run the tests under valgrind.
3245
3246
3247 Version 6.4 05-Sep-05
3248 ---------------------
3249
3250 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
3251 "--" to be printed when multiple files were scanned, even when none of the
3252 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
3253 consider it to be a bug, and have restored the previous behaviour.
3254
3255 2. A couple of code tidies to get rid of compiler warnings.
3256
3257 3. The pcretest program used to cheat by referring to symbols in the library
3258 whose names begin with _pcre_. These are internal symbols that are not
3259 really supposed to be visible externally, and in some environments it is
3260 possible to suppress them. The cheating is now confined to including
3261 certain files from the library's source, which is a bit cleaner.
3262
3263 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
3264 file's purpose clearer.
3265
3266 5. Reorganized pcre_ucp_findchar().
3267
3268
3269 Version 6.3 15-Aug-05
3270 ---------------------
3271
3272 1. The file libpcre.pc.in did not have general read permission in the tarball.
3273
3274 2. There were some problems when building without C++ support:
3275
3276 (a) If C++ support was not built, "make install" and "make test" still
3277 tried to test it.
3278
3279 (b) There were problems when the value of CXX was explicitly set. Some
3280 changes have been made to try to fix these, and ...
3281
3282 (c) --disable-cpp can now be used to explicitly disable C++ support.
3283
3284 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
3285 backslash in a target when C++ was disabled. This confuses some
3286 versions of "make", apparently. Using an intermediate variable solves
3287 this. (Same for CPP_LOBJ.)
3288
3289 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
3290 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
3291 necessary on certain architectures.
3292
3293 4. Added a setting of -export-symbols-regex to the link command to remove
3294 those symbols that are exported in the C sense, but actually are local
3295 within the library, and not documented. Their names all begin with
3296 "_pcre_". This is not a perfect job, because (a) we have to except some
3297 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
3298 available (and never for static libraries). I have made a note to try to
3299 find a way round (a) in the future.
3300
3301
3302 Version 6.2 01-Aug-05
3303 ---------------------
3304
3305 1. There was no test for integer overflow of quantifier values. A construction
3306 such as {1111111111111111} would give undefined results. What is worse, if
3307 a minimum quantifier for a parenthesized subpattern overflowed and became
3308 negative, the calculation of the memory size went wrong. This could have
3309 led to memory overwriting.
3310
3311 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
3312
3313 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
3314 operating environments where this matters.
3315
3316 4. Applied Giuseppe Maxia's patch to add additional features for controlling
3317 PCRE options from within the C++ wrapper.
3318
3319 5. Named capturing subpatterns were not being correctly counted when a pattern
3320 was compiled. This caused two problems: (a) If there were more than 100
3321 such subpatterns, the calculation of the memory needed for the whole
3322 compiled pattern went wrong, leading to an overflow error. (b) Numerical
3323 back references of the form \12, where the number was greater than 9, were
3324 not recognized as back references, even though there were sufficient
3325 previous subpatterns.
3326
3327 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
3328 versions of gcc, e.g. 2.95.4.
3329
3330
3331 Version 6.1 21-Jun-05
3332 ---------------------
3333
3334 1. There was one reference to the variable "posix" in pcretest.c that was not
3335 surrounded by "#if !defined NOPOSIX".
3336
3337 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
3338 the cross-check on the old pcre_info() function, for the benefit of the
3339 cut-down version of PCRE that is currently imported into Exim.
3340
3341 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
3342 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
3343 patterns that start (?i) but I don't think that matters. The use of (?i) is
3344 just an example; this all applies to the other options as well.
3345
3346 4. Since libtool seems to echo the compile commands it is issuing, the output
3347 from "make" can be reduced a bit by putting "@" in front of each libtool
3348 compile command.
3349
3350 5. Patch from the folks at Google for configure.in to be a bit more thorough
3351 in checking for a suitable C++ installation before trying to compile the
3352 C++ stuff. This should fix a reported problem when a compiler was present,
3353 but no suitable headers.
3354
3355 6. The man pages all had just "PCRE" as their title. I have changed them to
3356 be the relevant file name. I have also arranged that these names are
3357 retained in the file doc/pcre.txt, which is a concatenation in text format
3358 of all the man pages except the little individual ones for each function.
3359
3360 7. The NON-UNIX-USE file had not been updated for the different set of source
3361 files that come with release 6. I also added a few comments about the C++
3362 wrapper.
3363
3364
3365 Version 6.0 07-Jun-05
3366 ---------------------
3367
3368 1. Some minor internal re-organization to help with my DFA experiments.
3369
3370 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
3371 didn't matter for the library itself when fully configured, but did matter
3372 when compiling without UCP support, or within Exim, where the ucp files are
3373 not imported.
3374
3375 3. Refactoring of the library code to split up the various functions into
3376 different source modules. The addition of the new DFA matching code (see
3377 below) to a single monolithic source would have made it really too
3378 unwieldy, quite apart from causing all the code to be include in a
3379 statically linked application, when only some functions are used. This is
3380 relevant even without the DFA addition now that patterns can be compiled in
3381 one application and matched in another.
3382
3383 The downside of splitting up is that there have to be some external
3384 functions and data tables that are used internally in different modules of
3385 the library but which are not part of the API. These have all had their
3386 names changed to start with "_pcre_" so that they are unlikely to clash
3387 with other external names.
3388
3389 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
3390 a different (DFA) algorithm. Although it is slower than the original
3391 function, it does have some advantages for certain types of matching
3392 problem.
3393
3394 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
3395 including restarting after a partial match.
3396
3397 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
3398 defined when compiling for Windows was sent to me. I have put it into the
3399 code, though I have no means of testing or verifying it.
3400
3401 7. Added the pcre_refcount() auxiliary function.
3402
3403 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
3404 match before or at the first newline in the subject string. In pcretest,
3405 the /f option on a pattern can be used to set this.
3406
3407 9. A repeated \w when used in UTF-8 mode with characters greater than 256
3408 would behave wrongly. This has been present in PCRE since release 4.0.
3409
3410 10. A number of changes to the pcregrep command:
3411
3412 (a) Refactored how -x works; insert ^(...)$ instead of setting
3413 PCRE_ANCHORED and checking the length, in preparation for adding
3414 something similar for -w.
3415
3416 (b) Added the -w (match as a word) option.
3417
3418 (c) Refactored the way lines are read and buffered so as to have more
3419 than one at a time available.
3420
3421 (d) Implemented a pcregrep test script.
3422
3423 (e) Added the -M (multiline match) option. This allows patterns to match
3424 over several lines of the subject. The buffering ensures that at least
3425 8K, or the rest of the document (whichever is the shorter) is available
3426 for matching (and similarly the previous 8K for lookbehind assertions).
3427
3428 (f) Changed the --help output so that it now says
3429
3430 -w, --word-regex(p)
3431
3432 instead of two lines, one with "regex" and the other with "regexp"
3433 because that confused at least one person since the short forms are the
3434 same. (This required a bit of code, as the output is generated
3435 automatically from a table. It wasn't just a text change.)
3436
3437 (g) -- can be used to terminate pcregrep options if the next thing isn't an
3438 option but starts with a hyphen. Could be a pattern or a path name
3439 starting with a hyphen, for instance.
3440
3441 (h) "-" can be given as a file name to represent stdin.
3442
3443 (i) When file names are being printed, "(standard input)" is used for
3444 the standard input, for compatibility with GNU grep. Previously
3445 "<stdin>" was used.
3446
3447 (j) The option --label=xxx can be used to supply a name to be used for
3448 stdin when file names are being printed. There is no short form.
3449
3450 (k) Re-factored the options decoding logic because we are going to add
3451 two more options that take data. Such options can now be given in four
3452 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
3453
3454 (l) Added the -A, -B, and -C options for requesting that lines of context
3455 around matches be printed.
3456
3457 (m) Added the -L option to print the names of files that do not contain
3458 any matching lines, that is, the complement of -l.
3459
3460 (n) The return code is 2 if any file cannot be opened, but pcregrep does
3461 continue to scan other files.
3462
3463 (o) The -s option was incorrectly implemented. For compatibility with other
3464 greps, it now suppresses the error message for a non-existent or non-
3465 accessible file (but not the return code). There is a new option called
3466 -q that suppresses the output of matching lines, which was what -s was
3467 previously doing.
3468
3469 (p) Added --include and --exclude options to specify files for inclusion
3470 and exclusion when recursing.
3471
3472 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
3473 Hopefully, it now does.
3474
3475 12. Missing cast in pcre_study().
3476
3477 13. Added an "uninstall" target to the makefile.
3478
3479 14. Replaced "extern" in the function prototypes in Makefile.in with
3480 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
3481 world, but is set differently for Windows.
3482
3483 15. Added a second compiling function called pcre_compile2(). The only
3484 difference is that it has an extra argument, which is a pointer to an
3485 integer error code. When there is a compile-time failure, this is set
3486 non-zero, in addition to the error test pointer being set to point to an
3487 error message. The new argument may be NULL if no error number is required
3488 (but then you may as well call pcre_compile(), which is now just a
3489 wrapper). This facility is provided because some applications need a
3490 numeric error indication, but it has also enabled me to tidy up the way
3491 compile-time errors are handled in the POSIX wrapper.
3492
3493 16. Added VPATH=.libs to the makefile; this should help when building with one
3494 prefix path and installing with another. (Or so I'm told by someone who
3495 knows more about this stuff than I do.)
3496
3497 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
3498 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
3499 match everything, including newlines. This is not POSIX-compatible, but
3500 somebody wanted the feature. From pcretest it can be activated by using
3501 both the P and the s flags.
3502
3503 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
3504
3505 19. libpcre.pc was being incorrectly installed as executable.
3506
3507 20. A couple of places in pcretest check for end-of-line by looking for '\n';
3508 it now also looks for '\r' so that it will work unmodified on Windows.
3509
3510 21. Added Google's contributed C++ wrapper to the distribution.
3511
3512 22. Added some untidy missing memory free() calls in pcretest, to keep
3513 Electric Fence happy when testing.
3514
3515
3516
3517 Version 5.0 13-Sep-04
3518 ---------------------
3519
3520 1. Internal change: literal characters are no longer packed up into items
3521 containing multiple characters in a single byte-string. Each character
3522 is now matched using a separate opcode. However, there may be more than one
3523 byte in the character in UTF-8 mode.
3524
3525 2. The pcre_callout_block structure has two new fields: pattern_position and
3526 next_item_length. These contain the offset in the pattern to the next match
3527 item, and its length, respectively.
3528
3529 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
3530 insertion of callouts before each pattern item. Added the /C option to
3531 pcretest to make use of this.
3532
3533 4. On the advice of a Windows user, the lines
3534
3535 #if defined(_WIN32) || defined(WIN32)
3536 _setmode( _fileno( stdout ), 0x8000 );
3537 #endif /* defined(_WIN32) || defined(WIN32) */
3538
3539 have been added to the source of pcretest. This apparently does useful
3540 magic in relation to line terminators.
3541
3542 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
3543 for the benefit of those environments where the "b" makes a difference.
3544
3545 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
3546 to know about it. I have put a hack into configure.in that adds in code
3547 to set GCC=yes if CC=icc. This seems to end up at a point in the
3548 generated configure script that is early enough to affect the setting of
3549 compiler options, which is what is needed, but I have no means of testing
3550 whether it really works. (The user who reported this had patched the
3551 generated configure script, which of course I cannot do.)
3552
3553 LATER: After change 22 below (new libtool files), the configure script
3554 seems to know about icc (and also ecc). Therefore, I have commented out
3555 this hack in configure.in.
3556
3557 7. Added support for pkg-config (2 patches were sent in).
3558
3559 8. Negated POSIX character classes that used a combination of internal tables
3560 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
3561 [[:^ascii]]. Typically, they would match almost any characters. The other
3562 POSIX classes were not broken in this way.
3563
3564 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
3565 to find the match, as PCRE was deluded into thinking that the match had to
3566 start at the start point or following a newline. The same bug applied to
3567 patterns with negative forward assertions or any backward assertions
3568 preceding ".*" at the start, unless the pattern required a fixed first
3569 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
3570
3571 10. In UTF-8 mode, when moving forwards in the subject after a failed match
3572 starting at the last subject character, bytes beyond the end of the subject
3573 string were read.
3574
3575 11. Renamed the variable "class" as "classbits" to make life easier for C++
3576 users. (Previously there was a macro definition, but it apparently wasn't
3577 enough.)
3578
3579 12. Added the new field "tables" to the extra data so that tables can be passed
3580 in at exec time, or the internal tables can be re-selected. This allows
3581 a compiled regex to be saved and re-used at a later time by a different
3582 program that might have everything at different addresses.
3583
3584 13. Modified the pcre-config script so that, when run on Solaris, it shows a
3585 -R library as well as a -L library.
3586
3587 14. The debugging options of pcretest (-d on the command line or D on a
3588 pattern) showed incorrect output for anything following an extended class
3589 that contained multibyte characters and which was followed by a quantifier.
3590
3591 15. Added optional support for general category Unicode character properties
3592 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
3593 support. It adds about 90K to the size of the library. The meanings of the
3594 inbuilt class escapes such as \d and \s have NOT been changed.
3595
3596 16. Updated pcredemo.c to include calls to free() to release the memory for the
3597 compiled pattern.
3598
3599 17. The generated file chartables.c was being created in the source directory
3600 instead of in the building directory. This caused the build to fail if the
3601 source directory was different from the building directory, and was
3602 read-only.
3603
3604 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
3605 file. No doubt somebody will tell me if they don't make sense... Also added
3606 Dan Mooney's comments about building on OpenVMS.
3607
3608 19. Added support for partial matching via the PCRE_PARTIAL option for
3609 pcre_exec() and the \P data escape in pcretest.
3610
3611 20. Extended pcretest with 3 new pattern features:
3612
3613 (i) A pattern option of the form ">rest-of-line" causes pcretest to
3614 write the compiled pattern to the file whose name is "rest-of-line".
3615 This is a straight binary dump of the data, with the saved pointer to
3616 the character tables forced to be NULL. The study data, if any, is
3617 written too. After writing, pcretest reads a new pattern.
3618
3619 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
3620 compiled pattern from the given file. There must not be any
3621 occurrences of "<" in the file name (pretty unlikely); if there are,
3622 pcretest will instead treat the initial "<" as a pattern delimiter.
3623 After reading in the pattern, pcretest goes on to read data lines as
3624 usual.
3625
3626 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
3627 and 16-bit fields in a compiled pattern, to simulate a pattern that
3628 was compiled on a host of opposite endianness.
3629
3630 21. The pcre-exec() function can now cope with patterns that were compiled on
3631 hosts of opposite endianness, with this restriction:
3632
3633 As for any compiled expression that is saved and used later, the tables
3634 pointer field cannot be preserved; the extra_data field in the arguments
3635 to pcre_exec() should be used to pass in a tables address if a value
3636 other than the default internal tables were used at compile time.
3637
3638 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
3639 now diagnosed as an error. Previously, most of the time, a negative number
3640 would have been treated as zero, but if in addition "ovector" was passed as
3641 NULL, a crash could occur.
3642
3643 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
3644 new versions from the libtool 1.5 distribution (the last one is a copy of
3645 a file called libtool.m4). This seems to have fixed the need to patch
3646 "configure" to support Darwin 1.3 (which I used to do). However, I still
3647 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
3648 workstation).
3649
3650 24. Changed the PCRE licence to be the more standard "BSD" licence.
3651
3652
3653 Version 4.5 01-Dec-03
3654 ---------------------
3655
3656 1. There has been some re-arrangement of the code for the match() function so
3657 that it can be compiled in a version that does not call itself recursively.
3658 Instead, it keeps those local variables that need separate instances for
3659 each "recursion" in a frame on the heap, and gets/frees frames whenever it
3660 needs to "recurse". Keeping track of where control must go is done by means
3661 of setjmp/longjmp. The whole thing is implemented by a set of macros that
3662 hide most of the details from the main code, and operates only if
3663 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
3664 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
3665 operating.
3666
3667 To make it easier for callers to provide specially tailored get/free
3668 functions for this usage, two new functions, pcre_stack_malloc, and
3669 pcre_stack_free, are used. They are always called in strict stacking order,
3670 and the size of block requested is always the same.
3671
3672 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
3673 PCRE has been compiled to use the stack or the heap for recursion. The
3674 -C option of pcretest uses this to show which version is compiled.
3675
3676 A new data escape \S, is added to pcretest; it causes the amounts of store
3677 obtained and freed by both kinds of malloc/free at match time to be added
3678 to the output.
3679
3680 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
3681 what's available on my current Linux desktop machine.
3682
3683 3. When matching a UTF-8 string, the test for a valid string at the start has
3684 been extended. If start_offset is not zero, PCRE now checks that it points
3685 to a byte that is the start of a UTF-8 character. If not, it returns
3686 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
3687 this is necessary because there may be backward assertions in the pattern.
3688 When matching the same subject several times, it may save resources to use
3689 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
3690
3691 4. The code for checking the validity of UTF-8 strings has been tightened so
3692 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
3693 containing "overlong sequences".
3694
3695 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
3696 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
3697 should have been "&", but it just so happened that all the cases this let
3698 through by mistake were picked up later in the function.
3699
3700 6. I had used a variable called "isblank" - this is a C99 function, causing
3701 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3702
3703 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3704 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3705 so the version is shown; (c) stop "make test" echoing "./RunTest".
3706
3707 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3708
3709 9. The prototype for memmove() for systems that don't have it was using
3710 size_t, but the inclusion of the header that defines size_t was later. I've
3711 moved the #includes for the C headers earlier to avoid this.
3712
3713 10. Added some adjustments to the code to make it easier to compiler on certain
3714 special systems:
3715
3716 (a) Some "const" qualifiers were missing.
3717 (b) Added the macro EXPORT before all exported functions; by default this
3718 is defined to be empty.
3719 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3720 that it reads its output file name as an argument instead of writing
3721 to the standard output and assuming this can be redirected.
3722
3723 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3724 class containing characters with values greater than 255, PCRE compilation
3725 went into a loop.
3726
3727 12. A recursive reference to a subpattern that was within another subpattern
3728 that had a minimum quantifier of zero caused PCRE to crash. For example,
3729 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3730 recursion. If the recursively-called subpattern itself had a zero repeat,
3731 that was OK.
3732
3733 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3734 buffer into which it was copied (for escape processing) was still set at
3735 1024, so long lines caused crashes.
3736
3737 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3738 "internal error: code overflow...". This applied to any character class
3739 that was followed by a possessive quantifier.
3740
3741 15. Modified the Makefile to add libpcre.la as a prerequisite for
3742 libpcreposix.la because I was told this is needed for a parallel build to
3743 work.
3744
3745 16. If a pattern that contained .* following optional items at the start was
3746 studied, the wrong optimizing data was generated, leading to matching
3747 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3748 matching string must start with a or b or c. The correct conclusion for
3749 this pattern is that a match can start with any character.
3750
3751
3752 Version 4.4 13-Aug-03
3753 ---------------------
3754
3755 1. In UTF-8 mode, a character class containing characters with values between
3756 127 and 255 was not handled correctly if the compiled pattern was studied.
3757 In fixing this, I have also improved the studying algorithm for such
3758 classes (slightly).
3759
3760 2. Three internal functions had redundant arguments passed to them. Removal
3761 might give a very teeny performance improvement.
3762
3763 3. Documentation bug: the value of the capture_top field in a callout is *one
3764 more than* the number of the hightest numbered captured substring.
3765
3766 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3767 in incorrectly linking with a previously installed version. They now link
3768 explicitly with libpcre.la.
3769
3770 5. configure.in no longer needs to recognize Cygwin specially.
3771
3772 6. A problem in pcre.in for Windows platforms is fixed.
3773
3774 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3775 pcretest, it used to include the size of the study block as part of its
3776 output. Unfortunately, the structure contains a field that has a different
3777 size on different hardware architectures. This meant that the tests that
3778 showed this size failed. As the block is currently always of a fixed size,
3779 this information isn't actually particularly useful in pcretest output, so
3780 I have just removed it.
3781
3782 8. Three pre-processor statements accidentally did not start in column 1.
3783 Sadly, there are *still* compilers around that complain, even though
3784 standard C has not required this for well over a decade. Sigh.
3785
3786 9. In pcretest, the code for checking callouts passed small integers in the
3787 callout_data field, which is a void * field. However, some picky compilers
3788 complained about the casts involved for this on 64-bit systems. Now
3789 pcretest passes the address of the small integer instead, which should get
3790 rid of the warnings.
3791
3792 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3793 both compile and run time, and gives an error if an invalid UTF-8 sequence
3794 is found. There is a option for disabling this check in cases where the
3795 string is known to be correct and/or the maximum performance is wanted.
3796
3797 11. In response to a bug report, I changed one line in Makefile.in from
3798
3799 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3800 to
3801 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3802
3803 to look similar to other lines, but I have no way of telling whether this
3804 is the right thing to do, as I do not use Windows. No doubt I'll get told
3805 if it's wrong...
3806
3807
3808 Version 4.3 21-May-03
3809 ---------------------
3810
3811 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3812 Makefile.
3813
3814 2. Some refactoring to improve the quality of the code:
3815
3816 (i) The utf8_table... variables are now declared "const".
3817
3818 (ii) The code for \cx, which used the "case flipping" table to upper case
3819 lower case letters, now just substracts 32. This is ASCII-specific,
3820 but the whole concept of \cx is ASCII-specific, so it seems
3821 reasonable.
3822
3823 (iii) PCRE was using its character types table to recognize decimal and
3824 hexadecimal digits in the pattern. This is silly, because it handles
3825 only 0-9, a-f, and A-F, but the character types table is locale-
3826 specific, which means strange things might happen. A private
3827 table is now used for this - though it costs 256 bytes, a table is
3828 much faster than multiple explicit tests. Of course, the standard
3829 character types table is still used for matching digits in subject
3830 strings against \d.
3831
3832 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3833 ending in _t are). So I've renamed it as ESC_tee.
3834
3835 3. The first argument for regexec() in the POSIX wrapper should have been
3836 defined as "const".
3837
3838 4. Changed pcretest to use malloc() for its buffers so that they can be
3839 Electric Fenced for debugging.
3840
3841 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3842 to read one or more bytes before the start of the subject string. Often this
3843 had no effect on PCRE's behaviour, but in some circumstances it could
3844 provoke a segmentation fault.
3845
3846 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3847 to try to read one or more bytes before the start of the subject string.
3848
3849 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3850 UTF-8 support could misbehave in various ways if the subject string
3851 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3852 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3853 back over UTF-8 characters.)
3854
3855
3856 Version 4.2 14-Apr-03
3857 ---------------------
3858
3859 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3860
3861 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3862 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3863 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3864 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3865 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3866 and BUILD_EXEEXT
3867 Note: automatic setting of the BUILD variables is not yet working
3868 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3869 compile-time but not at link-time
3870 [LINK]: use for linking executables only
3871 make different versions for Windows and non-Windows
3872 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3873 libraries
3874 [LINK_FOR_BUILD]: new variable
3875 [OBJEXT]: use throughout
3876 [EXEEXT]: use throughout
3877 <winshared>: new target
3878 <wininstall>: new target
3879 <dftables.o>: use native compiler
3880 <dftables>: use native linker
3881 <install>: handle Windows platform correctly
3882 <clean>: ditto
3883 <check>: ditto
3884 copy DLL to top builddir before testing
3885
3886 As part of these changes, -no-undefined was removed again. This was reported
3887 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3888 in any case.
3889
3890 3. Some tidies to get rid of compiler warnings:
3891
3892 . In the match_data structure, match_limit was an unsigned long int, whereas
3893 match_call_count was an int. I've made them both unsigned long ints.
3894
3895 . In pcretest the fact that a const uschar * doesn't automatically cast to
3896 a void * provoked a warning.
3897
3898 . Turning on some more compiler warnings threw up some "shadow" variables
3899 and a few more missing casts.
3900
3901 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3902 option, a class that contained a single character with a value between 128
3903 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3904
3905 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3906 option, a class that contained several characters, but with at least one
3907 whose value was between 128 and 255 caused PCRE to crash.
3908
3909
3910 Version 4.1 12-Mar-03
3911 ---------------------
3912
3913 1. Compiling with gcc -pedantic found a couple of places where casts were
3914 needed, and a string in dftables.c that was longer than standard compilers are
3915 required to support.
3916
3917 2. Compiling with Sun's compiler found a few more places where the code could
3918 be tidied up in order to avoid warnings.
3919
3920 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3921 first of these names is deprecated in the latest Autoconf in favour of the name
3922 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3923 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3924 analogy I have changed it to CFLAGS_FOR_BUILD.
3925
3926 4. Added -no-undefined to the linking command in the Makefile, because this is
3927 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3928 linking step for the pcreposix library.
3929
3930 5. PCRE was failing to diagnose the case of two named groups with the same
3931 name.
3932
3933 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3934 literal character that is needed in the subject for a match, and scans along to
3935 ensure that it is present before embarking on the full matching process. This
3936 saves time in cases of nested unlimited repeats that are never going to match.
3937 Problem: the scan can take a lot of time if the subject is very long (e.g.
3938 megabytes), thus penalizing straightforward matches. It is now done only if the
3939 amount of subject to be scanned is less than 1000 bytes.
3940
3941 7. A lesser problem with the same optimization is that it was recording the
3942 first character of an anchored pattern as "needed", thus provoking a search
3943 right along the subject, even when the first match of the pattern was going to
3944 fail. The "needed" character is now not set for anchored patterns, unless it
3945 follows something in the pattern that is of non-fixed length. Thus, it still
3946 fulfils its original purpose of finding quick non-matches in cases of nested
3947 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3948
3949
3950 Version 4.0 17-Feb-03
3951 ---------------------
3952
3953 1. If a comment in an extended regex that started immediately after a meta-item
3954 extended to the end of string, PCRE compiled incorrect data. This could lead to
3955 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3956
3957 2. Moved to autoconf 2.53 and libtool 1.4.2.
3958
3959 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3960 the special perltest8 script is no longer needed - all the tests can be run
3961 from a single perltest script.
3962
3963 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3964 by \s. It has now been removed in PCRE. This means it isn't recognized as
3965 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3966 class [:space:] *does* include VT, thereby creating a mess.
3967
3968 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3969 space and tab.
3970
3971 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3972 its new features into the main test script, reducing the number of scripts.
3973
3974 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3975 were backward compatible, and made the (?i) apply to the whole pattern, as if
3976 /i were given. Now it behaves more logically, and applies the option setting
3977 only to what follows. PCRE has been changed to follow suit. However, if it
3978 finds options settings right at the start of the pattern, it extracts them into
3979 the global options, as before. Thus, they show up in the info data.
3980
3981 8. Added support for the \Q...\E escape sequence. Characters in between are
3982 treated as literals. This is slightly different from Perl in that $ and @ are
3983 also handled as literals inside the quotes. In Perl, they will cause variable
3984 interpolation. Note the following examples:
3985
3986 Pattern PCRE matches Perl matches
3987
3988 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3989 \Qabc\$xyz\E abc\$xyz abc\$xyz
3990 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3991
3992 For compatibility with Perl, \Q...\E sequences are recognized inside character
3993 classes as well as outside them.
3994
3995 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3996 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3997 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3998 signed/unsigned warnings.
3999
4000 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4001 option for pcretest, so I've replaced it by a simple function that does just
4002 that job.
4003
4004 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4005 "pcregrep -".
4006
4007 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4008 Java package. This provides some syntactic sugar for simple cases of what my
4009 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4010 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4011 item, you can use this simplified notation. Note that only makes sense with
4012 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4013 greediness, whatever the setting of the PCRE_UNGREEDY option.
4014
4015 13. A change of greediness default within a pattern was not taking effect at
4016 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
4017 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
4018 was abstracted outside.
4019
4020 14. PCRE now supports the \G assertion. It is true when the current matching
4021 position is at the start point of the match. This differs from \A when the
4022 starting offset is non-zero. Used with the /g option of pcretest (or similar
4023 code), it works in the same way as it does for Perl's /g option. If all
4024 alternatives of a regex begin with \G, the expression is anchored to the start
4025 match position, and the "anchored" flag is set in the compiled expression.
4026
4027 15. Some bugs concerning the handling of certain option changes within patterns
4028 have been fixed. These applied to options other than (?ims). For example,
4029 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
4030 been the other way round. Some of this was related to change 7 above.
4031
4032 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
4033 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
4034 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
4035 POSIX classes only within a class (e.g. /[[:alpha:]]/).
4036
4037 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
4038 mode. Unlike ".", it always matches newline, whatever the setting of
4039 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
4040 assertions. Perl allows it, but it doesn't (in general) work because it can't
4041 calculate the length of the lookbehind. At least, that's the case for Perl
4042 5.8.0 - I've been told they are going to document that it doesn't work in
4043 future.
4044
4045 18. Added an error diagnosis for escapes that PCRE does not support: these are
4046 \L, \l, \N, \P, \p, \U, \u, and \X.
4047
4048 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
4049 reading past the end of the pattern in cases such as /[abcd/.
4050
4051 20. PCRE was getting more memory than necessary for patterns with classes that
4052 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
4053
4054 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
4055 compiling PCRE for use with Virtual Pascal.
4056
4057 22. Small fix to the Makefile to make it work properly if the build is done
4058 outside the source tree.
4059
4060 23. Added a new extension: a condition to go with recursion. If a conditional
4061 subpattern starts with (?(R) the "true" branch is used if recursion has
4062 happened, whereas the "false" branch is used only at the top level.
4063
4064 24. When there was a very long string of literal characters (over 255 bytes
4065 without UTF support, over 250 bytes with UTF support), the computation of how
4066 much memory was required could be incorrect, leading to segfaults or other
4067 strange effects.
4068
4069 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
4070 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
4071 there was a subsequent back reference to those brackets. This meant that, for
4072 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
4073 possible to check for precisely this case. All we can do is abandon the
4074 optimization if .* occurs inside capturing brackets when there are any back
4075 references whatsoever. (See below for a better fix that came later.)
4076
4077 26. The handling of the optimization for finding the first character of a
4078 non-anchored pattern, and for finding a character that is required later in the
4079 match were failing in some cases. This didn't break the matching; it just
4080 failed to optimize when it could. The way this is done has been re-implemented.
4081
4082 27. Fixed typo in error message for invalid (?R item (it said "(?p").
4083
4084 28. Added a new feature that provides some of the functionality that Perl
4085 provides with (?{...}). The facility is termed a "callout". The way it is done
4086 in PCRE is for the caller to provide an optional function, by setting
4087 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
4088 global variable. By default it is unset, which disables all calling out. To get
4089 the function called, the regex must include (?C) at appropriate points. This
4090 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
4091 This provides a means of identifying different callout points. When PCRE
4092 reaches such a point in the regex, if pcre_callout has been set, the external
4093 function is called. It is provided with data in a structure called
4094 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
4095 matching continues; if it returns a non-zero value, the match at the current
4096 point fails. However, backtracking will occur if possible. [This was changed
4097 later and other features added - see item 49 below.]
4098
4099 29. pcretest is upgraded to test the callout functionality. It provides a
4100 callout function that displays information. By default, it shows the start of
4101 the match and the current position in the text. There are some new data escapes
4102 to vary what happens:
4103
4104 \C+ in addition, show current contents of captured substrings
4105 \C- do not supply a callout function
4106 \C!n return 1 when callout number n is reached
4107 \C!n!m return 1 when callout number n is reached for the mth time
4108
4109 30. If pcregrep was called with the -l option and just a single file name, it
4110 output "<stdin>" if a match was found, instead of the file name.
4111
4112 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
4113 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
4114 pcre_exec(). This saves a malloc/free per call. The default value of
4115 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
4116 when configuring.
4117
4118 32. The default maximum size of a compiled pattern is 64K. There have been a
4119 few cases of people hitting this limit. The code now uses macros to handle the
4120 storing of links as offsets within the compiled pattern. It defaults to 2-byte
4121 links, but this can be changed to 3 or 4 bytes by --with-link-size when
4122 configuring. Tests 2 and 5 work only with 2-byte links because they output
4123 debugging information about compiled patterns.
4124
4125 33. Internal code re-arrangements:
4126
4127 (a) Moved the debugging function for printing out a compiled regex into
4128 its own source file (printint.c) and used #include to pull it into
4129 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
4130 separate copies.
4131
4132 (b) Defined the list of op-code names for debugging as a macro in
4133 internal.h so that it is next to the definition of the opcodes.
4134
4135 (c) Defined a table of op-code lengths for simpler skipping along compiled
4136 code. This is again a macro in internal.h so that it is next to the
4137 definition of the opcodes.
4138
4139 34. Added support for recursive calls to individual subpatterns, along the
4140 lines of Robin Houston's patch (but implemented somewhat differently).
4141
4142 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
4143 allow it to read and process whole directories in Win32. This code was
4144 contributed by Lionel Fourquaux; it has not been tested by me.
4145
4146 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
4147 used to name a group. Names consist of alphanumerics and underscores, and must
4148 be unique. Back references use the syntax (?P=name) and recursive calls use
4149 (?P>name) which is a PCRE extension to the Python extension. Groups still have
4150 numbers. The function pcre_fullinfo() can be used after compilation to extract
4151 a name/number map. There are three relevant calls:
4152
4153 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
4154 PCRE_INFO_NAMECOUNT yields the number of entries
4155 PCRE_INFO_NAMETABLE yields a pointer to the map.
4156
4157 The map is a vector of fixed-size entries. The size of each entry depends on
4158 the length of the longest name used. The first two bytes of each entry are the
4159 group number, most significant byte first. There follows the corresponding
4160 name, zero terminated. The names are in alphabetical order.
4161
4162 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
4163 case instead of 255. Making it the same both with and without UTF-8 support
4164 means that the same test output works with both.
4165
4166 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
4167 calling malloc() with a zero argument.
4168
4169 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
4170 optimization. I've improved things by keeping a bitmap of backreferences with
4171 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
4172 fact referenced, the optimization can be applied. It is unlikely that a
4173 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
4174 the match to follow \n) will appear inside brackets with a number greater than
4175 31, but if it does, any back reference > 31 suppresses the optimization.
4176
4177 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
4178 of disabling numbered capturing parentheses. Any opening parenthesis that is
4179 not followed by ? behaves as if it were followed by ?: but named parentheses
4180 can still be used for capturing (and they will acquire numbers in the usual
4181 way).
4182
4183 41. Redesigned the return codes from the match() function into yes/no/error so
4184 that errors can be passed back from deep inside the nested calls. A malloc
4185 failure while inside a recursive subpattern call now causes the
4186 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
4187
4188 42. It is now possible to set a limit on the number of times the match()
4189 function is called in a call to pcre_exec(). This facility makes it possible to
4190 limit the amount of recursion and backtracking, though not in a directly
4191 obvious way, because the match() function is used in a number of different
4192 circumstances. The count starts from zero for each position in the subject
4193 string (for non-anchored patterns). The default limit is, for compatibility, a
4194 large number, namely 10 000 000. You can change this in two ways:
4195
4196 (a) When configuring PCRE before making, you can use --with-match-limit=n
4197 to set a default value for the compiled library.
4198
4199 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
4200 a different value is set. See 45 below.
4201
4202 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
4203
4204 43. Added a new function pcre_config(int, void *) to enable run-time extraction
4205 of things that can be changed at compile time. The first argument specifies
4206 what is wanted and the second points to where the information is to be placed.
4207 The current list of available information is:
4208
4209 PCRE_CONFIG_UTF8
4210
4211 The output is an integer that is set to one if UTF-8 support is available;
4212 otherwise it is set to zero.
4213
4214 PCRE_CONFIG_NEWLINE
4215
4216 The output is an integer that it set to the value of the code that is used for
4217 newline. It is either LF (10) or CR (13).
4218
4219 PCRE_CONFIG_LINK_SIZE
4220
4221 The output is an integer that contains the number of bytes used for internal
4222 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
4223
4224 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
4225
4226 The output is an integer that contains the threshold above which the POSIX
4227 interface uses malloc() for output vectors. See item 31 above.
4228
4229 PCRE_CONFIG_MATCH_LIMIT
4230
4231 The output is an unsigned integer that contains the default limit of the number
4232 of match() calls in a pcre_exec() execution. See 42 above.
4233
4234 44. pcretest has been upgraded by the addition of the -C option. This causes it
4235 to extract all the available output from the new pcre_config() function, and to
4236 output it. The program then exits immediately.
4237
4238 45. A need has arisen to pass over additional data with calls to pcre_exec() in
4239 order to support additional features. One way would have been to define
4240 pcre_exec2() (for example) with extra arguments, but this would not have been
4241 extensible, and would also have required all calls to the original function to
4242 be mapped to the new one. Instead, I have chosen to extend the mechanism that
4243 is used for passing in "extra" data from pcre_study().
4244
4245 The pcre_extra structure is now exposed and defined in pcre.h. It currently
4246 contains the following fields:
4247
4248 flags a bitmap indicating which of the following fields are set
4249 study_data opaque data from pcre_study()
4250 match_limit a way of specifying a limit on match() calls for a specific
4251 call to pcre_exec()
4252 callout_data data for callouts (see 49 below)
4253
4254 The flag bits are also defined in pcre.h, and are
4255
4256 PCRE_EXTRA_STUDY_DATA
4257 PCRE_EXTRA_MATCH_LIMIT
4258 PCRE_EXTRA_CALLOUT_DATA
4259
4260 The pcre_study() function now returns one of these new pcre_extra blocks, with
4261 the actual study data pointed to by the study_data field, and the
4262 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
4263 before. That is, this change is entirely upwards-compatible and requires no
4264 change to existing code.
4265
4266 If you want to pass in additional data to pcre_exec(), you can either place it
4267 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
4268 block.
4269
4270 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
4271 data string contains the escape sequence \M, pcretest calls pcre_exec() several
4272 times with different match limits, until it finds the minimum value needed for
4273 pcre_exec() to complete. The value is then output. This can be instructive; for
4274 most simple matches the number is quite small, but for pathological cases it
4275 gets very large very quickly.
4276
4277 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
4278 returns the size of the data block pointed to by the study_data field in a
4279 pcre_extra block, that is, the value that was passed as the argument to
4280 pcre_malloc() when PCRE was getting memory in which to place the information
4281 created by pcre_study(). The fourth argument should point to a size_t variable.
4282 pcretest has been extended so that this information is shown after a successful
4283 pcre_study() call when information about the compiled regex is being displayed.
4284
4285 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
4286 because what follows is always an absolute path. (Later: it turns out that this
4287 is more than cosmetic for MinGW, because it doesn't like empty path
4288 components.)
4289
4290 49. Some changes have been made to the callout feature (see 28 above):
4291
4292 (i) A callout function now has three choices for what it returns:
4293
4294 0 => success, carry on matching
4295 > 0 => failure at this point, but backtrack if possible
4296 < 0 => serious error, return this value from pcre_exec()
4297
4298 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
4299 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
4300 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
4301 use by callout functions. It will never be used by PCRE itself.
4302
4303 (ii) The pcre_extra structure (see 45 above) has a void * field called
4304 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
4305 pcre_callout_block structure has a field of the same name. The contents of
4306 the field passed in the pcre_extra structure are passed to the callout
4307 function in the corresponding field in the callout block. This makes it
4308 easier to use the same callout-containing regex from multiple threads. For
4309 testing, the pcretest program has a new data escape
4310
4311 \C*n pass the number n (may be negative) as callout_data
4312
4313 If the callout function in pcretest receives a non-zero value as
4314 callout_data, it returns that value.
4315
4316 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
4317 there were some redundant $(CFLAGS) in commands that are now specified as
4318 $(LINK), which already includes $(CFLAGS).
4319
4320 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
4321 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
4322 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
4323 one-byte characters throughout. Note that case-insensitive matching applies
4324 only to characters whose values are less than 256. PCRE doesn't support the
4325 notion of cases for higher-valued characters.
4326
4327 (i) A character class whose characters are all within 0-255 is handled as
4328 a bit map, and the map is inverted for negative classes. Previously, a
4329 character > 255 always failed to match such a class; however it should
4330 match if the class was a negative one (e.g. [^ab]). This has been fixed.
4331
4332 (ii) A negated character class with a single character < 255 is coded as
4333 "not this character" (OP_NOT). This wasn't working properly when the test
4334 character was multibyte, either singly or repeated.
4335
4336 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
4337 mode, for example: \x{100}{2,3}.
4338
4339 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
4340 singly or repeated) now correctly test multibyte characters. However,
4341 PCRE doesn't recognize any characters with values greater than 255 as
4342 digits, spaces, or word characters. Such characters always match \D, \S,
4343 and \W, and never match \d, \s, or \w.
4344
4345 (v) Classes may now contain characters and character ranges with values
4346 greater than 255. For example: [ab\x{100}-\x{400}].
4347
4348 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
4349 PCRE in UTF-8 mode.
4350
4351 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
4352 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
4353 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
4354 value.)
4355
4356 53. The single man page has become too large. I have therefore split it up into
4357 a number of separate man pages. These also give rise to individual HTML pages;
4358 these are now put in a separate directory, and there is an index.html page that
4359 lists them all. Some hyperlinking between the pages has been installed.
4360
4361 54. Added convenience functions for handling named capturing parentheses.
4362
4363 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
4364 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
4365 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
4366 are faulted.
4367
4368 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
4369 calling configure. These values are used when compiling the dftables.c program
4370 which is run to generate the source of the default character tables. They
4371 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
4372 you will need to set these values.
4373
4374 57. Updated the building process for Windows DLL, as provided by Fred Cox.
4375
4376
4377 Version 3.9 02-Jan-02
4378 ---------------------
4379
4380 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
4381
4382 2. If --disable-static was given, the building process failed when trying to
4383 build pcretest and pcregrep. (For some reason it was using libtool to compile
4384 them, which is not right, as they aren't part of the library.)
4385
4386
4387 Version 3.8 18-Dec-01
4388 ---------------------
4389
4390 1. The experimental UTF-8 code was completely screwed up. It was packing the
4391 bytes in the wrong order. How dumb can you get?
4392
4393
4394 Version 3.7 29-Oct-01
4395 ---------------------
4396
4397 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
4398 This caused pcretest, when used on the test data, to segfault. Unfortunately,
4399 this didn't happen under Solaris 8, where I normally test things.
4400
4401 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
4402 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
4403 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
4404 this fix an hour or so after the initial 3.7 release.)
4405
4406
4407 Version 3.6 23-Oct-01
4408 ---------------------
4409
4410 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
4411 offsets passed as NULL with zero offset count.
4412
4413 2. The config.guess and config.sub files had not been updated when I moved to
4414 the latest autoconf.
4415
4416
4417 Version 3.5 15-Aug-01
4418 ---------------------
4419
4420 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
4421 had been forgotten.
4422
4423 2. By using declared but undefined structures, we can avoid using "void"
4424 definitions in pcre.h while keeping the internal definitions of the structures
4425 private.
4426
4427 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
4428 user point of view, this means that both static and shared libraries are built
4429 by default, but this can be individually controlled. More of the work of
4430 handling this static/shared cases is now inside libtool instead of PCRE's make
4431 file.
4432
4433 4. The pcretest utility is now installed along with pcregrep because it is
4434 useful for users (to test regexs) and by doing this, it automatically gets
4435 relinked by libtool. The documentation has been turned into a man page, so
4436 there are now .1, .txt, and .html versions in /doc.
4437
4438 5. Upgrades to pcregrep:
4439 (i) Added long-form option names like gnu grep.
4440 (ii) Added --help to list all options with an explanatory phrase.
4441 (iii) Added -r, --recursive to recurse into sub-directories.
4442 (iv) Added -f, --file to read patterns from a file.
4443
4444 6. pcre_exec() was referring to its "code" argument before testing that
4445 argument for NULL (and giving an error if it was NULL).
4446
4447 7. Upgraded Makefile.in to allow for compiling in a different directory from
4448 the source directory.
4449
4450 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
4451 options bits, the pointer it was passed was to an int instead of to an unsigned
4452 long int. This mattered only on 64-bit systems.
4453
4454 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
4455 generated) instead of pcre.in, which it its source. Also made the same change
4456 in several of the .c files.
4457
4458 10. A new release of gcc defines printf() as a macro, which broke pcretest
4459 because it had an ifdef in the middle of a string argument for printf(). Fixed
4460 by using separate calls to printf().
4461
4462 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
4463 script, to force use of CR or LF instead of \n in the source. On non-Unix
4464 systems, the value can be set in config.h.
4465
4466 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
4467 absolute limit. Changed the text of the error message to make this clear, and
4468 likewise updated the man page.
4469
4470 13. The limit of 99 on the number of capturing subpatterns has been removed.
4471 The new limit is 65535, which I hope will not be a "real" limit.
4472
4473
4474 Version 3.4 22-Aug-00
4475 ---------------------
4476
4477 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
4478
4479 2. Diagnose condition (?(0) as an error instead of crashing on matching.
4480
4481
4482 Version 3.3 01-Aug-00
4483 ---------------------
4484
4485 1. If an octal character was given, but the value was greater than \377, it
4486 was not getting masked to the least significant bits, as documented. This could
4487 lead to crashes in some systems.
4488
4489 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
4490 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
4491
4492 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
4493 These just pass their arguments on to (pcre_free)(), but they are provided
4494 because some uses of PCRE bind it to non-C systems that can call its functions,
4495 but cannot call free() or pcre_free() directly.
4496
4497 4. Add "make test" as a synonym for "make check". Corrected some comments in
4498 the Makefile.
4499
4500 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
4501 Makefile.
4502
4503 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
4504 command called pgrep for grepping around the active processes.
4505
4506 7. Added the beginnings of support for UTF-8 character strings.
4507
4508 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
4509 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
4510 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
4511 out for the ar command.)
4512
4513
4514 Version 3.2 12-May-00
4515 ---------------------
4516
4517 This is purely a bug fixing release.
4518
4519 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
4520 of ZA. This was just one example of several cases that could provoke this bug,
4521 which was introduced by change 9 of version 2.00. The code for breaking
4522 infinite loops after an iteration that matches an empty string was't working
4523 correctly.
4524
4525 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
4526 when matched against abbab (for example). After matching an empty string, it
4527 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
4528 caused it to match further down the string than it should.
4529
4530 3. The code contained an inclusion of sys/types.h. It isn't clear why this
4531 was there because it doesn't seem to be needed, and it causes trouble on some
4532 systems, as it is not a Standard C header. It has been removed.
4533
4534 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
4535 were reported on the Macintosh. The changes were from
4536
4537 while ((c = *(++ptr)) != 0 && c != '\n');
4538 to
4539 while ((c = *(++ptr)) != 0 && c != '\n') ;
4540
4541 Totally extraordinary, but if that's what it takes...
4542
4543 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
4544 available. Added HAVE_BCOPY and an autoconf test for it; if neither
4545 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
4546 assumes the way PCRE uses memmove() (always moving upwards).
4547
4548 6. PCRE is being used in one environment where strchr() is not available. There
4549 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
4550 faster code anyway.
4551
4552
4553 Version 3.1 09-Feb-00
4554 ---------------------
4555
4556 The only change in this release is the fixing of some bugs in Makefile.in for
4557 the "install" target:
4558
4559 (1) It was failing to install pcreposix.h.
4560
4561 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
4562
4563
4564 Version 3.0 01-Feb-00
4565 ---------------------
4566
4567 1. Add support for the /+ modifier to perltest (to output $` like it does in
4568 pcretest).
4569
4570 2. Add support for the /g modifier to perltest.
4571
4572 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
4573 matches null strings.
4574
4575 4. Fix perltest so that it doesn't do unwanted things when fed an empty
4576 pattern. Perl treats empty patterns specially - it reuses the most recent
4577 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
4578 effect.
4579
4580 5. The POSIX interface was broken in that it was just handing over the POSIX
4581 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
4582 required a bigger vector, with some working space on the end. This means that
4583 the POSIX wrapper now has to get and free some memory, and copy the results.
4584
4585 6. Added some simple autoconf support, placing the test data and the
4586 documentation in separate directories, re-organizing some of the
4587 information files, and making it build pcre-config (a GNU standard). Also added
4588 libtool support for building PCRE as a shared library, which is now the
4589 default.
4590
4591 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
4592 09 are not valid octal constants. Single digits will be used for minor values
4593 less than 10.
4594
4595 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
4596 existing programs that set these in the POSIX interface can use PCRE without
4597 modification.
4598
4599 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
4600 return all that pcre_info() returns, plus additional data. The pcre_info()
4601 function is retained for compatibility, but is considered to be obsolete.
4602
4603 10. Added experimental recursion feature (?R) to handle one common case that
4604 Perl 5.6 will be able to do with (?p{...}).
4605
4606 11. Added support for POSIX character classes like [:alpha:], which Perl is
4607 adopting.
4608
4609
4610 Version 2.08 31-Aug-99
4611 ----------------------
4612
4613 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
4614 trying to match at the startoffset position, but instead was moving forward to
4615 the next newline as if a previous match had failed.
4616
4617 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
4618 and could get into a loop if a null string was matched other than at the start
4619 of the subject.
4620
4621 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
4622 be distinguished at compile time, and for completeness also added PCRE_DATE.
4623
4624 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
4625 in GnuWin32 environments.
4626
4627
4628 Version 2.07 29-Jul-99
4629 ----------------------
4630
4631 1. The documentation is now supplied in plain text form and HTML as well as in
4632 the form of man page sources.
4633
4634 2. C++ compilers don't like assigning (void *) values to other pointer types.
4635 In particular this affects malloc(). Although there is no problem in Standard
4636 C, I've put in casts to keep C++ compilers happy.
4637
4638 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
4639 should be (const char *).
4640
4641 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
4642 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
4643 However, I haven't made this a standard facility. The documentation doesn't
4644 mention it, and the Makefile doesn't support it.
4645
4646 5. The Makefile now contains an "install" target, with editable destinations at
4647 the top of the file. The pcretest program is not installed.
4648
4649 6. pgrep -V now gives the PCRE version number and date.
4650
4651 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
4652 causing the entire string to be ignored, instead of just the last character.
4653
4654 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
4655 non-matching string, it can take a very, very long time, even for strings of
4656 quite modest length, because of the nested recursion. PCRE now does better in
4657 some of these cases. It does this by remembering the last required literal
4658 character in the pattern, and pre-searching the subject to ensure it is present
4659 before running the real match. In other words, it applies a heuristic to detect
4660 some types of certain failure quickly, and in the above example, if presented
4661 with a string that has no trailing " it gives "no match" very quickly.
4662
4663 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
4664 other alternatives are tried instead.
4665
4666
4667 Version 2.06 09-Jun-99
4668 ----------------------
4669
4670 1. Change pcretest's output for amount of store used to show just the code
4671 space, because the remainder (the data block) varies in size between 32-bit and
4672 64-bit systems.
4673
4674 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
4675 start matching at. This allows lookbehinds to work when searching for multiple
4676 occurrences in a string.
4677
4678 3. Added additional options to pcretest for testing multiple occurrences:
4679
4680 /+ outputs the rest of the string that follows a match
4681 /g loops for multiple occurrences, using the new startoffset argument
4682 /G loops for multiple occurrences by passing an incremented pointer
4683
4684 4. PCRE wasn't doing the "first character" optimization for patterns starting
4685 with \b or \B, though it was doing it for other lookbehind assertions. That is,
4686 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
4687 the letter 'x'. On long subject strings, this gives a significant speed-up.
4688
4689
4690 Version 2.05 21-Apr-99
4691 ----------------------
4692
4693 1. Changed the type of magic_number from int to long int so that it works
4694 properly on 16-bit systems.
4695
4696 2. Fixed a bug which caused patterns starting with .* not to work correctly
4697 when the subject string contained newline characters. PCRE was assuming
4698 anchoring for such patterns in all cases, which is not correct because .* will
4699 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
4700 DOTALL is set at top level; otherwise it knows that patterns starting with .*
4701 must be retried after every newline in the subject.
4702
4703
4704 Version 2.04 18-Feb-99
4705 ----------------------
4706
4707 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4708 computation of the store needed to hold the pattern was incorrect (too large).
4709 If such patterns were nested a few deep, this could multiply and become a real
4710 problem.
4711
4712 2. Added /M option to pcretest to show the memory requirement of a specific
4713 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4714
4715 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4716 compiled in such a way that the backtracking after subsequent failure was
4717 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4718 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4719
4720
4721 Version 2.03 02-Feb-99
4722 ----------------------
4723
4724 1. Fixed typo and small mistake in man page.
4725
4726 2. Added 4th condition (GPL supersedes if conflict) and created separate
4727 LICENCE file containing the conditions.
4728
4729 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4730 Perl, that is the internal \ allows the delimiter to be included in the
4731 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4732 the final delimiter, add \ to the end of the pattern (to test the error).
4733
4734 4. Added the convenience functions for extracting substrings after a successful
4735 match. Updated pcretest to make it able to test these functions.
4736
4737
4738 Version 2.02 14-Jan-99
4739 ----------------------
4740
4741 1. Initialized the working variables associated with each extraction so that
4742 their saving and restoring doesn't refer to uninitialized store.
4743
4744 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4745 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4746 fix the problem.
4747
4748 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4749 calls, and wasn't printing the correct value for compiling calls. Increased the
4750 default value of LOOPREPEAT, and the number of significant figures in the
4751 times.
4752
4753 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4754
4755 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4756 a building problem on Windows NT with a FAT file system.
4757
4758
4759 Version 2.01 21-Oct-98
4760 ----------------------
4761
4762 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4763 to character tables built by pcre_maketables() in the current locale. If NULL
4764 is passed, the default tables are used.
4765
4766
4767 Version 2.00 24-Sep-98
4768 ----------------------
4769
4770 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4771 it any more.
4772
4773 2. Allow quantification of (?>) groups, and make it work correctly.
4774
4775 3. The first character computation wasn't working for (?>) groups.
4776
4777 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4778 end of the subject) and add 5.005's \z, which really does match only at the
4779 very end of the subject.
4780
4781 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4782
4783 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4784 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4785 localized options. All options to pcre_study() were also removed.
4786
4787 7. Add other new features from 5.005:
4788
4789 $(?<= positive lookbehind
4790 $(?<! negative lookbehind
4791 (?imsx-imsx) added the unsetting capability
4792 such a setting is global if at outer level; local otherwise
4793 (?imsx-imsx:) non-capturing groups with option setting
4794 (?(cond)re|re) conditional pattern matching
4795
4796 A backreference to itself in a repeated group matches the previous
4797 captured string.
4798
4799 8. General tidying up of studying (both automatic and via "study")
4800 consequential on the addition of new assertions.
4801
4802 9. As in 5.005, unlimited repeated groups that could match an empty substring
4803 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4804 runtime if any iteration does actually match an empty substring.
4805
4806 10. Include the RunTest script in the distribution.
4807
4808 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4809 discrepancies, some of which were old and were also with respect to 5.004. They
4810 have now been fixed.
4811
4812
4813 Version 1.09 28-Apr-98
4814 ----------------------
4815
4816 1. A negated single character class followed by a quantifier with a minimum
4817 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4818 program crashes, or just wrong answers. This did not apply to negated classes
4819 containing more than one character, or to minima other than one.
4820
4821
4822 Version 1.08 27-Mar-98
4823 ----------------------
4824
4825 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4826
4827 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4828 latter must appear before anything that relies on it in the pattern.
4829
4830
4831 Version 1.07 16-Feb-98
4832 ----------------------
4833
4834 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4835 repeat of a potentially empty string).
4836
4837
4838 Version 1.06 23-Jan-98
4839 ----------------------
4840
4841 1. Added Markus Oberhumer's little patches for C++.
4842
4843 2. Literal strings longer than 255 characters were broken.
4844
4845
4846 Version 1.05 23-Dec-97
4847 ----------------------
4848
4849 1. Negated character classes containing more than one character were failing if
4850 PCRE_CASELESS was set at run time.
4851
4852
4853 Version 1.04 19-Dec-97
4854 ----------------------
4855
4856 1. Corrected the man page, where some "const" qualifiers had been omitted.
4857
4858 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4859 input syntax.
4860
4861 3. Fixed memory leak which occurred when a regex with back references was
4862 matched with an offsets vector that wasn't big enough. The temporary memory
4863 that is used in this case wasn't being freed if the match failed.
4864
4865 4. Tidied pcretest to ensure it frees memory that it gets.
4866
4867 5. Temporary memory was being obtained in the case where the passed offsets
4868 vector was exactly big enough.
4869
4870 6. Corrected definition of offsetof() from change 5 below.
4871
4872 7. I had screwed up change 6 below and broken the rules for the use of
4873 setjmp(). Now fixed.
4874
4875
4876 Version 1.03 18-Dec-97
4877 ----------------------
4878
4879 1. A erroneous regex with a missing opening parenthesis was correctly
4880 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4881 on some systems.
4882
4883 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4884 it was reported that one broken compiler failed on the former because "code" is
4885 also an independent variable.
4886
4887 3. The erroneous regex a[]b caused an array overrun reference.
4888
4889 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4890 fail on data ending with that character. (It was going on too far, and checking
4891 the next character, typically a binary zero.) This was specific to the
4892 optimized code for single-character negative classes.
4893
4894 5. Added a contributed patch from the TIN world which does the following:
4895
4896 + Add an undef for memmove, in case the the system defines a macro for it.
4897
4898 + Add a definition of offsetof(), in case there isn't one. (I don't know
4899 the reason behind this - offsetof() is part of the ANSI standard - but
4900 it does no harm).
4901
4902 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4903 most of the places where whitespace preceded '#'. I have given up and
4904 allowed the remaining 2 cases to be at the margin.
4905
4906 + Rename some variables in pcre to eliminate shadowing. This seems very
4907 pedantic, but does no harm, of course.
4908
4909 6. Moved the call to setjmp() into its own function, to get rid of warnings
4910 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4911
4912 7. Constructs such as \d{8,} were compiling into the equivalent of
4913 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4914 outcome, but in this particular case used more store than had been allocated,
4915 which caused the bug to be discovered because it threw up an internal error.
4916
4917 8. The debugging code in both pcre and pcretest for outputting the compiled
4918 form of a regex was going wrong in the case of back references followed by
4919 curly-bracketed repeats.
4920
4921
4922 Version 1.02 12-Dec-97
4923 ----------------------
4924
4925 1. Typos in pcre.3 and comments in the source fixed.
4926
4927 2. Applied a contributed patch to get rid of places where it used to remove
4928 'const' from variables, and fixed some signed/unsigned and uninitialized
4929 variable warnings.
4930
4931 3. Added the "runtest" target to Makefile.
4932
4933 4. Set default compiler flag to -O2 rather than just -O.
4934
4935
4936 Version 1.01 19-Nov-97
4937 ----------------------
4938
4939 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4940 like /([ab]*)*/, that is, for classes with more than one character in them.
4941
4942 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4943 as /((?>a*))*/ (a PCRE_EXTRA facility).
4944
4945
4946 Version 1.00 18-Nov-97
4947 ----------------------
4948
4949 1. Added compile-time macros to support systems such as SunOS4 which don't have
4950 memmove() or strerror() but have other things that can be used instead.
4951
4952 2. Arranged that "make clean" removes the executables.
4953
4954
4955 Version 0.99 27-Oct-97
4956 ----------------------
4957
4958 1. Fixed bug in code for optimizing classes with only one character. It was
4959 initializing a 32-byte map regardless, which could cause it to run off the end
4960 of the memory it had got.
4961
4962 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4963
4964
4965 Version 0.98 22-Oct-97
4966 ----------------------
4967
4968 1. Fixed bug in code for handling temporary memory usage when there are more
4969 back references than supplied space in the ovector. This could cause segfaults.
4970
4971
4972 Version 0.97 21-Oct-97
4973 ----------------------
4974
4975 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4976
4977 2. Optimized negated single characters not to use a bit map.
4978
4979 3. Brought error texts together as macro definitions; clarified some of them;
4980 fixed one that was wrong - it said "range out of order" when it meant "invalid
4981 escape sequence".
4982
4983 4. Changed some char * arguments to const char *.
4984
4985 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4986
4987 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4988 pcretest.
4989
4990
4991 Version 0.96 16-Oct-97
4992 ----------------------
4993
4994 1. Added a simple "pgrep" utility to the distribution.
4995
4996 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4997 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4998 where "ddd" means "one or more decimal digits".
4999
5000 3. Fixed serious bug. If a pattern had a back reference, but the call to
5001 pcre_exec() didn't supply a large enough ovector to record the related
5002 identifying subpattern, the match always failed. PCRE now remembers the number
5003 of the largest back reference, and gets some temporary memory in which to save
5004 the offsets during matching if necessary, in order to ensure that
5005 backreferences always work.
5006
5007 4. Increased the compatibility with Perl in a number of ways:
5008
5009 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5010 to request this handling. The option can be set at compile or exec time.
5011
5012 (b) $ matches before a terminating newline by default; an option
5013 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
5014 mode). The option can be set at compile or exec time.
5015
5016 (c) The handling of \ followed by a digit other than 0 is now supposed to be
5017 the same as Perl's. If the decimal number it represents is less than 10
5018 or there aren't that many previous left capturing parentheses, an octal
5019 escape is read. Inside a character class, it's always an octal escape,
5020 even if it is a single digit.
5021
5022 (d) An escaped but undefined alphabetic character is taken as a literal,
5023 unless PCRE_EXTRA is set. Currently this just reserves the remaining
5024 escapes.
5025
5026 (e) {0} is now permitted. (The previous item is removed from the compiled
5027 pattern).
5028
5029 5. Changed all the names of code files so that the basic parts are no longer
5030 than 10 characters, and abolished the teeny "globals.c" file.
5031
5032 6. Changed the handling of character classes; they are now done with a 32-byte
5033 bit map always.
5034
5035 7. Added the -d and /D options to pcretest to make it possible to look at the
5036 internals of compilation without having to recompile pcre.
5037
5038
5039 Version 0.95 23-Sep-97
5040 ----------------------
5041
5042 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
5043 \x20 at the start of a run of normal characters. These were being treated as
5044 real characters, instead of the source characters being re-checked.
5045
5046
5047 Version 0.94 18-Sep-97
5048 ----------------------
5049
5050 1. The functions are now thread-safe, with the caveat that the global variables
5051 containing pointers to malloc() and free() or alternative functions are the
5052 same for all threads.
5053
5054 2. Get pcre_study() to generate a bitmap of initial characters for non-
5055 anchored patterns when this is possible, and use it if passed to pcre_exec().
5056
5057
5058 Version 0.93 15-Sep-97
5059 ----------------------
5060
5061 1. /(b)|(:+)/ was computing an incorrect first character.
5062
5063 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
5064 but not actually doing anything yet.
5065
5066 3. Treat "-" characters in classes that cannot be part of ranges as literals,
5067 as Perl does (e.g. [-az] or [az-]).
5068
5069 4. Set the anchored flag if a branch starts with .* or .*? because that tests
5070 all possible positions.
5071
5072 5. Split up into different modules to avoid including unneeded functions in a
5073 compiled binary. However, compile and exec are still in one module. The "study"
5074 function is split off.
5075
5076 6. The character tables are now in a separate module whose source is generated
5077 by an auxiliary program - but can then be edited by hand if required. There are
5078 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
5079 toupper() in the code.
5080
5081 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
5082 make them global. Abolish the function for setting them, as the caller can now
5083 set them directly.
5084
5085
5086 Version 0.92 11-Sep-97
5087 ----------------------
5088
5089 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
5090 (e.g. /a{1,3}/) was broken (I mis-optimized it).
5091
5092 2. Caseless matching was not working in character classes if the characters in
5093 the pattern were in upper case.
5094
5095 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
5096
5097 4. Make PCRE_ANCHORED public and accept as a compile option.
5098
5099 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
5100 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
5101 pass them.
5102
5103 6. Give an error if bad option bits passed at compile or run time.
5104
5105 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
5106 pcretest to cause it to pass that flag.
5107
5108 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
5109 options, and the first character, if set.
5110
5111 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
5112
5113
5114 Version 0.91 10-Sep-97
5115 ----------------------
5116
5117 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
5118 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
5119
5120 2. PCRE was looping on encountering an indefinitely repeated back reference to
5121 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
5122 Perl does - treats the match as successful.
5123
5124 ****

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5